Repository navigation
fix(learn): keep traceback tail in tool-error digest preview - #2596
Merged
chopratejas merged 1 commit intoJul 27, 2026
Merged
Conversation
The error preview in _format_tool_call sliced output[:200] head-only, so a traceback's root cause (ExceptionType: message), which sits at the tail, was discarded and the digest showed only the preamble. Truncate keeping both head and tail so the diagnosis survives. Fixes headroomlabs-ai#2590
tenderdeve
requested review from
DevanshiVyas,
JerrettDavis and
chopratejas
as code owners
July 27, 2026 09:46
Contributor
PR governanceThis PR follows the template and is marked ready for human review. |
|
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
Merged
chopratejas
added a commit
that referenced
this pull request
Jul 29, 2026
🤖 I have created a release *beep* *boop* --- <details><summary>0.33.0</summary> ## [0.33.0](v0.32.0...v0.33.0) (2026-07-29) ### Features * **lossless:** factor shared directory prefix in the grep search fold ([#2547](#2547)) ([7dc9a97](7dc9a97)) * **metrics:** record per-extension token savings ([#2371](#2371)) ([02eb90f](02eb90f)) * **opencode:** ship the transport plugin in pip installs ([#2601](#2601)) ([f54f04f](f54f04f)) * **opencode:** support Copilot subscription backend for headroom models ([#2441](#2441)) ([#2445](#2445)) ([9089e7f](9089e7f)) * **proxy/hooks:** run fold-only (stream-safe) turn hooks on streaming OpenAI chat ([#2549](#2549)) ([a6d4921](a6d4921)) * **proxy/savings:** aggregate tool-schema savings into Metrics + all reporting sinks ([#2546](#2546)) ([9f1ffef](9f1ffef)) * **proxy:** label GitHub Copilot traffic as "copilot" in the outcome… ([#2377](#2377)) ([d7a8cdb](d7a8cdb)) * **proxy:** make /v1/compress usable as a gateway/Kong sidecar ([#2458](#2458)) ([1329ed7](1329ed7)) * **proxy:** model-aware cold-prefix hook — reasoning compaction (Kimi/GLM) + cold recompaction (CC) ([#2555](#2555)) ([cb8f4b6](cb8f4b6)) * **proxy:** route selected external compressors through the content router ([#2388](#2388)) ([e3c7964](e3c7964)) * **proxy:** select built-in compressors via --compressor + registry inventory ([#2373](#2373)) ([56c7d4a](56c7d4a)) * **rust:** add structured prose offload plumbing ([#334](#334)) ([#2378](#2378)) ([9e07785](9e07785)) * **rust:** port CodeCompressor AST compressor to Rust (parity-only) ([#1154](#1154)) ([e530de5](e530de5)) * **rust:** port Kompress ML prose compressor to Rust (parity-only) ([#1153](#1153)) ([83e27e5](83e27e5)) * **telemetry:** record provider cache read/write/uncached tokens per request ([#2450](#2450)) ([bec4cce](bec4cce)) * **transforms:** add compressed signal + dispatch code_aware/html/diff via registry ([#2400](#2400)) ([7ebda67](7ebda67)) * **transforms:** add pluggable compressor registry + headroom.compressor entry point ([#2370](#2370)) ([a02073e](a02073e)) * **transforms:** dispatch kompress/text via the compressor registry + forward question ([#2411](#2411)) ([446ec26](446ec26)) * **transforms:** dispatch smart_crusher via the compressor registry (defer kompress/text ML boundary) ([#2404](#2404)) ([7c7bf43](7c7bf43)) * **transforms:** make built-in compressors real Compressor implementations (adapters) ([#2391](#2391)) ([981616c](981616c)) * **wrap:** boost Serena — symbol-first guidance, wrap-time pre-index, repo-language scoping ([#2425](#2425)) ([fd0e1a8](fd0e1a8)) * **wrap:** default code-memory to Serena (dashboard browser off) behind unified --code-memory ([#2413](#2413)) ([6e4425a](6e4425a)) * **wrap:** reduce-at-source — SAFE quiet-CLI env defaults for the launched agent ([#2548](#2548)) ([c990cfb](c990cfb)) ### Bug Fixes * **backends/litellm:** guard None completion_tokens in usage mapping ([#2322](#2322)) ([44a174f](44a174f)) * **backends:** don't crash the OpenAI->Anthropic converter on empty choices ([#2484](#2484)) ([43a7b57](43a7b57)) * **cache:** preserve cache_control ttl when re-anchoring a breakpoint ([#2651](#2651)) ([e0d2cd0](e0d2cd0)) * **cache:** preserve client cache_control ttl when consolidating breakpoints ([#2382](#2382)) ([8906d3a](8906d3a)) * **ccr:** guard empty/malformed OpenAI choices in _extract_assistant_message ([#2389](#2389)) ([89319fb](89319fb)) * **ccr:** sliding idle-window TTL with max-lifetime ceiling in the Rust core backends ([#2604](#2604)) ([#2631](#2631)) ([e825588](e825588)) * **ci:** align Ruff tooling versions ([#2406](#2406)) ([2bb14d1](2bb14d1)) * **cli:** warn when Headroom proxy URL leaks into the shell after unwrap claude ([#2238](#2238)) ([#2571](#2571)) ([904bc67](904bc67)) * **codex:** detect keyring-backed ChatGPT auth ([#2478](#2478)) ([46293f4](46293f4)) * **compression:** report source-line span in CCR compression marker ([#2597](#2597)) ([18e1c3c](18e1c3c)) * **copilot:** derive GHE credential host from API URL ([#800](#800)) ([#2511](#2511)) ([4a8157f](4a8157f)) * **copilot:** normalize subscription API routing ([#2441](#2441)) ([#2455](#2455)) ([2eca5ee](2eca5ee)) * **copilot:** preserve /v1 for the Anthropic /v1/messages endpoint ([#2409](#2409)) ([#2414](#2414)) ([c400f90](c400f90)) * **deps:** bump mcp to 1.28.1 to clear 3 high-severity CVEs ([#2348](#2348)) ([a90be94](a90be94)) * **grok:** preserve business-seat auth while routing only inference ([#2514](#2514)) ([e4076bb](e4076bb)) * **image:** reuse image models instead of rebuilding them per request ([#2513](#2513)) ([#2536](#2536)) ([2a63ec7](2a63ec7)) * **install:** carry upstream-routing env overrides into supervised deployments ([#2429](#2429)) ([170b04a](170b04a)) * **install:** default to cache mode, matching `headroom proxy` ([#1893](#1893) follow-up) ([#2563](#2563)) ([b121223](b121223)) * **install:** migrate deployments off the retired chopratejas image repo ([#2427](#2427)) ([17ff13c](17ff13c)) * **install:** use CREATE_NO_WINDOW instead of DETACHED_PROCESS on Windows ([#2527](#2527)) ([045f3df](045f3df)) * **kompress:** raise the default execution-slot wait ([#2456](#2456)) ([5bd2266](5bd2266)) * **learn:** detect the active OpenCode database ([#2587](#2587)) ([f74d874](f74d874)) * **learn:** keep traceback tail in tool-error digest preview ([#2596](#2596)) ([85e8699](85e8699)) * **learn:** treat unreadable candidate paths as absent in project decode ([#2446](#2446)) ([a09ba6c](a09ba6c)) * **mcp:** pin mcp dependency to <2.0.0 to prevent server startup crash ([#2642](#2642)) ([b3f016b](b3f016b)) * **proxy/cost:** count Gemini thinking tokens in output usage ([#2639](#2639)) ([22b707f](22b707f)) * **proxy/cost:** record each request's savings exactly once (drop 3 double-counts) ([#2545](#2545)) ([0845b26](0845b26)) * **proxy/cost:** warn once per model when pricing lookup fails ([#2504](#2504)) ([#2535](#2535)) ([fa47637](fa47637)) * **proxy/gemini:** None-guard token counts from usageMetadata ([#2347](#2347)) ([f64aac9](f64aac9)) * **proxy/gemini:** tolerate malformed parts on the compression path ([#2486](#2486)) ([07cf547](07cf547)) * **proxy/metrics:** move the savings-ledger append off the event loop ([#2439](#2439)) ([4aac068](4aac068)) * **proxy/openai:** cache under looked-up messages ([#2420](#2420)) ([7052d52](7052d52)) * **proxy/openai:** don't record Codex WS savings without input accounting ([#2493](#2493)) ([2195ba7](2195ba7)) * **proxy/openai:** feed chat/completions traffic into the traffic learner ([#2333](#2333)) ([6cdfd3f](6cdfd3f)) * **proxy/openai:** None-guard usage token counts on the chat path ([#2431](#2431)) ([313c290](313c290)) * **proxy/openai:** replay incremental events in buffered Responses SSE ([#2410](#2410)) ([#2415](#2415)) ([0cbc0e8](0cbc0e8)) * **proxy/output-shaping:** tolerate a non-string system block text in steering ([#2435](#2435)) ([3e97671](3e97671)) * **proxy/perf:** count turn-hook message folds in token accounting ([#2520](#2520)) ([c371d5a](c371d5a)) * **proxy/perf:** tokenizer-consistent token accounting + surface tool-schema savings ([#2542](#2542)) ([1cc53c9](1cc53c9)) * **proxy/streaming:** tolerate malformed content in _response_to_sse ([#2481](#2481)) ([77b26c0](77b26c0)) * **proxy:** keep buffered CCR streams alive ([#2479](#2479)) ([a2e42fb](a2e42fb)) * **proxy:** keep core tools and the client's ToolSearch resident for PascalCase clients ([#2647](#2647)) ([1d29738](1d29738)) * **proxy:** offload OpenAI and Gemini tokenizer counting off the event loop ([#2498](#2498)) ([806d2e4](806d2e4)) * **proxy:** promote Kompress health after runtime load ([#2402](#2402)) ([54526bc](54526bc)) * **proxy:** reassemble server_tool_use.input from streamed partial_json ([#2449](#2449)) ([8c8fae0](8c8fae0)) * **proxy:** report deferred Kompress status and promote health from cache ([#2564](#2564)) ([d50cfab](d50cfab)) * **proxy:** skip max_tokens rename for backend-routed openai chat ([#2401](#2401)) ([d6a1af4](d6a1af4)) * **release:** publish Windows wheel + sdist (disable PyPI attestations, [#112](#112)) ([#2405](#2405)) ([f9cbdd6](f9cbdd6)) * **release:** sync generated version metadata on the release branch ([#2659](#2659)) ([5383c6b](5383c6b)) * **rust:** port CJK-aware relevance-query matching to CodeCompressor ([#2634](#2634)) ([e86c639](e86c639)) * **security:** exclude compromised ast-grep-cli 0.44.1 (supply-chain trojan) ([#2342](#2342)) ([494fb5a](494fb5a)) * **tokenizers:** price Claude against a real BPE (tiktoken o200k) not a char estimate ([#2543](#2543)) ([285176b](285176b)) * **transforms/cross-turn-dedup:** don't renumber-fold zero-padded line prefixes ([#2369](#2369)) ([f4070c4](f4070c4)) * **transforms/kompress-remote:** keep compress fail-open on malformed 200 ([#2320](#2320)) ([b759990](b759990)) * **wrap:** emit bare dotted keys for Codex --config overrides ([#2383](#2383)) ([f57e959](f57e959)) * **wrap:** make RTK opt-in (off by default) across wrap subcommands ([#2344](#2344)) ([44136ed](44136ed)) * **wrap:** skip Serena project setup outside real project roots ([#2574](#2574)) ([0994ea0](0994ea0)) * **wrap:** stop same-port persistent routing during claude unwrap ([#2340](#2340)) ([#2350](#2350)) ([cf5fa64](cf5fa64)) ### Performance Improvements * **content_router:** dedupe content detection ([#2419](#2419)) ([9b016f2](9b016f2)) ### Dependencies * bump the cargo-minor-patch group with 10 updates ([#2284](#2284)) ([3266ed7](3266ed7)) * bump the npm-minor-patch group across 3 directories with 7 updates ([#2276](#2276)) ([961866b](961866b)) ### Code Refactoring * **transforms:** dispatch simple built-in strategies via the compressor registry ([#2399](#2399)) ([fc9c63f](fc9c63f)) * **wrap:** retire tokensave; Serena is the code-memory MCP ([#2499](#2499)) ([5d23a0a](5d23a0a)) </details> --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
_format_tool_callinheadroom/learn/analyzer.pybuilt the error preview with a head-only slice —tc.output[:200]. For tracebacks the root cause (ExceptionType: message) is at the tail, so the digest showed onlyTraceback (most recent call last):plus the first frame and dropped the actual diagnosis. The issue reports 46% of 715 measured errors were truncated past the 200-char head.Closes #2590
Type of Change
Changes Made
_truncate_head_tail()helper that collapses newlines and, when over budget, keeps both the head and the tail joined by…._format_tool_callnow uses it for error output so the exception line survives truncation. Short errors are returned unchanged (no marker).Testing
pytest)ruff check .)mypy headroom)Test Output
Real Behavior Proof
KeyError: 'the-actual-root-cause'at the tail) as a failing tool call and built the digest.TracebackandKeyError: 'the-actual-root-cause', separated by…; short errors have no….Review Readiness
Checklist
CHANGELOG.mdAdditional Notes
Truncation budget stays at 200 chars (now split head/tail). mypy not run locally; happy to adjust if CI flags anything.