Skip to content

feat: effort level display in model bracket - #471

Merged
jarrodwatts merged 1 commit into
jarrodwatts:mainfrom
zelinewang:feat/effort-level-display
Apr 20, 2026
Merged

jarrodwatts merged 1 commit into
jarrodwatts:mainfrom
zelinewang:feat/effort-level-display

Conversation

@zelinewang

Copy link
Copy Markdown
Contributor

Summary

  • Displays reasoning effort level inside the model bracket: [Opus 4.6 ●max]
  • Opt-in via display.showEffortLevel: true (off by default)
  • Data source: Reads stdin.effort first (future-proof for when Claude Code adds it — see Include reasoning effort level in statusline JSON data anthropics/claude-code#39399, #50060, #50577, #45805), falls back to parent process CLI args (--effort flag)
  • Progressive fill symbols for visual clarity: ○ low · ◔ medium · ◑ high · ◕ xhigh · ● max
  • Unknown future levels display as plain text (forward-compatible)
  • Works in both compact and expanded layouts

Bug Fix (setup.md)

The bash version selector in commands/setup.md doesn't filter non-semver directories, unlike the PowerShell version which has Where-Object { $_.Name -match '^\d+(\.\d+)+$' }. This causes the HUD to silently use the wrong directory if backup/temp directories exist (e.g., 0.0.10.bak).

Fix: Added grep -E '^[0-9]+\.[0-9]+\.[0-9]+\t' before sort in all bash templates.

Design Decisions

Decision Rationale
Model bracket placement Effort is a property of the model's behavior — natural to group together
stdin.effort priority Authoritative source when available; ready for upstream support
Parent process fallback ps -o args= detects --effort CLI flag at session start
No settings.json fallback Settings reflect defaults, not per-session state — would be misleading
String-based (not enum) Forward-compatible with any future effort levels Claude Code may add

Configuration

{
  "display": {
    "showEffortLevel": true
  }
}

Visual Examples

[Opus 4.6 ●max] ████░░░░░░ 45%     ← effort max
[Sonnet 4.6 ◑high] ██░░░░░░░░ 20%   ← effort high  
[Opus 4.6] ████░░░░░░ 45%           ← effort not detected (graceful)

Test Plan

  • All 399 existing tests pass (0 failures)
  • 8 new effort tests pass (stdin priority, all levels, normalization, unknown levels)
  • npm run build succeeds
  • Manual test: compact layout shows [Model ●max]
  • Manual test: expanded layout shows [Model ●max]
  • Manual test: unknown level shows plain text (no symbol)
  • Verify with real Claude Code session (requires showEffortLevel: true in config)

Closes #201
Relates to anthropics/claude-code#39399, #50060, #50577, #45805

Displays the current reasoning effort level inside the model bracket
(e.g., [Opus 4.6 ●max]) when `display.showEffortLevel` is enabled.

Data source priority:
1. stdin.effort field (future-proof — ready for when Claude Code
   exposes effort in statusline JSON, per issues #39399, #50060)
2. Parent process CLI args (detects --effort flag at session start)

Supports all known levels with progressive fill symbols:
  ○ low | ◔ medium | ◑ high | ◕ xhigh | ● max

Unknown future levels display as plain text (no symbol), ensuring
forward compatibility as Claude Code adds/removes effort levels.

Also fixes: version selector in setup.md bash templates now filters
non-semver directories (e.g., .bak, -old) via grep before sort,
matching the existing PowerShell behavior.

Closes jarrodwatts#201, relates to anthropics/claude-code#39399
@zelinewang
zelinewang requested a review from jarrodwatts as a code owner April 20, 2026 10:43

@jarrodwatts jarrodwatts left a comment

Copy link
Copy Markdown
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The opt-in model-bracket placement fits the HUD better than the environment line variants, the data source prefers real session state over stale defaults, and the rebased branch passed the full test suite on current main. Merging.

@jarrodwatts
jarrodwatts merged commit a5fa6ff into jarrodwatts:main Apr 20, 2026
2 checks passed
prezis added a commit to prezis/claude-hud that referenced this pull request Apr 30, 2026
This fork adds a 4th resolution step to claude-hud's effort detection so
the badge works for users on Claude Code < 2.1.115 (no stdin.effort) and
for users who set effortLevel persistently in settings.json instead of
via --effort CLI flag.

Resolution priority (highest -> lowest):
  1. stdin.effort string (PR jarrodwatts#471 path, upstream)
  2. stdin.effort.level object (PR jarrodwatts#491, upstream, CC 2.1.115+)
  3. --effort CLI flag from parent process (upstream)
  4. settings.json effortLevel (NEW, this fork) -- reads from
     ${cwd}/.claude/settings.local.json, then settings.json, then
     ${CLAUDE_CONFIG_DIR:-~/.claude}/settings.json. Most-specific wins.

Behavior change: display.showEffortLevel default flipped false -> true
(this fork's whole purpose is to make effort visible by default).

Source changes (already in eb58786 via auto-commit hook): src/effort.ts
adds readSettingsEffort + cwd param; src/index.ts threads stdin.cwd;
src/config.ts flips default; tests/effort.test.js +8 tests for the
fallback (project-local > project > user precedence, malformed JSON,
non-string values, stdin still wins).

This commit: README banner explaining fork differences + CHANGELOG entry.

Tests: 519/520 pass (1 pre-existing skip).

Upstream-able: source changes are surgical and additive; PR candidate.
hseunghwan added a commit to hseunghwan/claude-hud that referenced this pull request Jun 20, 2026
stdin reports ultracode as plain `xhigh`, so the distinction is recovered from the transcript: the ultra_effort_enter / ultra_effort_exit attachment markers combined with the immediate /effort command output, last signal wins. The exit attachment can lag a turn, so the /effort output is what clears a stale label the instant effort drops back to plain xhigh.

The /effort signal matches only a complete <local-command-stdout>...</local-command-stdout> block anchored at the start of a user record, so prose that merely quotes or echoes the phrase (in user or assistant text) cannot flip the state.

Also drops the ps-based parent-process --effort fallback (jarrodwatts#471); effort now comes solely from Claude Code stdin, removing a subprocess spawn on every statusline refresh.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01K9mvZZfMk6WyKiTxnV4GAG
hseunghwan added a commit to hseunghwan/claude-hud that referenced this pull request Jun 20, 2026
stdin reports ultracode as plain xhigh, so the distinction is recovered from the transcript: the ultra_effort_enter / ultra_effort_exit attachment markers combined with the immediate /effort command output, last signal wins. The exit attachment can lag a turn, so the /effort output is what clears a stale label the instant effort drops back to plain xhigh.

The /effort signal matches only a complete <local-command-stdout>...</local-command-stdout> block anchored at the start of a user record, so prose that merely quotes or echoes the phrase (in user or assistant text) cannot flip the state.

Also drops the ps-based parent-process --effort fallback (jarrodwatts#471); effort now comes solely from Claude Code stdin, removing a subprocess spawn on every statusline refresh.
hseunghwan added a commit to hseunghwan/claude-hud that referenced this pull request Jun 20, 2026
stdin reports ultracode as a normal level (xhigh), so the distinction is recovered from the transcript: the ultra_effort_enter / ultra_effort_exit attachment markers combined with the immediate /effort command output, last signal wins. The exit attachment can lag a turn, so the /effort output is what clears a stale label the instant effort drops back.

When the marker is active the reported level is wrapped as ultracode(<level>), taken as-is rather than assumed to be xhigh. The /effort signal matches a <local-command-stdout> block anchored at the start of a user record, so prose that merely quotes the phrase cannot flip the state.

Also drops the ps-based parent-process --effort fallback (jarrodwatts#471); effort now comes solely from Claude Code stdin, removing a subprocess spawn on every statusline refresh.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01K9mvZZfMk6WyKiTxnV4GAG
hseunghwan added a commit to hseunghwan/claude-hud that referenced this pull request Jun 20, 2026
stdin reports ultracode as a normal level (xhigh), so the distinction is recovered from the transcript: the ultra_effort_enter / ultra_effort_exit attachment markers combined with the immediate /effort command output, last signal wins. The exit attachment can lag a turn, so the /effort output is what clears a stale label the instant effort drops back.

When the marker is active the reported level is wrapped as ultracode(<level>), taken as-is rather than assumed to be xhigh. The /effort signal matches a <local-command-stdout> block anchored at the start of a user record, so prose that merely quotes the phrase cannot flip the state.

Also drops the ps-based parent-process --effort fallback (jarrodwatts#471); effort now comes solely from Claude Code stdin, removing a subprocess spawn on every statusline refresh.
hseunghwan added a commit to hseunghwan/claude-hud that referenced this pull request Jun 20, 2026
Claude Code reports ultracode as a normal level (xhigh) in stdin, so the
statusline recovers it from the transcript: the ultra_effort_enter /
ultra_effort_exit attachment markers and the immediate /effort command
output, latest signal wins. When active, the reported level is wrapped as
ultracode(<level>), taken as-is rather than assumed to be xhigh.

Also drop the ps-based parent-process --effort fallback (jarrodwatts#471); the effort
label now comes solely from Claude Code stdin, removing a subprocess spawn
on every statusline refresh.

Co-authored-by: Claude <[email protected]>
jarrodwatts pushed a commit to hseunghwan/claude-hud that referenced this pull request Jul 13, 2026
Claude Code reports ultracode as a normal level (xhigh) in stdin, so the
statusline recovers it from the transcript: the ultra_effort_enter /
ultra_effort_exit attachment markers and the immediate /effort command
output, latest signal wins. When active, the reported level is wrapped as
ultracode(<level>), taken as-is rather than assumed to be xhigh.

Also drop the ps-based parent-process --effort fallback (jarrodwatts#471); the effort
label now comes solely from Claude Code stdin, removing a subprocess spawn
on every statusline refresh.

Co-authored-by: Claude <[email protected]>
jarrodwatts pushed a commit that referenced this pull request Jul 13, 2026
Claude Code reports ultracode as a normal level (xhigh) in stdin, so the
statusline recovers it from the transcript: the ultra_effort_enter /
ultra_effort_exit attachment markers and the immediate /effort command
output, latest signal wins. When active, the reported level is wrapped as
ultracode(<level>), taken as-is rather than assumed to be xhigh.

Also drop the ps-based parent-process --effort fallback (#471); the effort
label now comes solely from Claude Code stdin, removing a subprocess spawn
on every statusline refresh.

Co-authored-by: Claude <[email protected]>
UnpxreTW added a commit to UnpxreTW/claude-hud that referenced this pull request Jul 13, 2026
* feat: add advisor model display (display.showAdvisor)

Surfaces the model configured via Claude Code's `/advisor` command so it
sits inline on the project line:

  [Opus 4.7 (1M context)] │ AthenaKB git:(main*) │ Advisor: Opus 4.7
  Context ██░░░░░░░░ 19%

## How it works

Claude Code stamps an `advisorModel` field onto every assistant record
in the session transcript when `/advisor` is configured. The transcript
parser already walks every line for tool / agent / todo state, so this
just captures the latest non-empty `advisorModel` into the existing
`TranscriptData` shape with no extra IO.

`renderProjectLine` then appends the prettified label (e.g.
`claude-opus-4-7` → `Opus 4.7`) as another part on the same row,
respecting the existing ` │ ` separator and ANSI handling. A
`display.showAdvisor` boolean (default false) gates the whole block;
`display.advisorOverride` lets users force a custom label if needed.

## What's included

- `TranscriptData.advisorModel` field + parser capture
- TRANSCRIPT_CACHE_VERSION bumped 5 → 6 to invalidate caches that
  predate the new field
- `display.showAdvisor` / `display.advisorOverride` config + validation
- `renderProjectLine` integration + `prettifyAdvisorId` helper that
  handles both canonical IDs and short aliases
- i18n labels for en (`Advisor`) and zh-Hans (`顾问`)
- `configure.md` Element Mapping + Turn Off/On flow entries
- 15 new tests covering ID prettification, render gating, transcript
  capture (latest-wins, empty-string-ignored, missing-field fallback),
  and config validation

## Notes

- `dist/` is intentionally untouched per CONTRIBUTING.md
- One pre-existing failure remains in `countConfigs cache: miss on
  nested rules additions` — it fails identically on main without any
  of my changes

* feat: add customLinePosition config to control custom line placement (jarrodwatts#575)

Adds display.customLinePosition so users can place customLine before the model badge while preserving the existing default placement.

* build: compile dist/ [auto]

* fix: stabilize context fallback and todo rendering (jarrodwatts#579)

Fixes context fallback for streaming zero frames and guards todo rendering when content is missing.\n\nCloses jarrodwatts#576.\nCloses jarrodwatts#577.

* build: compile dist/ [auto]

* fix(advisor): security hardening pass on advisor display

Addresses review feedback from jarrodwatts#573:

- Sanitize both `ctx.transcript.advisorModel` and `display.advisorOverride`
  through the existing `sanitizeDisplayText` helper before rendering so
  ANSI ESC, OSC, C0/C1 control, and bidi sequences cannot reach the
  terminal. Defence-in-depth: re-sanitize the prettified result too.
- Cap advisor display at 64 chars at the render layer
  (MAX_ADVISOR_DISPLAY_LEN), regardless of source.
- In the transcript parser, capture `advisorModel` only from records
  whose `type === 'assistant'` (matching the documented schema) and
  hard-cap the captured value at 64 chars (ADVISOR_MODEL_MAX_LEN) so a
  malformed log cannot persist an oversized value through the JSON
  cache. Bumps TRANSCRIPT_CACHE_VERSION 6 → 7. The cache deserializer
  also re-applies the cap as belt-and-braces against legacy entries.
- Add 9 new tests covering: CSI/OSC stripping (ANSI ESC + BEL + DEL
  bytes), bidi mark stripping (RLO/PDF), oversize transcript truncation,
  oversize override truncation (display-layer), sanitize-collapses-to-
  empty fallback, non-assistant record rejection (user / custom-title /
  system), inline placement in expanded layout (renderProjectLine), and
  inline placement in compact layout (renderSessionLine).
- Document `display.showAdvisor` and `display.advisorOverride` in
  README.md and README.zh.md configuration tables.

`npm test`: 637 / 639 pass + 1 skip + 1 pre-existing unrelated failure
(`countConfigs cache: miss on nested rules additions`, fails identically
on `main`).

* feat: add optional autoCompactWindow context denominator

* feat: use autoCompactWindow as token-display denominator too

* release: prep v0.1.0

* docs: clarify plugin release versioning

* fix: validate autoCompactWindow as an integer

* build: compile dist/ [auto]

* build: compile dist/ [auto]

* build(deps-dev): bump @types/node from 25.9.1 to 25.9.2

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.1 to 25.9.2.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 25.9.2
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

* feat: add showSkills / showMcp HUD display elements

* feat(usage): render balance_label alongside stdin rate_limits

Previously, external-snapshot balance_label was shown only when stdin
rate_limits were absent (mutually exclusive). Rate-limit windows (5h/7d)
and a prepaid/metered dollar balance are complementary data, so both
should be visible when available.

Changes:
- index.ts: when stdin provides rate_limits, also read the external
  snapshot for a balance_label (only when externalUsagePath is set)
  and merge it onto usageData
- session-line.ts: render balance_label as an additive segment after
  the windows block rather than as an exclusive leading branch;
  standalone balance (no window data) preserves existing behaviour
- tests/render.test.js: add test for combined windows + balance display

Backward-compatible: users without externalUsagePath configured see
no behaviour change (no extra file read, no extra segment).

* fix: render external balance with usage windows

* fix: harden skills and mcp activity rendering

* build: compile dist/ [auto]

* fix: stabilize setup and bash tool targets

* build: compile dist/ [auto]

* release: 0.1.1

* fix(i18n): translate "token" as 词元 instead of 令牌 in Simplified Chinese

令牌 is the established Chinese rendering of "token" in the
authentication/security sense only (access token → 访问令牌). For the
NLP/LLM sense — the unit of text shown by the HUD's token counters —
the standard term is 词元, officially designated as the Chinese name
for the AI-domain term "token" by the China National Committee for
Terms in Sciences and Technologies on 2026-03-25.

dist/ is intentionally untouched; it is auto-compiled by the
build-dist workflow.

* docs: refresh CLAUDE.md architecture overview to match current source

The File Structure section predated many modules that now exist in src/
(skills-mcp-line, advisor, cost, external-usage, speed-tracker, i18n,
width handling, and the newer render/lines entries), and the opt-in
lines list was missing the Skills/MCP activity lines and advisor label.

Update both so agents reading CLAUDE.md get an accurate map of the
codebase. Docs only, no behavior change.

Co-Authored-By: Claude Fable 5 <[email protected]>

* feat: add fallback speed estimation via transcript file growth

When the model provider does not return output_tokens (e.g. non-standard
proxies), the HUD shows no token/s speed at all.

This adds a fallback that estimates speed from transcript file size
growth (~4 bytes/token heuristic) when output_tokens is unavailable.

The primary output_tokens path remains unchanged - the fallback only
activates when the provider skips the field.

* test: harden transcript speed fallback

* build: compile dist/ [auto]

* feat: show session compaction count behind opt-in showCompactions setting

* feat: add CLAUDE_HUD_DISABLE env var kill-switch to disable the HUD

* docs: sync configure.md with current config schema

The guided /configure command doc had drifted from src/config.ts. Several
display elements were added to the schema (and rendered by the HUD) without
being surfaced in the configure flow, so users had no guided way to toggle
them and the Element Mapping table was incomplete.

Missing elements now documented and added to the Turn On / Turn Off lists:
- showEffortLevel (reasoning level: low/medium/high/xhigh/max)
- showOutputStyle (current output style name)
- showCost (session cost)
- showSkills, showMcp (activity elements added in the skills/mcp commit)
- showMemoryUsage
- showPromptCache (+ promptCacheTtlSeconds)
- showClaudeCodeVersion
- showAddedDirs (+ addedDirsLayout)
- showResetLabel

Also:
- Element Mapping table now lists every show* boolean in the schema,
  including showModel / showContextBar (previously mislabeled "not
  configurable" — they are configurable booleans that default to true).
- Removed the internal contradiction where usageValue was listed both as an
  un-edited "advanced" setting and in the Element Mapping table.
- Expanded the advanced-settings note to cover the keys the guided flow
  preserves but does not edit (elementOrder, mergeGroups, modelFormat,
  autoCompactWindow, sevenDayThreshold, advisorOverride, externalUsage*, etc).
- Updated the Full preset definition to enable the newly documented elements.

No behavior change to the renderer; this is a documentation/UX sync so the
guided configure flow matches what the HUD can actually display.

* build(deps-dev): bump @types/node from 25.9.2 to 25.9.3

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.2 to 25.9.3.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 25.9.3
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

* docs: sync Chinese compactions config docs

* docs: cover remaining configure display options

* build: compile dist/ [auto]

* release: 0.2.0

* fix: merge seven-day usage from external snapshot

Merge missing seven-day usage from a configured external snapshot when stdin only supplies the five-hour window. Adds regression coverage for the partial-stdin fallback path. Verified locally with npm ci, npm run build, and node --test tests/index.test.js tests/external-usage.test.js.

* fix(security): require opt-in for extra command

Require CLAUDE_HUD_ALLOW_EXTRA_CMD before executing --extra-cmd, tighten cache file permissions, and document the local command trust boundary. Maintainer follow-up removed generated dist artifacts from the PR and added extra opt-in parsing coverage. Verified locally with npm ci, npm test, and node --test tests/extra-cmd.test.js. Fixes jarrodwatts#618.

* fix(cost): correct Opus 4.5 pricing

Update Opus 4.5+ local cost estimates to the official standard Anthropic rate of $5/M input and $25/M output while keeping older Opus 4.0/4.1 estimates on $15/M input and $75/M output. Adds focused coverage for Opus 4.5/4.6, Bedrock-style model IDs, and the older fallback. Verified locally with npm ci, npm run build, and node --test tests/core.test.js tests/render.test.js. GitHub CI passed on Node 18 and 20. Fixes jarrodwatts#622. Supersedes jarrodwatts#624.

* build: compile dist/ [auto]

* release: 0.2.1

Bump package, lockfile, plugin metadata, and marketplace metadata to 0.2.1. Document the jarrodwatts#617, jarrodwatts#619, and jarrodwatts#625 fixes in CHANGELOG.md. Verified locally with npm ci, npm run build, npm test, npm run test:coverage, version metadata checks, and dist/index.js presence. GitHub CI passed on Node 18 and 20.

* refactor: extract duplicated code into shared utilities (jarrodwatts#634)

- formatTokens: 3 identical copies → src/utils/format.ts
- formatContextValue: 2 identical copies → src/utils/format.ts
- hyperlink/getFileHref/safeHyperlink: 2 copies each → src/utils/hyperlinks.ts
- ANSI/control/bidi sanitization: 4 implementations → src/utils/sanitize.ts
- truncateDesc/truncateContent: 2 copies → src/utils/truncate.ts
- DEBUG flag inline checks: replaced with createDebug() from debug.ts

11 files updated to import from 4 new shared utility modules.
No behavioral changes; all existing tests pass.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* fix: harden cache file permissions and imports for security (jarrodwatts#632)

- speed-tracker: write cache files with 0o600 and directories with 0o700,
  matching the permission model used by all other cache writers (transcript,
  context-cache, config-cache, version). Previously used default umask,
  leaving token counts and timestamps world-readable on shared systems.

- transcript.ts, config-reader.ts: use node: protocol prefix for built-in
  module imports (fs, path, os, readline). Bare specifiers can be hijacked
  by a same-named package in node_modules; the node: prefix guarantees the
  Node.js built-in.

- external-usage: require externalUsagePath to be an absolute path before
  reading, consistent with the write-path validation in
  resolveSnapshotWritePath(). Relative paths are now silently ignored.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Improve error handling: add debug logging to silent catch blocks (jarrodwatts#633)

Replace empty catch {} blocks with catch (err) and debug() logging
across 11 source files. This makes errors visible when DEBUG=claude-hud:*
is set, while preserving the non-fatal fallback behavior required by a
statusline plugin that runs every ~300ms.

Files improved:
- stdin.ts: stream setup, JSON parse, and stream error handlers
- git.ts: branch, status, diff, ahead/behind, and remote URL commands
- transcript.ts: file stat, cache read/write, line parse, stream errors
- config.ts: config file load failure
- config-reader.ts: path comparison, sentinel stat, cache read/write
- context-cache.ts: cache read/write, sweep operations
- speed-tracker.ts: cache read/write, file size tracking, legacy cleanup
- version.ts: binary stat, cache read/write, exec, executable check
- external-usage.ts: snapshot comparison, directory check, write, read
- memory.ts: /proc/meminfo, vm_stat, and memory calculation
- effort.ts: parent process effort detection

All catch blocks now capture the error variable and pass it to the
existing createDebug() utility with contextual messages including
file paths and operation descriptions.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Add unit tests for modules with lowest coverage (jarrodwatts#635)

Add comprehensive tests targeting the 6 modules with the least test coverage:
- cost.ts: 87.64% → 100% statements, 75.6% → 100% branch
- identity.ts: 87.96% → 96.29% statements, 40.74% → 76.31% branch
- session-tokens.ts: 85.71% → 100% statements, 76.92% → 100% branch
- usage.ts: 85.95% → 97.94% statements, 82.29% → 88.88% branch
- memory.ts: 77.41% → 79.03% statements, 73.33% → 85.71% branch
- version.ts: 72.78% → 74.68% statements, 66.66% → 80.3% branch

Overall coverage improved from 93.56% → 94.6% statements, 84.54% → 86.23% branch.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* test(cost): clarify Bedrock test intent and add real-session coverage (jarrodwatts#636)

Add a comment explaining that the existing bedrockOpus46 test exercises
regex normalization (display_name only), not real Bedrock pricing behavior.

Add a complementary test confirming that real Bedrock sessions (with
model.id containing 'anthropic.claude-') correctly skip cost estimation
via the isBedrockModelId early return.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* feat: optionally show provider label before model

Add opt-in provider label placement for custom proxy workflows while preserving the existing default model badge order. Compact and expanded layouts now share the same badge formatter.

Closes jarrodwatts#620

* build: compile dist/ [auto]

* test: lock down usage and speed cache hardening

Document the absolute read-path requirement for external usage snapshots and add focused tests for relative read rejection plus private speed-cache directory and file modes.

* release: 0.3.0

Bump plugin, package, lockfile, and marketplace metadata to 0.3.0 and document the provider label plus cache-hardening changes.

* test: handle non-ASCII entrypoint paths (jarrodwatts#655)

* refactor(i18n): let locales own layout via a tiny interpolate() helper (jarrodwatts#647)

* feat: add opt-in showRoutedCost for Bedrock/Vertex cost (jarrodwatts#648)

Closes jarrodwatts#621

* build(deps-dev): bump @types/node from 25.9.3 to 26.1.1 (jarrodwatts#657)

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.3 to 26.1.1.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 26.1.1
  dependency-type: direct:development
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* fix: correct effort suffix placement and guard check inconsistencies (jarrodwatts#650)

* fix: correct effort suffix placement and guard check inconsistencies

the effort level tag (like "max") was showing up after the provider name
instead of after the model — so you'd see "Opus | Bedrock max" when it
should be "Opus max | Bedrock". small thing but it looked wrong.

also fixed a few spots where opt-in features (showDuration, showConfigCounts)
were using !== false checks instead of === true, which could show them
unexpectedly if the config object was partially constructed.

cleaned up duplicate debug namespaces too — two modules were both using
'context' and two were using 'config', which made debug output confusing.

* test: cover effort and opt-in render guards

---------

Co-authored-by: Jarrod Watts <[email protected]>

* feat(display): show auth method and logged-in account on the first line (jarrodwatts#652)

* feat(display): show auth method and logged-in account on the first line

Adds an opt-in segment at the end of the first HUD line showing the
subscription plan of the current login (e.g. "Claude Max 20x") and the
logged-in account, derived from the oauthAccount block that Claude Code
persists in {CLAUDE_CONFIG_DIR}.json.

New display options (all default off):
- display.showAuth: plan label, e.g. "Claude Max 20x" / "API Key"
- display.showAuthUser: email local part (fallback: displayName)
- display.authUserLength: truncation length, 0 = full (default 8)

* fix(auth): strip whole ANSI sequences via shared sanitizeDisplayText helper

Replaces the local codepoint-class regex with the existing
sanitizeDisplayText() util so CSI/OSC sequences are removed entirely
instead of leaving parameter bytes behind (evil\x1b[31m@… -> evil@…),
matching how other externally-sourced display text is sanitized.

* fix(auth): prefer active API key over stale OAuth profile

---------

Co-authored-by: BBleae <[email protected]>
Co-authored-by: Jarrod Watts <[email protected]>

* fix: Fix duplicate counting issue in assistant message usage statistics (jarrodwatts#646)

* fix: Fix duplicate counting issue in assistant message usage statistics

* Harden transcript usage deduplication

---------

Co-authored-by: yuanliu <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>

* build(deps-dev): bump typescript from 6.0.3 to 7.0.2 (jarrodwatts#656)

Bumps [typescript](https://github.com/microsoft/TypeScript) from 6.0.3 to 7.0.2.
- [Release notes](https://github.com/microsoft/TypeScript/releases)
- [Commits](https://github.com/microsoft/TypeScript/commits)

---
updated-dependencies:
- dependency-name: typescript
  dependency-version: 7.0.2
  dependency-type: direct:development
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Fix symlinked rules and compact cache tokens (jarrodwatts#659)

Co-authored-by: Codex Maintainer <[email protected]>

* build: compile dist/ [auto]

* feat(effort): show ultracode sessions as ultracode(xhigh) (jarrodwatts#640)

Claude Code reports ultracode as a normal level (xhigh) in stdin, so the
statusline recovers it from the transcript: the ultra_effort_enter /
ultra_effort_exit attachment markers and the immediate /effort command
output, latest signal wins. When active, the reported level is wrapped as
ultracode(<level>), taken as-is rather than assumed to be xhigh.

Also drop the ps-based parent-process --effort fallback (jarrodwatts#471); the effort
label now comes solely from Claude Code stdin, removing a subprocess spawn
on every statusline refresh.

Co-authored-by: Claude <[email protected]>

* build: compile dist/ [auto]

* feat(i18n): add Traditional Chinese (zh-Hant / zh-TW) locale (jarrodwatts#645)

- Add src/i18n/zh-Hant.ts with full Traditional Chinese translations
- Update Language type to include 'zh-Hant' and 'zh-TW'
- Register zh-Hant locale and zh-TW alias in i18n index
- Update validateLanguage() in config.ts to accept zh-Hant and zh-TW
- Fix session-time.ts: replace getLanguage() === 'zh' with isCjkLanguage()
  so zh-Hant also gets correct no-space formatting (e.g. '5m前' not '5m 前')
- Add test coverage for zh-Hant and zh-TW in i18n.test.js

Co-authored-by: YutenC <[email protected]>

* build: compile dist/ [auto]

* feat: add display.modelSource for proxy users (auto/stdin/transcript) (jarrodwatts#643)

* feat: add display.modelSource for proxy users (auto/stdin/transcript)

Users behind API proxies (cc-switch, LiteLLM, OpenRouter passthrough)
see the wrong model name because Claude Code reports the configured
model, not the one the proxy actually served.

This PR adds a `display.modelSource` config option:

- `auto` (default): uses stdin model for Claude models, transcript model
  for non-Claude. Zero impact on existing users.
- `stdin`: always use Claude Code's reported model (current behavior).
- `transcript`: always use the model from the API response's
  `message.model` field. Best for proxy users who want the real model.

Implementation:
- TranscriptData.lastAssistantModel: captures the actual model from
  each assistant message's `message.model` field
- stdin.resolveModelName(): resolves model based on modelSource config
- project.ts + session-line.ts: use resolveModelName instead of
  getModelName
- Cache serialization updated to persist lastAssistantModel

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* Harden transcript model display

---------

Co-authored-by: Claude Opus 4.7 (1M context) <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>

* build: compile dist/ [auto]

* release: 0.4.0 (jarrodwatts#660)

Co-authored-by: Codex Release <[email protected]>

---------

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: yjtdkj <[email protected]>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Matt Van Horn <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Kang Seong Jae <[email protected]>
Co-authored-by: Claude <[email protected]>
Co-authored-by: diaojz <[email protected]>
Co-authored-by: ViewWay <[email protected]>
Co-authored-by: John Costa <[email protected]>
Co-authored-by: tgoldblatt <[email protected]>
Co-authored-by: CDD-Jarvis <[email protected]>
Co-authored-by: jorgelaranjo <[email protected]>
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Longado <[email protected]>
Co-authored-by: kywoo26 <[email protected]>
Co-authored-by: Semianchuk Vitalii <[email protected]>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: 晴猫 <[email protected]>
Co-authored-by: BBleae <[email protected]>
Co-authored-by: yuanliu <[email protected]>
Co-authored-by: yuanliu <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>
Co-authored-by: Codex Maintainer <[email protected]>
Co-authored-by: hyunsh <[email protected]>
Co-authored-by: YutenC <[email protected]>
Co-authored-by: YutenC <[email protected]>
Co-authored-by: CctoctoFX <[email protected]>
Co-authored-by: Codex Release <[email protected]>
UnpxreTW added a commit to UnpxreTW/claude-hud that referenced this pull request Jul 15, 2026
* feat: add advisor model display (display.showAdvisor)

Surfaces the model configured via Claude Code's `/advisor` command so it
sits inline on the project line:

  [Opus 4.7 (1M context)] │ AthenaKB git:(main*) │ Advisor: Opus 4.7
  Context ██░░░░░░░░ 19%

## How it works

Claude Code stamps an `advisorModel` field onto every assistant record
in the session transcript when `/advisor` is configured. The transcript
parser already walks every line for tool / agent / todo state, so this
just captures the latest non-empty `advisorModel` into the existing
`TranscriptData` shape with no extra IO.

`renderProjectLine` then appends the prettified label (e.g.
`claude-opus-4-7` → `Opus 4.7`) as another part on the same row,
respecting the existing ` │ ` separator and ANSI handling. A
`display.showAdvisor` boolean (default false) gates the whole block;
`display.advisorOverride` lets users force a custom label if needed.

## What's included

- `TranscriptData.advisorModel` field + parser capture
- TRANSCRIPT_CACHE_VERSION bumped 5 → 6 to invalidate caches that
  predate the new field
- `display.showAdvisor` / `display.advisorOverride` config + validation
- `renderProjectLine` integration + `prettifyAdvisorId` helper that
  handles both canonical IDs and short aliases
- i18n labels for en (`Advisor`) and zh-Hans (`顾问`)
- `configure.md` Element Mapping + Turn Off/On flow entries
- 15 new tests covering ID prettification, render gating, transcript
  capture (latest-wins, empty-string-ignored, missing-field fallback),
  and config validation

## Notes

- `dist/` is intentionally untouched per CONTRIBUTING.md
- One pre-existing failure remains in `countConfigs cache: miss on
  nested rules additions` — it fails identically on main without any
  of my changes

* feat: add customLinePosition config to control custom line placement (jarrodwatts#575)

Adds display.customLinePosition so users can place customLine before the model badge while preserving the existing default placement.

* build: compile dist/ [auto]

* fix: stabilize context fallback and todo rendering (jarrodwatts#579)

Fixes context fallback for streaming zero frames and guards todo rendering when content is missing.\n\nCloses jarrodwatts#576.\nCloses jarrodwatts#577.

* build: compile dist/ [auto]

* fix(advisor): security hardening pass on advisor display

Addresses review feedback from jarrodwatts#573:

- Sanitize both `ctx.transcript.advisorModel` and `display.advisorOverride`
  through the existing `sanitizeDisplayText` helper before rendering so
  ANSI ESC, OSC, C0/C1 control, and bidi sequences cannot reach the
  terminal. Defence-in-depth: re-sanitize the prettified result too.
- Cap advisor display at 64 chars at the render layer
  (MAX_ADVISOR_DISPLAY_LEN), regardless of source.
- In the transcript parser, capture `advisorModel` only from records
  whose `type === 'assistant'` (matching the documented schema) and
  hard-cap the captured value at 64 chars (ADVISOR_MODEL_MAX_LEN) so a
  malformed log cannot persist an oversized value through the JSON
  cache. Bumps TRANSCRIPT_CACHE_VERSION 6 → 7. The cache deserializer
  also re-applies the cap as belt-and-braces against legacy entries.
- Add 9 new tests covering: CSI/OSC stripping (ANSI ESC + BEL + DEL
  bytes), bidi mark stripping (RLO/PDF), oversize transcript truncation,
  oversize override truncation (display-layer), sanitize-collapses-to-
  empty fallback, non-assistant record rejection (user / custom-title /
  system), inline placement in expanded layout (renderProjectLine), and
  inline placement in compact layout (renderSessionLine).
- Document `display.showAdvisor` and `display.advisorOverride` in
  README.md and README.zh.md configuration tables.

`npm test`: 637 / 639 pass + 1 skip + 1 pre-existing unrelated failure
(`countConfigs cache: miss on nested rules additions`, fails identically
on `main`).

* feat: add optional autoCompactWindow context denominator

* feat: use autoCompactWindow as token-display denominator too

* release: prep v0.1.0

* docs: clarify plugin release versioning

* fix: validate autoCompactWindow as an integer

* build: compile dist/ [auto]

* build: compile dist/ [auto]

* build(deps-dev): bump @types/node from 25.9.1 to 25.9.2

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.1 to 25.9.2.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 25.9.2
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

* feat: add showSkills / showMcp HUD display elements

* feat(usage): render balance_label alongside stdin rate_limits

Previously, external-snapshot balance_label was shown only when stdin
rate_limits were absent (mutually exclusive). Rate-limit windows (5h/7d)
and a prepaid/metered dollar balance are complementary data, so both
should be visible when available.

Changes:
- index.ts: when stdin provides rate_limits, also read the external
  snapshot for a balance_label (only when externalUsagePath is set)
  and merge it onto usageData
- session-line.ts: render balance_label as an additive segment after
  the windows block rather than as an exclusive leading branch;
  standalone balance (no window data) preserves existing behaviour
- tests/render.test.js: add test for combined windows + balance display

Backward-compatible: users without externalUsagePath configured see
no behaviour change (no extra file read, no extra segment).

* fix: render external balance with usage windows

* fix: harden skills and mcp activity rendering

* build: compile dist/ [auto]

* fix: stabilize setup and bash tool targets

* build: compile dist/ [auto]

* release: 0.1.1

* fix(i18n): translate "token" as 词元 instead of 令牌 in Simplified Chinese

令牌 is the established Chinese rendering of "token" in the
authentication/security sense only (access token → 访问令牌). For the
NLP/LLM sense — the unit of text shown by the HUD's token counters —
the standard term is 词元, officially designated as the Chinese name
for the AI-domain term "token" by the China National Committee for
Terms in Sciences and Technologies on 2026-03-25.

dist/ is intentionally untouched; it is auto-compiled by the
build-dist workflow.

* docs: refresh CLAUDE.md architecture overview to match current source

The File Structure section predated many modules that now exist in src/
(skills-mcp-line, advisor, cost, external-usage, speed-tracker, i18n,
width handling, and the newer render/lines entries), and the opt-in
lines list was missing the Skills/MCP activity lines and advisor label.

Update both so agents reading CLAUDE.md get an accurate map of the
codebase. Docs only, no behavior change.

Co-Authored-By: Claude Fable 5 <[email protected]>

* feat: add fallback speed estimation via transcript file growth

When the model provider does not return output_tokens (e.g. non-standard
proxies), the HUD shows no token/s speed at all.

This adds a fallback that estimates speed from transcript file size
growth (~4 bytes/token heuristic) when output_tokens is unavailable.

The primary output_tokens path remains unchanged - the fallback only
activates when the provider skips the field.

* test: harden transcript speed fallback

* build: compile dist/ [auto]

* feat: show session compaction count behind opt-in showCompactions setting

* feat: add CLAUDE_HUD_DISABLE env var kill-switch to disable the HUD

* docs: sync configure.md with current config schema

The guided /configure command doc had drifted from src/config.ts. Several
display elements were added to the schema (and rendered by the HUD) without
being surfaced in the configure flow, so users had no guided way to toggle
them and the Element Mapping table was incomplete.

Missing elements now documented and added to the Turn On / Turn Off lists:
- showEffortLevel (reasoning level: low/medium/high/xhigh/max)
- showOutputStyle (current output style name)
- showCost (session cost)
- showSkills, showMcp (activity elements added in the skills/mcp commit)
- showMemoryUsage
- showPromptCache (+ promptCacheTtlSeconds)
- showClaudeCodeVersion
- showAddedDirs (+ addedDirsLayout)
- showResetLabel

Also:
- Element Mapping table now lists every show* boolean in the schema,
  including showModel / showContextBar (previously mislabeled "not
  configurable" — they are configurable booleans that default to true).
- Removed the internal contradiction where usageValue was listed both as an
  un-edited "advanced" setting and in the Element Mapping table.
- Expanded the advanced-settings note to cover the keys the guided flow
  preserves but does not edit (elementOrder, mergeGroups, modelFormat,
  autoCompactWindow, sevenDayThreshold, advisorOverride, externalUsage*, etc).
- Updated the Full preset definition to enable the newly documented elements.

No behavior change to the renderer; this is a documentation/UX sync so the
guided configure flow matches what the HUD can actually display.

* build(deps-dev): bump @types/node from 25.9.2 to 25.9.3

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.2 to 25.9.3.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 25.9.3
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

* docs: sync Chinese compactions config docs

* docs: cover remaining configure display options

* build: compile dist/ [auto]

* release: 0.2.0

* fix: merge seven-day usage from external snapshot

Merge missing seven-day usage from a configured external snapshot when stdin only supplies the five-hour window. Adds regression coverage for the partial-stdin fallback path. Verified locally with npm ci, npm run build, and node --test tests/index.test.js tests/external-usage.test.js.

* fix(security): require opt-in for extra command

Require CLAUDE_HUD_ALLOW_EXTRA_CMD before executing --extra-cmd, tighten cache file permissions, and document the local command trust boundary. Maintainer follow-up removed generated dist artifacts from the PR and added extra opt-in parsing coverage. Verified locally with npm ci, npm test, and node --test tests/extra-cmd.test.js. Fixes jarrodwatts#618.

* fix(cost): correct Opus 4.5 pricing

Update Opus 4.5+ local cost estimates to the official standard Anthropic rate of $5/M input and $25/M output while keeping older Opus 4.0/4.1 estimates on $15/M input and $75/M output. Adds focused coverage for Opus 4.5/4.6, Bedrock-style model IDs, and the older fallback. Verified locally with npm ci, npm run build, and node --test tests/core.test.js tests/render.test.js. GitHub CI passed on Node 18 and 20. Fixes jarrodwatts#622. Supersedes jarrodwatts#624.

* build: compile dist/ [auto]

* release: 0.2.1

Bump package, lockfile, plugin metadata, and marketplace metadata to 0.2.1. Document the jarrodwatts#617, jarrodwatts#619, and jarrodwatts#625 fixes in CHANGELOG.md. Verified locally with npm ci, npm run build, npm test, npm run test:coverage, version metadata checks, and dist/index.js presence. GitHub CI passed on Node 18 and 20.

* refactor: extract duplicated code into shared utilities (jarrodwatts#634)

- formatTokens: 3 identical copies → src/utils/format.ts
- formatContextValue: 2 identical copies → src/utils/format.ts
- hyperlink/getFileHref/safeHyperlink: 2 copies each → src/utils/hyperlinks.ts
- ANSI/control/bidi sanitization: 4 implementations → src/utils/sanitize.ts
- truncateDesc/truncateContent: 2 copies → src/utils/truncate.ts
- DEBUG flag inline checks: replaced with createDebug() from debug.ts

11 files updated to import from 4 new shared utility modules.
No behavioral changes; all existing tests pass.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* fix: harden cache file permissions and imports for security (jarrodwatts#632)

- speed-tracker: write cache files with 0o600 and directories with 0o700,
  matching the permission model used by all other cache writers (transcript,
  context-cache, config-cache, version). Previously used default umask,
  leaving token counts and timestamps world-readable on shared systems.

- transcript.ts, config-reader.ts: use node: protocol prefix for built-in
  module imports (fs, path, os, readline). Bare specifiers can be hijacked
  by a same-named package in node_modules; the node: prefix guarantees the
  Node.js built-in.

- external-usage: require externalUsagePath to be an absolute path before
  reading, consistent with the write-path validation in
  resolveSnapshotWritePath(). Relative paths are now silently ignored.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Improve error handling: add debug logging to silent catch blocks (jarrodwatts#633)

Replace empty catch {} blocks with catch (err) and debug() logging
across 11 source files. This makes errors visible when DEBUG=claude-hud:*
is set, while preserving the non-fatal fallback behavior required by a
statusline plugin that runs every ~300ms.

Files improved:
- stdin.ts: stream setup, JSON parse, and stream error handlers
- git.ts: branch, status, diff, ahead/behind, and remote URL commands
- transcript.ts: file stat, cache read/write, line parse, stream errors
- config.ts: config file load failure
- config-reader.ts: path comparison, sentinel stat, cache read/write
- context-cache.ts: cache read/write, sweep operations
- speed-tracker.ts: cache read/write, file size tracking, legacy cleanup
- version.ts: binary stat, cache read/write, exec, executable check
- external-usage.ts: snapshot comparison, directory check, write, read
- memory.ts: /proc/meminfo, vm_stat, and memory calculation
- effort.ts: parent process effort detection

All catch blocks now capture the error variable and pass it to the
existing createDebug() utility with contextual messages including
file paths and operation descriptions.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Add unit tests for modules with lowest coverage (jarrodwatts#635)

Add comprehensive tests targeting the 6 modules with the least test coverage:
- cost.ts: 87.64% → 100% statements, 75.6% → 100% branch
- identity.ts: 87.96% → 96.29% statements, 40.74% → 76.31% branch
- session-tokens.ts: 85.71% → 100% statements, 76.92% → 100% branch
- usage.ts: 85.95% → 97.94% statements, 82.29% → 88.88% branch
- memory.ts: 77.41% → 79.03% statements, 73.33% → 85.71% branch
- version.ts: 72.78% → 74.68% statements, 66.66% → 80.3% branch

Overall coverage improved from 93.56% → 94.6% statements, 84.54% → 86.23% branch.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* test(cost): clarify Bedrock test intent and add real-session coverage (jarrodwatts#636)

Add a comment explaining that the existing bedrockOpus46 test exercises
regex normalization (display_name only), not real Bedrock pricing behavior.

Add a complementary test confirming that real Bedrock sessions (with
model.id containing 'anthropic.claude-') correctly skip cost estimation
via the isBedrockModelId early return.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* feat: optionally show provider label before model

Add opt-in provider label placement for custom proxy workflows while preserving the existing default model badge order. Compact and expanded layouts now share the same badge formatter.

Closes jarrodwatts#620

* build: compile dist/ [auto]

* test: lock down usage and speed cache hardening

Document the absolute read-path requirement for external usage snapshots and add focused tests for relative read rejection plus private speed-cache directory and file modes.

* release: 0.3.0

Bump plugin, package, lockfile, and marketplace metadata to 0.3.0 and document the provider label plus cache-hardening changes.

* test: handle non-ASCII entrypoint paths (jarrodwatts#655)

* refactor(i18n): let locales own layout via a tiny interpolate() helper (jarrodwatts#647)

* feat: add opt-in showRoutedCost for Bedrock/Vertex cost (jarrodwatts#648)

Closes jarrodwatts#621

* build(deps-dev): bump @types/node from 25.9.3 to 26.1.1 (jarrodwatts#657)

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.3 to 26.1.1.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 26.1.1
  dependency-type: direct:development
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* fix: correct effort suffix placement and guard check inconsistencies (jarrodwatts#650)

* fix: correct effort suffix placement and guard check inconsistencies

the effort level tag (like "max") was showing up after the provider name
instead of after the model — so you'd see "Opus | Bedrock max" when it
should be "Opus max | Bedrock". small thing but it looked wrong.

also fixed a few spots where opt-in features (showDuration, showConfigCounts)
were using !== false checks instead of === true, which could show them
unexpectedly if the config object was partially constructed.

cleaned up duplicate debug namespaces too — two modules were both using
'context' and two were using 'config', which made debug output confusing.

* test: cover effort and opt-in render guards

---------

Co-authored-by: Jarrod Watts <[email protected]>

* feat(display): show auth method and logged-in account on the first line (jarrodwatts#652)

* feat(display): show auth method and logged-in account on the first line

Adds an opt-in segment at the end of the first HUD line showing the
subscription plan of the current login (e.g. "Claude Max 20x") and the
logged-in account, derived from the oauthAccount block that Claude Code
persists in {CLAUDE_CONFIG_DIR}.json.

New display options (all default off):
- display.showAuth: plan label, e.g. "Claude Max 20x" / "API Key"
- display.showAuthUser: email local part (fallback: displayName)
- display.authUserLength: truncation length, 0 = full (default 8)

* fix(auth): strip whole ANSI sequences via shared sanitizeDisplayText helper

Replaces the local codepoint-class regex with the existing
sanitizeDisplayText() util so CSI/OSC sequences are removed entirely
instead of leaving parameter bytes behind (evil\x1b[31m@… -> evil@…),
matching how other externally-sourced display text is sanitized.

* fix(auth): prefer active API key over stale OAuth profile

---------

Co-authored-by: BBleae <[email protected]>
Co-authored-by: Jarrod Watts <[email protected]>

* fix: Fix duplicate counting issue in assistant message usage statistics (jarrodwatts#646)

* fix: Fix duplicate counting issue in assistant message usage statistics

* Harden transcript usage deduplication

---------

Co-authored-by: yuanliu <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>

* build(deps-dev): bump typescript from 6.0.3 to 7.0.2 (jarrodwatts#656)

Bumps [typescript](https://github.com/microsoft/TypeScript) from 6.0.3 to 7.0.2.
- [Release notes](https://github.com/microsoft/TypeScript/releases)
- [Commits](https://github.com/microsoft/TypeScript/commits)

---
updated-dependencies:
- dependency-name: typescript
  dependency-version: 7.0.2
  dependency-type: direct:development
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Fix symlinked rules and compact cache tokens (jarrodwatts#659)

Co-authored-by: Codex Maintainer <[email protected]>

* build: compile dist/ [auto]

* feat(effort): show ultracode sessions as ultracode(xhigh) (jarrodwatts#640)

Claude Code reports ultracode as a normal level (xhigh) in stdin, so the
statusline recovers it from the transcript: the ultra_effort_enter /
ultra_effort_exit attachment markers and the immediate /effort command
output, latest signal wins. When active, the reported level is wrapped as
ultracode(<level>), taken as-is rather than assumed to be xhigh.

Also drop the ps-based parent-process --effort fallback (jarrodwatts#471); the effort
label now comes solely from Claude Code stdin, removing a subprocess spawn
on every statusline refresh.

Co-authored-by: Claude <[email protected]>

* build: compile dist/ [auto]

* feat(i18n): add Traditional Chinese (zh-Hant / zh-TW) locale (jarrodwatts#645)

- Add src/i18n/zh-Hant.ts with full Traditional Chinese translations
- Update Language type to include 'zh-Hant' and 'zh-TW'
- Register zh-Hant locale and zh-TW alias in i18n index
- Update validateLanguage() in config.ts to accept zh-Hant and zh-TW
- Fix session-time.ts: replace getLanguage() === 'zh' with isCjkLanguage()
  so zh-Hant also gets correct no-space formatting (e.g. '5m前' not '5m 前')
- Add test coverage for zh-Hant and zh-TW in i18n.test.js

Co-authored-by: YutenC <[email protected]>

* build: compile dist/ [auto]

* feat: add display.modelSource for proxy users (auto/stdin/transcript) (jarrodwatts#643)

* feat: add display.modelSource for proxy users (auto/stdin/transcript)

Users behind API proxies (cc-switch, LiteLLM, OpenRouter passthrough)
see the wrong model name because Claude Code reports the configured
model, not the one the proxy actually served.

This PR adds a `display.modelSource` config option:

- `auto` (default): uses stdin model for Claude models, transcript model
  for non-Claude. Zero impact on existing users.
- `stdin`: always use Claude Code's reported model (current behavior).
- `transcript`: always use the model from the API response's
  `message.model` field. Best for proxy users who want the real model.

Implementation:
- TranscriptData.lastAssistantModel: captures the actual model from
  each assistant message's `message.model` field
- stdin.resolveModelName(): resolves model based on modelSource config
- project.ts + session-line.ts: use resolveModelName instead of
  getModelName
- Cache serialization updated to persist lastAssistantModel

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* Harden transcript model display

---------

Co-authored-by: Claude Opus 4.7 (1M context) <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>

* build: compile dist/ [auto]

* release: 0.4.0 (jarrodwatts#660)

Co-authored-by: Codex Release <[email protected]>

* fix weekly usage threshold default (jarrodwatts#662)

* build: compile dist/ [auto]

* release: 0.4.1 (jarrodwatts#663)

* fix: harden git refs and extra command labels (jarrodwatts#664)

Show detached tags and commits, preserve slash-separated branch links, and accept sanitized plain-text output for explicitly enabled extra commands. Sanitize compact Git refs and add focused regression coverage.

* build: compile dist/ [auto]

* release: 0.4.2 (jarrodwatts#665)

Publish the hardened Git ref and opt-in plain-output fixes from jarrodwatts#664 with aligned plugin, marketplace, package, lockfile, and changelog versions.

---------

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: yjtdkj <[email protected]>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Matt Van Horn <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Kang Seong Jae <[email protected]>
Co-authored-by: Claude <[email protected]>
Co-authored-by: diaojz <[email protected]>
Co-authored-by: ViewWay <[email protected]>
Co-authored-by: John Costa <[email protected]>
Co-authored-by: tgoldblatt <[email protected]>
Co-authored-by: CDD-Jarvis <[email protected]>
Co-authored-by: jorgelaranjo <[email protected]>
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Longado <[email protected]>
Co-authored-by: kywoo26 <[email protected]>
Co-authored-by: Semianchuk Vitalii <[email protected]>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: 晴猫 <[email protected]>
Co-authored-by: BBleae <[email protected]>
Co-authored-by: yuanliu <[email protected]>
Co-authored-by: yuanliu <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>
Co-authored-by: Codex Maintainer <[email protected]>
Co-authored-by: hyunsh <[email protected]>
Co-authored-by: YutenC <[email protected]>
Co-authored-by: YutenC <[email protected]>
Co-authored-by: CctoctoFX <[email protected]>
Co-authored-by: Codex Release <[email protected]>
Co-authored-by: hanboy587 <[email protected]>
UnpxreTW added a commit to UnpxreTW/claude-hud that referenced this pull request Jul 16, 2026
* feat: add advisor model display (display.showAdvisor)

Surfaces the model configured via Claude Code's `/advisor` command so it
sits inline on the project line:

  [Opus 4.7 (1M context)] │ AthenaKB git:(main*) │ Advisor: Opus 4.7
  Context ██░░░░░░░░ 19%

## How it works

Claude Code stamps an `advisorModel` field onto every assistant record
in the session transcript when `/advisor` is configured. The transcript
parser already walks every line for tool / agent / todo state, so this
just captures the latest non-empty `advisorModel` into the existing
`TranscriptData` shape with no extra IO.

`renderProjectLine` then appends the prettified label (e.g.
`claude-opus-4-7` → `Opus 4.7`) as another part on the same row,
respecting the existing ` │ ` separator and ANSI handling. A
`display.showAdvisor` boolean (default false) gates the whole block;
`display.advisorOverride` lets users force a custom label if needed.

## What's included

- `TranscriptData.advisorModel` field + parser capture
- TRANSCRIPT_CACHE_VERSION bumped 5 → 6 to invalidate caches that
  predate the new field
- `display.showAdvisor` / `display.advisorOverride` config + validation
- `renderProjectLine` integration + `prettifyAdvisorId` helper that
  handles both canonical IDs and short aliases
- i18n labels for en (`Advisor`) and zh-Hans (`顾问`)
- `configure.md` Element Mapping + Turn Off/On flow entries
- 15 new tests covering ID prettification, render gating, transcript
  capture (latest-wins, empty-string-ignored, missing-field fallback),
  and config validation

## Notes

- `dist/` is intentionally untouched per CONTRIBUTING.md
- One pre-existing failure remains in `countConfigs cache: miss on
  nested rules additions` — it fails identically on main without any
  of my changes

* feat: add customLinePosition config to control custom line placement (jarrodwatts#575)

Adds display.customLinePosition so users can place customLine before the model badge while preserving the existing default placement.

* build: compile dist/ [auto]

* fix: stabilize context fallback and todo rendering (jarrodwatts#579)

Fixes context fallback for streaming zero frames and guards todo rendering when content is missing.\n\nCloses jarrodwatts#576.\nCloses jarrodwatts#577.

* build: compile dist/ [auto]

* fix(advisor): security hardening pass on advisor display

Addresses review feedback from jarrodwatts#573:

- Sanitize both `ctx.transcript.advisorModel` and `display.advisorOverride`
  through the existing `sanitizeDisplayText` helper before rendering so
  ANSI ESC, OSC, C0/C1 control, and bidi sequences cannot reach the
  terminal. Defence-in-depth: re-sanitize the prettified result too.
- Cap advisor display at 64 chars at the render layer
  (MAX_ADVISOR_DISPLAY_LEN), regardless of source.
- In the transcript parser, capture `advisorModel` only from records
  whose `type === 'assistant'` (matching the documented schema) and
  hard-cap the captured value at 64 chars (ADVISOR_MODEL_MAX_LEN) so a
  malformed log cannot persist an oversized value through the JSON
  cache. Bumps TRANSCRIPT_CACHE_VERSION 6 → 7. The cache deserializer
  also re-applies the cap as belt-and-braces against legacy entries.
- Add 9 new tests covering: CSI/OSC stripping (ANSI ESC + BEL + DEL
  bytes), bidi mark stripping (RLO/PDF), oversize transcript truncation,
  oversize override truncation (display-layer), sanitize-collapses-to-
  empty fallback, non-assistant record rejection (user / custom-title /
  system), inline placement in expanded layout (renderProjectLine), and
  inline placement in compact layout (renderSessionLine).
- Document `display.showAdvisor` and `display.advisorOverride` in
  README.md and README.zh.md configuration tables.

`npm test`: 637 / 639 pass + 1 skip + 1 pre-existing unrelated failure
(`countConfigs cache: miss on nested rules additions`, fails identically
on `main`).

* feat: add optional autoCompactWindow context denominator

* feat: use autoCompactWindow as token-display denominator too

* release: prep v0.1.0

* docs: clarify plugin release versioning

* fix: validate autoCompactWindow as an integer

* build: compile dist/ [auto]

* build: compile dist/ [auto]

* build(deps-dev): bump @types/node from 25.9.1 to 25.9.2

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.1 to 25.9.2.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 25.9.2
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

* feat: add showSkills / showMcp HUD display elements

* feat(usage): render balance_label alongside stdin rate_limits

Previously, external-snapshot balance_label was shown only when stdin
rate_limits were absent (mutually exclusive). Rate-limit windows (5h/7d)
and a prepaid/metered dollar balance are complementary data, so both
should be visible when available.

Changes:
- index.ts: when stdin provides rate_limits, also read the external
  snapshot for a balance_label (only when externalUsagePath is set)
  and merge it onto usageData
- session-line.ts: render balance_label as an additive segment after
  the windows block rather than as an exclusive leading branch;
  standalone balance (no window data) preserves existing behaviour
- tests/render.test.js: add test for combined windows + balance display

Backward-compatible: users without externalUsagePath configured see
no behaviour change (no extra file read, no extra segment).

* fix: render external balance with usage windows

* fix: harden skills and mcp activity rendering

* build: compile dist/ [auto]

* fix: stabilize setup and bash tool targets

* build: compile dist/ [auto]

* release: 0.1.1

* fix(i18n): translate "token" as 词元 instead of 令牌 in Simplified Chinese

令牌 is the established Chinese rendering of "token" in the
authentication/security sense only (access token → 访问令牌). For the
NLP/LLM sense — the unit of text shown by the HUD's token counters —
the standard term is 词元, officially designated as the Chinese name
for the AI-domain term "token" by the China National Committee for
Terms in Sciences and Technologies on 2026-03-25.

dist/ is intentionally untouched; it is auto-compiled by the
build-dist workflow.

* docs: refresh CLAUDE.md architecture overview to match current source

The File Structure section predated many modules that now exist in src/
(skills-mcp-line, advisor, cost, external-usage, speed-tracker, i18n,
width handling, and the newer render/lines entries), and the opt-in
lines list was missing the Skills/MCP activity lines and advisor label.

Update both so agents reading CLAUDE.md get an accurate map of the
codebase. Docs only, no behavior change.

Co-Authored-By: Claude Fable 5 <[email protected]>

* feat: add fallback speed estimation via transcript file growth

When the model provider does not return output_tokens (e.g. non-standard
proxies), the HUD shows no token/s speed at all.

This adds a fallback that estimates speed from transcript file size
growth (~4 bytes/token heuristic) when output_tokens is unavailable.

The primary output_tokens path remains unchanged - the fallback only
activates when the provider skips the field.

* test: harden transcript speed fallback

* build: compile dist/ [auto]

* feat: show session compaction count behind opt-in showCompactions setting

* feat: add CLAUDE_HUD_DISABLE env var kill-switch to disable the HUD

* docs: sync configure.md with current config schema

The guided /configure command doc had drifted from src/config.ts. Several
display elements were added to the schema (and rendered by the HUD) without
being surfaced in the configure flow, so users had no guided way to toggle
them and the Element Mapping table was incomplete.

Missing elements now documented and added to the Turn On / Turn Off lists:
- showEffortLevel (reasoning level: low/medium/high/xhigh/max)
- showOutputStyle (current output style name)
- showCost (session cost)
- showSkills, showMcp (activity elements added in the skills/mcp commit)
- showMemoryUsage
- showPromptCache (+ promptCacheTtlSeconds)
- showClaudeCodeVersion
- showAddedDirs (+ addedDirsLayout)
- showResetLabel

Also:
- Element Mapping table now lists every show* boolean in the schema,
  including showModel / showContextBar (previously mislabeled "not
  configurable" — they are configurable booleans that default to true).
- Removed the internal contradiction where usageValue was listed both as an
  un-edited "advanced" setting and in the Element Mapping table.
- Expanded the advanced-settings note to cover the keys the guided flow
  preserves but does not edit (elementOrder, mergeGroups, modelFormat,
  autoCompactWindow, sevenDayThreshold, advisorOverride, externalUsage*, etc).
- Updated the Full preset definition to enable the newly documented elements.

No behavior change to the renderer; this is a documentation/UX sync so the
guided configure flow matches what the HUD can actually display.

* build(deps-dev): bump @types/node from 25.9.2 to 25.9.3

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.2 to 25.9.3.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 25.9.3
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

* docs: sync Chinese compactions config docs

* docs: cover remaining configure display options

* build: compile dist/ [auto]

* release: 0.2.0

* fix: merge seven-day usage from external snapshot

Merge missing seven-day usage from a configured external snapshot when stdin only supplies the five-hour window. Adds regression coverage for the partial-stdin fallback path. Verified locally with npm ci, npm run build, and node --test tests/index.test.js tests/external-usage.test.js.

* fix(security): require opt-in for extra command

Require CLAUDE_HUD_ALLOW_EXTRA_CMD before executing --extra-cmd, tighten cache file permissions, and document the local command trust boundary. Maintainer follow-up removed generated dist artifacts from the PR and added extra opt-in parsing coverage. Verified locally with npm ci, npm test, and node --test tests/extra-cmd.test.js. Fixes jarrodwatts#618.

* fix(cost): correct Opus 4.5 pricing

Update Opus 4.5+ local cost estimates to the official standard Anthropic rate of $5/M input and $25/M output while keeping older Opus 4.0/4.1 estimates on $15/M input and $75/M output. Adds focused coverage for Opus 4.5/4.6, Bedrock-style model IDs, and the older fallback. Verified locally with npm ci, npm run build, and node --test tests/core.test.js tests/render.test.js. GitHub CI passed on Node 18 and 20. Fixes jarrodwatts#622. Supersedes jarrodwatts#624.

* build: compile dist/ [auto]

* release: 0.2.1

Bump package, lockfile, plugin metadata, and marketplace metadata to 0.2.1. Document the jarrodwatts#617, jarrodwatts#619, and jarrodwatts#625 fixes in CHANGELOG.md. Verified locally with npm ci, npm run build, npm test, npm run test:coverage, version metadata checks, and dist/index.js presence. GitHub CI passed on Node 18 and 20.

* refactor: extract duplicated code into shared utilities (jarrodwatts#634)

- formatTokens: 3 identical copies → src/utils/format.ts
- formatContextValue: 2 identical copies → src/utils/format.ts
- hyperlink/getFileHref/safeHyperlink: 2 copies each → src/utils/hyperlinks.ts
- ANSI/control/bidi sanitization: 4 implementations → src/utils/sanitize.ts
- truncateDesc/truncateContent: 2 copies → src/utils/truncate.ts
- DEBUG flag inline checks: replaced with createDebug() from debug.ts

11 files updated to import from 4 new shared utility modules.
No behavioral changes; all existing tests pass.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* fix: harden cache file permissions and imports for security (jarrodwatts#632)

- speed-tracker: write cache files with 0o600 and directories with 0o700,
  matching the permission model used by all other cache writers (transcript,
  context-cache, config-cache, version). Previously used default umask,
  leaving token counts and timestamps world-readable on shared systems.

- transcript.ts, config-reader.ts: use node: protocol prefix for built-in
  module imports (fs, path, os, readline). Bare specifiers can be hijacked
  by a same-named package in node_modules; the node: prefix guarantees the
  Node.js built-in.

- external-usage: require externalUsagePath to be an absolute path before
  reading, consistent with the write-path validation in
  resolveSnapshotWritePath(). Relative paths are now silently ignored.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Improve error handling: add debug logging to silent catch blocks (jarrodwatts#633)

Replace empty catch {} blocks with catch (err) and debug() logging
across 11 source files. This makes errors visible when DEBUG=claude-hud:*
is set, while preserving the non-fatal fallback behavior required by a
statusline plugin that runs every ~300ms.

Files improved:
- stdin.ts: stream setup, JSON parse, and stream error handlers
- git.ts: branch, status, diff, ahead/behind, and remote URL commands
- transcript.ts: file stat, cache read/write, line parse, stream errors
- config.ts: config file load failure
- config-reader.ts: path comparison, sentinel stat, cache read/write
- context-cache.ts: cache read/write, sweep operations
- speed-tracker.ts: cache read/write, file size tracking, legacy cleanup
- version.ts: binary stat, cache read/write, exec, executable check
- external-usage.ts: snapshot comparison, directory check, write, read
- memory.ts: /proc/meminfo, vm_stat, and memory calculation
- effort.ts: parent process effort detection

All catch blocks now capture the error variable and pass it to the
existing createDebug() utility with contextual messages including
file paths and operation descriptions.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Add unit tests for modules with lowest coverage (jarrodwatts#635)

Add comprehensive tests targeting the 6 modules with the least test coverage:
- cost.ts: 87.64% → 100% statements, 75.6% → 100% branch
- identity.ts: 87.96% → 96.29% statements, 40.74% → 76.31% branch
- session-tokens.ts: 85.71% → 100% statements, 76.92% → 100% branch
- usage.ts: 85.95% → 97.94% statements, 82.29% → 88.88% branch
- memory.ts: 77.41% → 79.03% statements, 73.33% → 85.71% branch
- version.ts: 72.78% → 74.68% statements, 66.66% → 80.3% branch

Overall coverage improved from 93.56% → 94.6% statements, 84.54% → 86.23% branch.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* test(cost): clarify Bedrock test intent and add real-session coverage (jarrodwatts#636)

Add a comment explaining that the existing bedrockOpus46 test exercises
regex normalization (display_name only), not real Bedrock pricing behavior.

Add a complementary test confirming that real Bedrock sessions (with
model.id containing 'anthropic.claude-') correctly skip cost estimation
via the isBedrockModelId early return.

Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* feat: optionally show provider label before model

Add opt-in provider label placement for custom proxy workflows while preserving the existing default model badge order. Compact and expanded layouts now share the same badge formatter.

Closes jarrodwatts#620

* build: compile dist/ [auto]

* test: lock down usage and speed cache hardening

Document the absolute read-path requirement for external usage snapshots and add focused tests for relative read rejection plus private speed-cache directory and file modes.

* release: 0.3.0

Bump plugin, package, lockfile, and marketplace metadata to 0.3.0 and document the provider label plus cache-hardening changes.

* test: handle non-ASCII entrypoint paths (jarrodwatts#655)

* refactor(i18n): let locales own layout via a tiny interpolate() helper (jarrodwatts#647)

* feat: add opt-in showRoutedCost for Bedrock/Vertex cost (jarrodwatts#648)

Closes jarrodwatts#621

* build(deps-dev): bump @types/node from 25.9.3 to 26.1.1 (jarrodwatts#657)

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.3 to 26.1.1.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 26.1.1
  dependency-type: direct:development
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* fix: correct effort suffix placement and guard check inconsistencies (jarrodwatts#650)

* fix: correct effort suffix placement and guard check inconsistencies

the effort level tag (like "max") was showing up after the provider name
instead of after the model — so you'd see "Opus | Bedrock max" when it
should be "Opus max | Bedrock". small thing but it looked wrong.

also fixed a few spots where opt-in features (showDuration, showConfigCounts)
were using !== false checks instead of === true, which could show them
unexpectedly if the config object was partially constructed.

cleaned up duplicate debug namespaces too — two modules were both using
'context' and two were using 'config', which made debug output confusing.

* test: cover effort and opt-in render guards

---------

Co-authored-by: Jarrod Watts <[email protected]>

* feat(display): show auth method and logged-in account on the first line (jarrodwatts#652)

* feat(display): show auth method and logged-in account on the first line

Adds an opt-in segment at the end of the first HUD line showing the
subscription plan of the current login (e.g. "Claude Max 20x") and the
logged-in account, derived from the oauthAccount block that Claude Code
persists in {CLAUDE_CONFIG_DIR}.json.

New display options (all default off):
- display.showAuth: plan label, e.g. "Claude Max 20x" / "API Key"
- display.showAuthUser: email local part (fallback: displayName)
- display.authUserLength: truncation length, 0 = full (default 8)

* fix(auth): strip whole ANSI sequences via shared sanitizeDisplayText helper

Replaces the local codepoint-class regex with the existing
sanitizeDisplayText() util so CSI/OSC sequences are removed entirely
instead of leaving parameter bytes behind (evil\x1b[31m@… -> evil@…),
matching how other externally-sourced display text is sanitized.

* fix(auth): prefer active API key over stale OAuth profile

---------

Co-authored-by: BBleae <[email protected]>
Co-authored-by: Jarrod Watts <[email protected]>

* fix: Fix duplicate counting issue in assistant message usage statistics (jarrodwatts#646)

* fix: Fix duplicate counting issue in assistant message usage statistics

* Harden transcript usage deduplication

---------

Co-authored-by: yuanliu <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>

* build(deps-dev): bump typescript from 6.0.3 to 7.0.2 (jarrodwatts#656)

Bumps [typescript](https://github.com/microsoft/TypeScript) from 6.0.3 to 7.0.2.
- [Release notes](https://github.com/microsoft/TypeScript/releases)
- [Commits](https://github.com/microsoft/TypeScript/commits)

---
updated-dependencies:
- dependency-name: typescript
  dependency-version: 7.0.2
  dependency-type: direct:development
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build: compile dist/ [auto]

* Fix symlinked rules and compact cache tokens (jarrodwatts#659)

Co-authored-by: Codex Maintainer <[email protected]>

* build: compile dist/ [auto]

* feat(effort): show ultracode sessions as ultracode(xhigh) (jarrodwatts#640)

Claude Code reports ultracode as a normal level (xhigh) in stdin, so the
statusline recovers it from the transcript: the ultra_effort_enter /
ultra_effort_exit attachment markers and the immediate /effort command
output, latest signal wins. When active, the reported level is wrapped as
ultracode(<level>), taken as-is rather than assumed to be xhigh.

Also drop the ps-based parent-process --effort fallback (jarrodwatts#471); the effort
label now comes solely from Claude Code stdin, removing a subprocess spawn
on every statusline refresh.

Co-authored-by: Claude <[email protected]>

* build: compile dist/ [auto]

* feat(i18n): add Traditional Chinese (zh-Hant / zh-TW) locale (jarrodwatts#645)

- Add src/i18n/zh-Hant.ts with full Traditional Chinese translations
- Update Language type to include 'zh-Hant' and 'zh-TW'
- Register zh-Hant locale and zh-TW alias in i18n index
- Update validateLanguage() in config.ts to accept zh-Hant and zh-TW
- Fix session-time.ts: replace getLanguage() === 'zh' with isCjkLanguage()
  so zh-Hant also gets correct no-space formatting (e.g. '5m前' not '5m 前')
- Add test coverage for zh-Hant and zh-TW in i18n.test.js

Co-authored-by: YutenC <[email protected]>

* build: compile dist/ [auto]

* feat: add display.modelSource for proxy users (auto/stdin/transcript) (jarrodwatts#643)

* feat: add display.modelSource for proxy users (auto/stdin/transcript)

Users behind API proxies (cc-switch, LiteLLM, OpenRouter passthrough)
see the wrong model name because Claude Code reports the configured
model, not the one the proxy actually served.

This PR adds a `display.modelSource` config option:

- `auto` (default): uses stdin model for Claude models, transcript model
  for non-Claude. Zero impact on existing users.
- `stdin`: always use Claude Code's reported model (current behavior).
- `transcript`: always use the model from the API response's
  `message.model` field. Best for proxy users who want the real model.

Implementation:
- TranscriptData.lastAssistantModel: captures the actual model from
  each assistant message's `message.model` field
- stdin.resolveModelName(): resolves model based on modelSource config
- project.ts + session-line.ts: use resolveModelName instead of
  getModelName
- Cache serialization updated to persist lastAssistantModel

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* Harden transcript model display

---------

Co-authored-by: Claude Opus 4.7 (1M context) <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>

* build: compile dist/ [auto]

* release: 0.4.0 (jarrodwatts#660)

Co-authored-by: Codex Release <[email protected]>

* fix weekly usage threshold default (jarrodwatts#662)

* build: compile dist/ [auto]

* release: 0.4.1 (jarrodwatts#663)

* fix: harden git refs and extra command labels (jarrodwatts#664)

Show detached tags and commits, preserve slash-separated branch links, and accept sanitized plain-text output for explicitly enabled extra commands. Sanitize compact Git refs and add focused regression coverage.

* build: compile dist/ [auto]

* release: 0.4.2 (jarrodwatts#665)

Publish the hardened Git ref and opt-in plain-output fixes from jarrodwatts#664 with aligned plugin, marketplace, package, lockfile, and changelog versions.

* feat: render model-scoped weekly usage windows (jarrodwatts#669)

Parse and render bounded model-scoped weekly usage data from Claude Code stdin.

Harden untrusted labels and values, cover expanded and compact layouts, and prevent scoped-only data from overwriting shared external snapshots.

* build: compile dist/ [auto]

* build: prevent stale compiled artifacts (jarrodwatts#670)

Clean dist before TypeScript compilation and enforce source-to-artifact parity so deleted modules cannot remain directly package-addressable.

* build: compile dist/ [auto]

* release: 0.5.0 (jarrodwatts#671)

Ship model-scoped usage windows and prevent removed compiled modules from remaining in release packages.

---------

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: yjtdkj <[email protected]>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Matt Van Horn <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Kang Seong Jae <[email protected]>
Co-authored-by: Claude <[email protected]>
Co-authored-by: diaojz <[email protected]>
Co-authored-by: ViewWay <[email protected]>
Co-authored-by: John Costa <[email protected]>
Co-authored-by: tgoldblatt <[email protected]>
Co-authored-by: CDD-Jarvis <[email protected]>
Co-authored-by: jorgelaranjo <[email protected]>
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: Longado <[email protected]>
Co-authored-by: kywoo26 <[email protected]>
Co-authored-by: Semianchuk Vitalii <[email protected]>
Co-authored-by: Jarrod Watts <[email protected]>
Co-authored-by: 晴猫 <[email protected]>
Co-authored-by: BBleae <[email protected]>
Co-authored-by: yuanliu <[email protected]>
Co-authored-by: yuanliu <[email protected]>
Co-authored-by: Claude HUD Maintainer <[email protected]>
Co-authored-by: Codex Maintainer <[email protected]>
Co-authored-by: hyunsh <[email protected]>
Co-authored-by: YutenC <[email protected]>
Co-authored-by: YutenC <[email protected]>
Co-authored-by: CctoctoFX <[email protected]>
Co-authored-by: Codex Release <[email protected]>
Co-authored-by: hanboy587 <[email protected]>
Co-authored-by: Aster <[email protected]>
UnpxreTW added a commit to UnpxreTW/claude-hud that referenced this pull request Jul 21, 2026
* feat: add advisor model display (display.showAdvisor)

Surfaces the model configured via Claude Code's `/advisor` command so it
sits inline on the project line:

  [Opus 4.7 (1M context)] │ AthenaKB git:(main*) │ Advisor: Opus 4.7
  Context ██░░░░░░░░ 19%

## How it works

Claude Code stamps an `advisorModel` field onto every assistant record
in the session transcript when `/advisor` is configured. The transcript
parser already walks every line for tool / agent / todo state, so this
just captures the latest non-empty `advisorModel` into the existing
`TranscriptData` shape with no extra IO.

`renderProjectLine` then appends the prettified label (e.g.
`claude-opus-4-7` → `Opus 4.7`) as another part on the same row,
respecting the existing ` │ ` separator and ANSI handling. A
`display.showAdvisor` boolean (default false) gates the whole block;
`display.advisorOverride` lets users force a custom label if needed.

## What's included

- `TranscriptData.advisorModel` field + parser capture
- TRANSCRIPT_CACHE_VERSION bumped 5 → 6 to invalidate caches that
  predate the new field
- `display.showAdvisor` / `display.advisorOverride` config + validation
- `renderProjectLine` integration + `prettifyAdvisorId` helper that
  handles both canonical IDs and short aliases
- i18n labels for en (`Advisor`) and zh-Hans (`顾问`)
- `configure.md` Element Mapping + Turn Off/On flow entries
- 15 new tests covering ID prettification, render gating, transcript
  capture (latest-wins, empty-string-ignored, missing-field fallback),
  and config validation

## Notes

- `dist/` is intentionally untouched per CONTRIBUTING.md
- One pre-existing failure remains in `countConfigs cache: miss on
  nested rules additions` — it fails identically on main without any
  of my changes

* feat: add customLinePosition config to control custom line placement (jarrodwatts#575)

Adds display.customLinePosition so users can place customLine before the model badge while preserving the existing default placement.

* build: compile dist/ [auto]

* fix: stabilize context fallback and todo rendering (jarrodwatts#579)

Fixes context fallback for streaming zero frames and guards todo rendering when content is missing.\n\nCloses jarrodwatts#576.\nCloses jarrodwatts#577.

* build: compile dist/ [auto]

* fix(advisor): security hardening pass on advisor display

Addresses review feedback from jarrodwatts#573:

- Sanitize both `ctx.transcript.advisorModel` and `display.advisorOverride`
  through the existing `sanitizeDisplayText` helper before rendering so
  ANSI ESC, OSC, C0/C1 control, and bidi sequences cannot reach the
  terminal. Defence-in-depth: re-sanitize the prettified result too.
- Cap advisor display at 64 chars at the render layer
  (MAX_ADVISOR_DISPLAY_LEN), regardless of source.
- In the transcript parser, capture `advisorModel` only from records
  whose `type === 'assistant'` (matching the documented schema) and
  hard-cap the captured value at 64 chars (ADVISOR_MODEL_MAX_LEN) so a
  malformed log cannot persist an oversized value through the JSON
  cache. Bumps TRANSCRIPT_CACHE_VERSION 6 → 7. The cache deserializer
  also re-applies the cap as belt-and-braces against legacy entries.
- Add 9 new tests covering: CSI/OSC stripping (ANSI ESC + BEL + DEL
  bytes), bidi mark stripping (RLO/PDF), oversize transcript truncation,
  oversize override truncation (display-layer), sanitize-collapses-to-
  empty fallback, non-assistant record rejection (user / custom-title /
  system), inline placement in expanded layout (renderProjectLine), and
  inline placement in compact layout (renderSessionLine).
- Document `display.showAdvisor` and `display.advisorOverride` in
  README.md and README.zh.md configuration tables.

`npm test`: 637 / 639 pass + 1 skip + 1 pre-existing unrelated failure
(`countConfigs cache: miss on nested rules additions`, fails identically
on `main`).

* feat: add optional autoCompactWindow context denominator

* feat: use autoCompactWindow as token-display denominator too

* release: prep v0.1.0

* docs: clarify plugin release versioning

* fix: validate autoCompactWindow as an integer

* build: compile dist/ [auto]

* build: compile dist/ [auto]

* build(deps-dev): bump @types/node from 25.9.1 to 25.9.2

Bumps [@types/node](https://github.com/DefinitelyTyped/DefinitelyTyped/tree/HEAD/types/node) from 25.9.1 to 25.9.2.
- [Release notes](https://github.com/DefinitelyTyped/DefinitelyTyped/releases)
- [Commits](https://github.com/DefinitelyTyped/DefinitelyTyped/commits/HEAD/types/node)

---
updated-dependencies:
- dependency-name: "@types/node"
  dependency-version: 25.9.2
  dependency-type: direct:development
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

* feat: add showSkills / showMcp HUD display elements

* feat(usage): render balance_label alongside stdin rate_limits

Previously, external-snapshot balance_label was shown only when stdin
rate_limits were absent (mutually exclusive). Rate-limit windows (5h/7d)
and a prepaid/metered dollar balance are complementary data, so both
should be visible when available.

Changes:
- index.ts: when stdin provides rate_limits, also read the external
  snapshot for a balance_label (only when externalUsagePath is set)
  and merge it onto usageData
- session-line.ts: render balance_label as an additive segment after
  the windows block rather than as an exclusive leading branch;
  standalone balance (no window data) preserves existing behaviour
- tests/render.test.js: add test for combined windows + balance display

Backward-compatible: users without externalUsagePath configured see
no behaviour change (no extra file read, no extra segment).

* fix: render external balance with usage windows

* fix: harden skills and mcp activity rendering

* build: compile dist/ [auto]

* fix: stabilize setup and bash tool targets

* build: compile dist/ [auto]

* release: 0.1.1

* fix(i18n): translate "token" as 词元 instead of 令牌 in Simplified Chinese

令牌 is the established Chinese rendering of "token" in the
authentication/security sense only (access token → 访问令牌). For the
NLP/LLM sense — the unit of text shown by the HUD's token counters —
the standard term is 词元, officially designated as the Chinese name
for the AI-domain term "token" by the China National Committee for
Terms in Sciences and Technologies on 2026-03-25.

dist/ is intentionally untouched; it is auto-compiled by the
build-dist workflow.

* docs: refresh CLAUDE.md architecture overview to match current source

The File Structure section predated many modules that now exist in src/
(skills-mcp-line, advisor, cost, external-usage, speed-tracker, i18n,
width handling, and the newer render/lines entries), and the opt-in
lines list was missing the Skills/MCP activity lines and advisor label.

Update both so agents reading CLAUDE.md get an accurate map of the
codebase. Docs only, no behavior change.

Co-Authored-By: Claude Fable 5 <[email protected]>

* feat: add fallback speed estimation via transcript file growth

When the model provider does not return output_tokens (e.g. non-standard
proxies), the HUD shows no token/s speed at all.

This adds a fallback that estimates speed from transcript file size
growth (~4 bytes/token heuristic) when output_tokens is unavailable.

The primary output_tokens path remains unchanged - the fallback only
activates when the provider skips the field.

* test: harden transcript speed fallback

* build: compile dist/ [auto]

* feat: show session compaction count behind opt-in showCompactions setting

* feat: add CLAUDE_HUD_DISABLE env var kill-switch to disable the HUD

* docs: sync configure.md with current config schema

The guided /configure command doc had drifted from src/config.ts. Several
display elements were added to the schema (and rendered by the HUD) without
being surfaced in the configure flow, so users had no guided way to toggle
them and the Element Mapping table was incomplete.

Missing elements now documented and added to the Turn On / Turn Off lists:
- showEffortLevel (reasoning level: low/medium/high/xhigh/max)
- showOutputStyle (current output style name)
- showCost (session cost)
- showSkills, showMcp (activity elements added in the skills/mcp commit)
- showMemoryUsage
- showPromptCache (+ promptCacheTtlSeconds)
- showClaudeCodeVersion
- showAddedDirs (+ addedDirsLayout)
- showResetLabel

Also:
- Element Mapping table now lists every show* boolean in the schema,
  including showModel / showContextBar (previously mislabeled "not
  configurable" — they are configurable booleans that default to true).
- Removed the internal contradiction where usageValue was listed both as an
  un-edited "advanced" setting and in the Element Mapping table.
- Expanded the advanced-settings note to cover the keys the guided flow
  preserves but does not edit (elementOrder, mergeGroups, modelFormat,
  autoCompactWindow, sevenDayThreshold, advisorOverride, ex