Remove h5i claims, rewind, and rollback commands/features#175
Conversation
Rip out three commands and their backends:
- claims: module, CLI, ClaimsCommands enum, SessionStart live-claims
injection + $H5I_CLAIMS_FREQUENCY hint, capture/recall routing, the
h5i_claims_{add,list,prune} MCP tools, doctor validate_claim_files,
and the claims sections in embedded agent instructions + .claude docs.
- rewind: command, repository::rewind backend + tests, h5i_rewind MCP tool.
- rollback: command + rollback-only backends revert_commit,
list_ai_commits, find_matching_commit.
Build, clippy (-D warnings), and tests all green.
Follow-up to f02294e (feature removal). Update all reference docs and the marketing site so nothing advertises the removed commands: - MANUAL.md: drop the h5i rollback/rewind/claims sections, MCP tool + env-var (H5I_CLAIMS_FREQUENCY) + refs/h5i/shadow entries, and TOC/ migration/noun-table rows; regenerate docs/manual/index.html. - man/man1/h5i.1: remove rewind/rollback SS sections, claim(s) from command lists, h5i_rewind from MCP list, and the example blocks. - docs site: delete the 3 dedicated posts (intent-based-rollback, content-addressed-claims, reduce-claude-token-costs) + orphan claims-merkle.svg, fix sitemap/blog-index/inbound links, and scrub the claims '51% cost' narrative from homepage, features, and pitch.
vibe is one implementation (h5i_core::vibe, Commands::Vibe) that was exposed under both 'recall vibe' and 'audit vibe'. Since vibe is an AI-footprint audit, keep it under 'audit' and remove the duplicate 'recall' surface: drop the (recall, vibe) route + recall verb-list entry, the recall NounVerb help entry, and point the legacy hint at 'h5i audit vibe'. Docs: rename the MANUAL/man detail section to 'h5i audit vibe' (regenerate docs/manual), fix TOC/migration/verb tables, and update two site pages. Build, clippy, tests green.
Reorganize the manual so commands no longer scatter. Group strictly by the noun-verb CLI model with consistent heading levels (group = H2, command = H3, subcommand = H4): Core (init, resolve) · capture · recall · objects · audit · share · msg · env · hook · serve · mcp · appendices Fixes the mis-nesting where verb sections lived under the wrong group: - audit review/scan moved out of recall notes/context into audit - capture memory moved into capture; share memory push/pull into share - share push/pull lifted up beside share pr (were 2000 lines away) - hook claude/codex demoted under hook - init/resolve grouped into a new Core commands section Pure reorder + re-level: all body prose preserved verbatim (verified by byte-identical non-heading multiset). Rebuilt the Table of Contents and also fixed 7 internal anchors that were dangling in the original — the rendered docs/manual/index.html now has zero broken #links.
🪙 80% AI-authored · 4 files · prompt 28/100
💬 The ask
📍 What shipped
Note 🪴 Prompt maturity: 28/100 · developing · 3 prompts scored (75% of AI commits) 📊 heuristic breakdown
Readability (display-only): Flesch 83 · FK grade 4.0 · Gunning Fog 5.8 Offline classical-NLP score — no LLM. Higher = more specific, bounded, and grounded prompting. Tip ✅ Security scan cleanNo credentials leaked in this branch's diff. h5i scanned every added line against the secret rule pack and found nothing to rotate. Note Duplicate-code scan clean — no copy-paste blocks introduced. 🧠 Reasoning by file — 8049 nodes across 6 lanes, latest 24 only%%{init: {"flowchart": {"nodeSpacing": 42, "rankSpacing": 48, "diagramPadding": 14}, "themeVariables": {"fontSize": "18px"}} }%%
flowchart TB
subgraph lane_0["💭 Reasoning · 6 steps"]
direction LR
n_lane_0_0["THINK · Let me examine the MANUAL/man structure to see if there's a separate `au…"]:::t
n_lane_0_1["THINK · Let me gather the remaining specifics (TOC, tables, man page structure) …"]:::t
n_lane_0_2["THINK · Let me run the e2e noun-workflow integration test (it exercises noun rou…"]:::t
n_lane_0_3["THINK · When it completes I'll review the new structure and verification results…"]:::t
n_lane_0_4["THINK · I'll use the gh-slug scheme consistently (the TOC already uses it, and i…"]:::t
n_lane_0_5["THINK · Let me do a final quality review of the new structure — the TOC and the …"]:::t
n_lane_0_0 --> n_lane_0_1
n_lane_0_1 --> n_lane_0_2
n_lane_0_2 --> n_lane_0_3
n_lane_0_3 --> n_lane_0_4
n_lane_0_4 --> n_lane_0_5
end
subgraph lane_1["📄 MANUAL.md · 8 ops"]
direction LR
n_lane_1_0["READ"]:::o
n_lane_1_1["EDIT"]:::a
n_lane_1_2["READ"]:::o
n_lane_1_3["EDIT × 2"]:::a
n_lane_1_4["READ"]:::o
n_lane_1_5["EDIT × 2"]:::a
n_lane_1_0 --> n_lane_1_1
n_lane_1_1 --> n_lane_1_2
n_lane_1_2 --> n_lane_1_3
n_lane_1_3 --> n_lane_1_4
n_lane_1_4 --> n_lane_1_5
end
subgraph lane_2["📄 man/man1/h5i.1 · 4 ops"]
direction LR
n_lane_2_0["READ"]:::o
n_lane_2_1["EDIT"]:::a
n_lane_2_2["READ"]:::o
n_lane_2_3["EDIT"]:::a
n_lane_2_0 --> n_lane_2_1
n_lane_2_1 --> n_lane_2_2
n_lane_2_2 --> n_lane_2_3
end
subgraph lane_3["📄 docs/guides/ai-code-provenance/index.html · 3 ops"]
direction LR
n_lane_3_0["READ"]:::o
n_lane_3_1["EDIT × 2"]:::a
n_lane_3_0 --> n_lane_3_1
end
subgraph lane_4["📄 docs/blog/from-git-blame-to-ai-blame/index.html · 2 ops"]
direction LR
n_lane_4_0["READ"]:::o
n_lane_4_1["EDIT"]:::a
n_lane_4_0 --> n_lane_4_1
end
subgraph lane_5["🗂 Other · 1 node"]
direction LR
n_lane_5_0["READ · read headings.txt"]:::o
end
classDef o fill:#dbeafe,stroke:#1e3a8a,color:#0b1c4a;
classDef t fill:#fef3c7,stroke:#92400e,color:#3f2d05;
classDef a fill:#dcfce7,stroke:#166534,color:#0a2e16;
classDef n fill:#ede9fe,stroke:#5b21b6,color:#221251;
classDef m fill:#e5e7eb,stroke:#374151,color:#0b0f17;
Note 🪙 Token reduction — 6 captured tool outputs kept out of context: 21408 → 3047 tokens (85% saved, 18361 tokens). Full output is recoverable with By tool
📜 Per-commit provenance
|
No description provided.