testsAndMisc/docs/superpowers/evidence/claude-md-terse-rewrite-2026-08-16.json
Krzysztof kuhy Rudnicki 004b4b0b29
Some checks are pending
Pre-commit checks / pre-commit (push) Waiting to run
Shell tests / shell-tests (push) Waiting to run
Shell tests / arch-tests (push) Waiting to run
Cut CLAUDE.md 259 -> 141 lines: delete what hooks already enforce, fix what was wrong
This file loads into context on every turn of every session, so every line is
paid for repeatedly. Two thirds of it either duplicated a gate that already
fails the commit, or stated something untrue.

Deleted because a hook enforces it: the ruff/mypy/pylint/coverage settings,
"double-quote expansions" (shellcheck runs with no severity filter), absolute
imports, Google docstrings, type hints, and the inline-Python rule -- which
got a real gate in 17b20e5 rather than being dropped.

Corrected four claims that were simply false, all verified against the config:

- pytest-coverage was listed as a pre-push hook; it is stages:[pre-commit].
  Pre-push is prettier + ci-mirror.
- the root-symlink list named requirements.txt twice and omitted lint_python.sh
- "scripts/" described a directory that does not exist at repo root
- per-file-ignores were listed as S101/PLR2004/S310/S607/PLC0415; the actual
  set is ARG/D/PLC0415/PLR2004/S101/SLF001

Added the five top-level dirs the layout table omitted (poker-stakes, bottles,
bucket_catch, focus_owner, kcd2_dice_solver) and documented three gates that
were enforced but undocumented: the 4-file contract threshold, append-only
session logs, and the 2% jscpd limit.

Also recorded two places the gate is weaker than it looks: the mypy hook
disables ~20 error codes despite strict=true, and no-polling-antipatterns only
inspects *_loop/*_daemon/poll* functions.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011w2Pj4Ek3ZNMQVYcFHgXDY
2026-08-16 15:17:02 +02:00

40 lines
3.3 KiB
JSON

{
"intent": "Shrink CLAUDE.md from 259 to 141 lines by deleting rules that hooks already enforce, correcting claims that were factually wrong, and rewriting the remainder as HOW rather than WHY.",
"scope": [
"CLAUDE.md (rewritten)",
"Non-goal: no hook behaviour changes in this commit; the one new gate (no-inline-python) landed separately in 17b20e5"
],
"changes": [
"Deleted prose duplicating existing gates: ruff/mypy/pylint/coverage settings, 'double-quote expansions' (shellcheck SC2086, no severity filter), absolute imports (TID252), Google docstrings, type hints (ANN), the inline-Python rule (now the no-inline-python hook), and the evidence-file procedure detail.",
"Corrected four factual errors found by auditing the config: (1) pytest-coverage was listed as a pre-push hook but is stages:[pre-commit] -- pre-push is prettier + ci-mirror only; (2) the root-symlink list named requirements.txt twice and omitted lint_python.sh; (3) 'scripts/ -- workspace-level helpers' described a directory that does not exist at repo root; (4) per-file-ignores were listed as S101/PLR2004/S310/S607/PLC0415 when the actual set is ARG/D/PLC0415/PLR2004/S101/SLF001.",
"Added five top-level directories the layout table omitted: poker-stakes, bottles, bucket_catch, focus_owner, kcd2_dice_solver.",
"Documented three enforced-but-undocumented gates an agent would otherwise hit blind: the 4-file contract threshold, append-only session logs, and the 2% jscpd duplication limit.",
"Recorded two places where the gate is weaker than it looks: the mypy hook disables ~20 error codes despite strict=true in config, and no-polling-antipatterns only inspects *_loop/*_daemon/poll* functions."
],
"verification": [
{
"command": "bash ~/utils/scripts/check_file_length.sh CLAUDE.md",
"result": "pass",
"evidence": "259 -> 141 lines; exits 0, comfortably under the 250-line cap with headroom rather than trimming to exactly 250."
},
{
"command": "grep checks against check_python_location.sh, validate_evidence.py, check_append_only_sessions.sh, check_agent_contract.sh, check_polling_antipatterns.sh, .githooks/pre-commit",
"result": "pass",
"evidence": "Every retained claim verified against its source: allowlist dirs match; banned phrases are exactly 'should work'/'probably fine'/'seems right'; sessions regex is ^docs/superpowers/sessions/.*\\.(jsonl|log|txt)$; MULTI_FILE_THRESHOLD=4; in_poll awk guard matches *_loop|*_daemon|poll*; jscpd --threshold 2."
},
{
"command": "test -d for each of the 13 directories named in the layout table",
"result": "pass",
"evidence": "No MISSING lines -- every path in the table exists."
}
],
"risks": [
"Deleting enforced prose relies on the hooks staying wired. That is the intended trade: a gate that fails the commit is stronger than a sentence an agent may not read, and .pre-commit-config.yaml is now named as the source of truth.",
"The audit that classified each rule was produced by a subagent; every claim retained or corrected here was independently re-verified against the config files before editing."
],
"rollback": [
"git revert this commit to restore the 259-line version.",
"Validate with: bash ~/utils/scripts/check_file_length.sh CLAUDE.md (will then fail at 259 lines)."
]
}