docs: split CLAUDE.md's Cold Storage and Closed Backlog into non-imported files - #403
Conversation
…rted files CLAUDE.md is loaded in full every session, and its Closed Backlog section alone was over 60% of the file's ~4200 lines despite being pure historical record with no forward-looking action attached to any entry. Moved Cold Storage and Closed Backlog (plus the now-fully-closed Active Backlog item 7) into two new files, docs/agent-closed-backlog.md and docs/agent-cold-storage.md, leaving short pointer stubs in CLAUDE.md. Unlike docs/agent-lessons-learned.md / agent-interconnect.md / agent-ndjson.md, these two new files are deliberately NOT referenced via CLAUDE.md's @import mechanism, so they are read on demand rather than loaded into every session's context. CLAUDE.md shrinks from ~4200 to ~1065 lines with zero content loss (verified via substring-containment checks against the pre-move file). Item 7's closure was independently re-confirmed before moving it. Updated the three files that referenced item 7's literal "7." marker as a permanent MD029 lint exception (.markdownlint-cli2.jsonc, AGENTS.md, tools/run_sanity_sweep.sh) to reflect that the exception no longer applies -- markdownlint-cli2 on CLAUDE.md now reports 0 issues, confirmed locally. Updated two cross-references in docs/agent-ndjson.md and docs/agent-scratchlog.md that pointed at item 7's old CLAUDE.md location. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A4iE1BRSkUETwz237XeuTW
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: ASSERTIVE Plan: Pro Plus Run ID: 📒 Files selected for processing (2)
📜 Recent review details🧰 Additional context used📓 Path-based instructions (4)**/*📄 CodeRabbit inference engine (AGENTS.md)
Files:
**/*.{md,markdown}📄 CodeRabbit inference engine (AGENTS.md)
Files:
**/*.{bat,cmd,ps1,py,yml,yaml,json,md}📄 CodeRabbit inference engine (CLAUDE.md)
Files:
**/*.{bat,cmd,ps1,py,md}📄 CodeRabbit inference engine (CLAUDE.md)
Files:
🧠 Learnings (1)📓 Common learnings🔇 Additional comments (2)
📝 WalkthroughSummary by CodeRabbit
WalkthroughThe change moves closed backlog history and shelved ideas out of ChangesDocumentation archive and lint baseline
Estimated code review effort: 2 (Simple) | ~10 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docs/agent-closed-backlog.md`:
- Around line 3-5: Keep every path literal in docs/agent-closed-backlog.md on a
single source line within its backtick-delimited code span, including the
references near the top and the condabin\conda.bat path near the later section;
wrap surrounding prose outside the inline code span when necessary.
In `@docs/agent-cold-storage.md`:
- Around line 3-4: Keep the inline-code path `docs/agent-lessons-learned.md`
contiguous on a single Markdown line in the opening paragraph of
`docs/agent-cold-storage.md`, without changing the surrounding wording.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 62ff70f2-2f9b-4958-973d-3b2c0f202433
📒 Files selected for processing (8)
.markdownlint-cli2.jsoncAGENTS.mdCLAUDE.mddocs/agent-closed-backlog.mddocs/agent-cold-storage.mddocs/agent-ndjson.mddocs/agent-scratchlog.mdtools/run_sanity_sweep.sh
📜 Review details
⏰ Context from checks skipped due to timeout. (9)
- GitHub Check: auto_merge
- GitHub Check: Batch syntax/run check (justme-test)
- GitHub Check: Batch syntax/run check (contract-uv-fail)
- GitHub Check: Batch syntax/run check (contract-uv)
- GitHub Check: Batch syntax/run check (cache)
- GitHub Check: Batch syntax/run check (conda-full)
- GitHub Check: Batch syntax/run check (uv)
- GitHub Check: Batch syntax/run check (uv-dl-fallback)
- GitHub Check: Batch syntax/run check (real)
🧰 Additional context used
📓 Path-based instructions (5)
**/*
📄 CodeRabbit inference engine (AGENTS.md)
**/*: Work in the order Plan, Check the plan, Execute, Self-check/tests, and keep each loop to one change set.
Push every commit before it can be lost; do not leave commits unpushed at the end of a turn.
Files:
AGENTS.mddocs/agent-cold-storage.mddocs/agent-scratchlog.mdtools/run_sanity_sweep.shdocs/agent-ndjson.mddocs/agent-closed-backlog.mdCLAUDE.md
**/{CLAUDE.md,README.md,docs/**/*.md}
📄 CodeRabbit inference engine (AGENTS.md)
Cite run_setup.bat locations by stable label or subroutine name rather than line number, unless a line number provides immediate value for the introducing change.
Files:
docs/agent-cold-storage.mddocs/agent-scratchlog.mddocs/agent-ndjson.mddocs/agent-closed-backlog.mdCLAUDE.md
docs/*.md
📄 CodeRabbit inference engine (CLAUDE.md)
When a change teaches or invalidates a lesson, update the relevant knowledge document in the same commit; edit existing entries rather than only appending.
Files:
docs/agent-cold-storage.mddocs/agent-scratchlog.mddocs/agent-ndjson.mddocs/agent-closed-backlog.md
**/*.{py,bat,cmd,ps1,yml,yaml,json,sh}
📄 CodeRabbit inference engine (AGENTS.md)
Keep source text ASCII plain text and do not manually change line endings; follow .gitattributes.
Files:
tools/run_sanity_sweep.sh
CLAUDE.md
📄 CodeRabbit inference engine (AGENTS.md)
Run markdownlint-cli2 only on CLAUDE.md with the repository's MD029-only configuration; new Active Backlog entries must be bullets with the number in prose, not literal ordered-list markers.
Files:
CLAUDE.md
🧠 Learnings (1)
📓 Common learnings
Learnt from: CR
Repo: mixmansoundude/Python_vs_Windows
Timestamp: 2026-07-31T09:26:58.985Z
Learning: Freeze scope during an iteration loop; fix CI issues before test failures, assertions, and product code; implement exactly one feature slice and add exactly one test per loop.
🪛 LanguageTool
docs/agent-cold-storage.md
[style] ~50-~50: Consider an alternative for the overused word “exactly”.
Context: ...t manufacture that evidence, which is exactly why this belongs in Cold Storage rather...
(EXACTLY_PRECISELY)
[style] ~70-~70: Consider an alternative for the overused word “exactly”.
Context: ...just a version bump. Condition (b) is exactly what the "Next-pin probe concept" secti...
(EXACTLY_PRECISELY)
[grammar] ~74-~74: Ensure spelling is correct
Context: ...nal reasoning plus a deliberately blunt writeup of why Tier B's narrow, well-justifie...
(QB_NEW_EN_ORTHOGRAPHY_ERROR_IDS_1)
🔇 Additional comments (7)
.markdownlint-cli2.jsonc (1)
13-25: LGTM!AGENTS.md (1)
243-243: LGTM!tools/run_sanity_sweep.sh (1)
97-101: LGTM!CLAUDE.md (1)
16-30: LGTM!Also applies to: 85-88, 484-486, 737-742, 1059-1065
docs/agent-cold-storage.md (1)
1-2: LGTM!Also applies to: 5-116
docs/agent-ndjson.md (1)
533-534: LGTM!Also applies to: 550-552
docs/agent-scratchlog.md (1)
53-55: LGTM!
Two paths in docs/agent-cold-storage.md and docs/agent-closed-backlog.md were line-wrapped inside a backtick-delimited code span. Markdown renders that newline as a literal space, so docs/agent-lessons-learned.md rendered as "docs/agent-lessons- learned.md" and the condabin\conda.bat path similarly picked up a stray space. Reflowed both paragraphs so each path stays on one source line within its own code span. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A4iE1BRSkUETwz237XeuTW
batch-check.yml's Miniconda-availability step comments still pointed at CLAUDE.md for Active Backlog item 7, which PR #403 moved out to docs/agent-closed-backlog.md -- update both references to match the pattern already used in docs/agent-ndjson.md. Also replace the two non-ASCII cursor-block characters in demo-bootstrapper-output.md's postexec-checkpoint quotes with a plain ASCII underscore.
batch-check.yml's Miniconda-availability step comments still pointed at CLAUDE.md for Active Backlog item 7, which PR #403 moved out to docs/agent-closed-backlog.md -- update both references to match the pattern already used in docs/agent-ndjson.md. Also replace the two non-ASCII cursor-block characters in demo-bootstrapper-output.md's postexec-checkpoint quotes with a plain ASCII underscore.
… cleanup (#404) * docs: show the full disk-space warning block in Scenario 24 The scan previously quoted only the trailing [WARN] REQ-025 log line; the same code path (run_setup.bat's disk-space guard) also echoes three raw *** lines directly to the console before it, which a real user sees in the same run. Add them so the scenario matches what actually prints. * tools: add audit_console_messages.py for demo-doc coverage checks Productionized the ad hoc script used for the bottom-up console-message audit (echo/call:log lines in run_setup.bat vs. docs/demo-bootstrapper- output.md coverage), following the same hand-run, not-wired-into-CI pattern as tools/audit_batch_exit_paths.py, so a future re-review of the demo doc doesn't have to re-derive the extraction/matching logic. * fix: stale item-7 cross-refs and non-ASCII cursor glyphs (CodeRabbit) batch-check.yml's Miniconda-availability step comments still pointed at CLAUDE.md for Active Backlog item 7, which PR #403 moved out to docs/agent-closed-backlog.md -- update both references to match the pattern already used in docs/agent-ndjson.md. Also replace the two non-ASCII cursor-block characters in demo-bootstrapper-output.md's postexec-checkpoint quotes with a plain ASCII underscore. * docs: fix stale REQ-018 bullet re: activity-aware EXE verification CLAUDE.md Active Backlog item 9: README's REQ-018 bullet said the verification run is "force-stopped after a short interval even if running fine," which is the opposite of the activity-aware-kill behavior that actually shipped (only a completely silent process is force-stopped; any output at all keeps it running as long as needed). Update the bullet to describe the real condition, note the narrower --hidden-import repair-check re-run stays unconditionally time-boxed by design, and move the now-closed item 9 into docs/agent-closed-backlog.md per CLAUDE.md's own directive. No behavior changed -- documentation only. * fix: tci_justme misleading AllUsers WARN, plus a backlog-number collision Active Backlog item 11 (:tci_justme's WARN unconditionally claims "AllUsers install failed" even when AllUsers was only ever skipped, never attempted): :try_conda_install now sets HP_CONDA_ALLUSERS_ATTEMPTED right before the real install attempt, and :tci_justme branches its log line on that flag -- a genuine failure keeps the original WARN wording, a skip gets a new, honest INFO line. Added a regression assertion to tests/selfapps_justme.ps1 confirming the new wording fires and the old one doesn't, in the non-elevated scenario that test already exercises. Updated demo-doc Scenario 19 to match (marked [Extrapolated Branch] pending a fresh CI capture). While closing this out, found item 11 collided with an already-closed, differently-numbered item from 2026-07-25 (still correctly cited by docs/agent-ndjson.md) -- and the same check on item 9 (closed earlier this session) turned up an identical collision, cited by both docs/agent-ndjson.md and docs/agent-interconnect.md. Renumbered both to 16/17 (the next genuinely unused numbers) when archiving them into docs/agent-closed-backlog.md, and filed a new item 18 documenting that the remaining current items (8, 10, 12, 13, 14, 15) show the same number-reuse signature via a grep-based check, without yet doing the individual verification each one needs before it can be safely renumbered too -- left for a dedicated future pass. --------- Co-authored-by: Claude <noreply@anthropic.com>
Summary
CLAUDE.mdis loaded in full every session via Claude Code's context, and its "ClosedBacklog" section alone was over 60% of the file's ~4200 lines despite being a pure
historical record with no forward-looking action attached to any entry.
out into two new files:
docs/agent-closed-backlog.mdanddocs/agent-cold-storage.md,leaving short pointer stubs in
CLAUDE.md.docs/agent-lessons-learned.md/agent-interconnect.md/agent-ndjson.md, thesetwo new files are deliberately not referenced via CLAUDE.md's
@importmechanism, sothey're read on demand rather than loaded into every session's context — that's the whole
point of the move.
CLAUDE.mdshrinks from ~4200 to ~1065 lines with zero content loss (verified viasubstring-containment checks of every moved section against the pre-move file).
lines before moving it, not assumed from the task description alone.
7.marker as apermanent, expected MD029 lint exception (
.markdownlint-cli2.jsonc,AGENTS.md,tools/run_sanity_sweep.sh) to reflect that the exception no longer applies — confirmedlocally that
markdownlint-cli2 CLAUDE.mdnow reports 0 issues.docs/agent-ndjson.md,docs/agent-scratchlog.md) thatpointed at item 7's old location inside
CLAUDE.md.Test plan
python -m compileall -q .— cleanpython tools/check_delimiters.py run_setup.bat— clean (unchanged file)markdownlint-cli2 CLAUDE.md— 0 issues (previously 1 permanent/expected finding foritem 7's marker; confirms the exception no longer applies)
python -m yamllint .github/workflows/— clean (no workflow changes in this PR)python -m pytest tests/test_*.py -q— 398 passed, 41 skipped (no code changes)to its pre-move content via Python substring-containment checks
Generated by Claude Code