6.1 KiB
6.1 KiB
| type | schema | rules | |||||||
|---|---|---|---|---|---|---|---|---|---|
| journal |
|
|
Journal
2026-04-23
- Restructured tree:
tasks/→.claude/tasks/, created.claude/memory/(5 registries) +.claude/audits/. - Adapted CLAUDE.md + skills
onboard,init-project+ agentonboarder+lib/project-archetypes/dotfiles-meta.md. - Added CAPITALIZE step in
ship-feature,bugfix,hotfix,feat,commit-change+ created/closeskill for session-end ritual. - 2nd user verify-gate caught bugs:
.gitignorebroke tracking (fixed BDR-003); harden/validate dispatcher bash broken after audit move (LRN-002). - Audits routed to
.claude/audits/(seo/geo/harden/validate/code-clean) +MIGRATION.mdwritten for existing projects. - 9 atomic commits (
c721a36..a9606aa) via/commit-change— first real exec of Phase 4 CAPITALIZE. - Decisions logged: BDR-002, BDR-003. Learnings: LRN-002. Blockers: BLK-002.
- English-only rule enforced in all CAPITALIZE specs (commit
bfcca72); 9 existing entries retrofitted to English in follow-up commit.
2026-04-27
- Settings: switched
permissions.defaultModefrom"default"to"auto"and droppeddisableAutoMode: "disable"(BDR-004); reorganised top-level keys + addedeffortLevel: "xhigh"; removed stale rootTODO.md(already migrated to.claude/tasks/TODO.md). - Learning: Claude Code
disable*settings use sentinel string"disable", not boolean (LRN-003). - 3 atomic commits (
f7f033f..1421578) via/commit-change. - Animation lib autoflow added: new helper
lib/animation-lib-check.sh+ STEP 5e in/init-project(auto-install) + STEP 2.5 in/onboard(opt-in) + read-only detection inplugin-advisorPHASE 1/2/3 + signal inlib/design-gate.md+ scaffolder note.motionchosen over legacyframer-motion(BDR-005, LRN-004).
2026-05-03
- Added
JuliusBrussee/cavemanas 4th always-on plugin (BDR-006). Full install: plugin + standalone hooks + caveman-shrink MCP scaffold (snippet only, not auto-registered — proxy needs upstream wrapper, LRN-006). - Discovered two co-masking bugs:
claude plugin installdoesn't enable (LRN-005) +session-start.shhardcoded "✅ ON: security-guidance rtk superpowers" regardless of actual state. Addedenable_plugin()helper +plugin_enabled()detector readingenabledPluginsfromsettings.json. Banner now reflects reality. - Side fix: doctor.sh exited under
set -euo pipefailwhen gstack/skills/ missing — wrapped find in brace +|| true. - 3 atomic commits (
0184818..2ec7935).
2026-05-04
- Built skill profile system (BDR-007):
lib/profile.sh+lib/profiles/{design,dev,qa,audit,minimal}.profilepartition gstack + personal skills by purpose. Activation toggles symlinksskills/↔skills-disabled/. - Wired into
agents/plugin-advisor.md(DETECT call toprofile.sh current+ newPROFILEline in OUTPUT + new "Skill profiles" subsection in TOGGLING EXTERNAL TOOLS),lib/toggle-external.sh(header pointer),Makefile(4 targets),skills/profile/SKILL.md(/profileslash command). cmd_currenthonestly reports "full" when nogstack__*entry exists inskills-disabled/— avoids "100% match" trap when full gstack on.- Tested end-to-end: list/show/current/diff/set/reset/apply all green; shellcheck clean; symlink state restored after reset.
- Profile system v2 (BDR-008): extended
profile.shto toggle Claude plugins (claude plugin enable|disable) + MCP servers (magicvialib/toggle-external.sh). Added 4 new profiles:web,seo,web-full,backend. Refined existing profiles to useplugin@<marketplace>syntax +clientries. Always-on plugins protected byMANAGED_PLUGINSallowlist +PROTECTED_PLUGINSdenylist. - Verified:
set webenables ui-ux-pro-max + magic;set seodisables ui-ux-pro-max;set minimaldisables ui-ux-pro-max but spares caveman/security-guidance/superpowers.currentheuristic respects ties (web-full beats web at 100%).
2026-05-05
- Mandated caveman format on all
.claude/memory/*.mdwrites (BDR-009). Rule added to CLAUDE.md "Memory registries" section. Self-applied: CLAUDE.md prose compressed in same pass. - Compressed 5 existing registries via
/caveman:compress(decisions, learnings, blockers, journal, evals) — ~40% input-token reduction per session-start load. - Side chores: disabled
example-skills@anthropic-agent-skillsplugin in settings.json; gitignored*.original.mdcompress backups (recoverable via git history). - 4 atomic commits (
0275eed..639486a) via/commit-change.
2026-05-06
- darwin-skill round 1 across 18 personal skills. Mean 83.4 → 88.7 (+5.3). 16 keeps, 2 reverts (code-clean, doc — D2 dry_run noise). Branch
auto-optimize/skills-20260506-1730. 22 commits, 35 files changed. - Top gains (analyze +18.5, skills-perso +11.9, refactor +11.0, hotfix +9.0) all from same shape: edge-case table in agent file. Captured as LRN-008.
- LRN-009: dry_run ratchet too strict for skills already >91; LRN-010:
~/.claude/skills,agentssymlink to Documents/claude — git operations must run from there. - Audit report
.claude/audits/DARWIN-SKILL-OPTIMIZATION.md. Eval log~/.agents/skills/darwin-skill/results.tsv(38 rows). Branch awaits manual review before merge.
2026-05-07
- /client-handover gates SEO classique + GEO (IA) independently at ≥17/20 (BDR-010). Was: combined display only, gate fired on SEO alone. Now: 4-axis gate (SEO, GEO, HARDEN, VALIDATE), axis-aware fix loop, per-axis override transparency.
- Pattern captured as LRN-011: single subagent emits N gated scores → labeled extraction + axis-aware loop + per-axis escalation. Generalizes to future multi-metric audits (e.g. /harden split TLS/headers/redirects).
- 1 atomic commit
5569a80(feat(client-handover): split SEO + GEO scores, gate GEO at ≥17/20). Bash unit testedextract_score_labeledon 4 cases (new format, /100 normalize, legacy fallback, GEO UNKNOWN strict) — all OK.