RISK-2 drift fix: 'Не построен — follow-up' → points at the deployed extension
(.common/lib/pi-extensions/session-close-ritual.ts), records why agent_end not
agent_settled (live-verified), bump 0.4.0→0.5.0 (MINOR), dist rebuilt.
Reviewer findings (sovereign-catalog-wave2-review, inbox 2026-08-12T13:30Z):
F1 dist/project-discipline.skill was stale (0.1.2 vs 0.2.0) — rebuilt, now 0.2.1.
F2 description contradiction — leftover old Rule-4 text («grant via разреши
автопуш…» + duplicate (5)) survived my v0.2.0 edit; removed, kept the
'Activated by' activation line. Body Rule 4: dropped stale 'any grant' /
'the grant does not apply' references (no grant concept in free-push rule).
lint 0/0.
RED (2026-08-12, clean pi -p subprocesses): user phrases «прогриль мой план»,
«grill my plan», «хватит, точи план», «прогриль с доками» all resolved to
grilling, NOT grill-me/grill-with-docs. Root cause (verified in pi docs
skills.md): disable-model-invocation: true hides a skill from the system
prompt — the thin user-wrapper split (mattpocock's slash-command model) is
structurally unreachable in pi's natural-language matching.
GREEN: single grilling v0.2.0 with integrated Docs mode (trigger «с доками» /
«зафиксируй» → decision record to .wiki/concepts/<topic>-decision.md or
.brainstorm/ append). Re-run: 5/5 positive → grilling, 4/4 negative clean.
Removed grill-me + grill-with-docs (invisible in pi, dead weight); README
table updated; lint 0/0; ~/.claude/skills = 39; dist pruned+rebuilt.
poller-service-deploy-via-factory (fork 1). New skill v0.1.0 — installs the
standing-duty stack as platform-native services (systemd/launchd/winsw):
no node window, OS-supervised autostart+crash-restart, run-as-user, deploy-boundary.
- fetches winsw (pinned v2.12.0 + SHA256-verify, not vendored; STOP on placeholder)
- installs DISARMED: scope is runtime config (poller-scope.json), arming is a
separate operator step via the appeals-inbox pult; never carries POLLER_PROJECTS/DRY_RUN
- confirmation gates: discovery (read-only) -> plan -> writes; rollback section
- tears down the legacy start-worker.ps1 Scheduled Task (no double-claim)
- dist/setup-agents-task-runner.skill rebuilt
[skip-tdd: visual] — installer docs + OS service config, no testable pure logic.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace all mcp__markitdown__convert_to_markdown invocations with the
native `markitdown <path|url>` CLI (v0.1.6, on PATH). Outputs to stdout
or `-o <file>`; sees the full host filesystem, so the Docker bind-mount
caveat (host→container file:// translation, [Errno 2] /c:/Users/...) is
gone and that whole section is removed. Updated the ingest pattern (-o
straight into .wiki/raw/), gotchas table (command-not-found → check
`markitdown --version`), and contrast table header (CLI, not MCP).
Description triggers unchanged. PATCH bump 1.0.0→1.0.1; dist artifact
rebuilt.
Decommissioned the Docker MCP containers: no container is named
`markitdown-mcp` (the server spawns anonymous ones from
markitdown-mcp:latest, 3 had piled up); removed all by image ancestor.
Left mcpServers.markitdown in ~/.claude.json untouched (out of scope) —
flagged as a follow-up in the concept page.
Wiki: concepts/using-markitdown-cli-migration.md + index + log.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two findings from live-hook e2e smoke 2026-05-25 on this Windows machine:
(1) README snippet was pwsh-only — PS 7 Core isn't on stock Windows. PS 5.1
(`powershell`) is always present and the hook script runs cleanly under both.
README now leads with `powershell` and notes the `pwsh` swap for PS 7+ users.
(2) Missing caveat that hooks load at Claude Code session start — mid-session
edits to ~/.claude/settings.json don't activate the hook until CC restart.
Without this note user would think the hook is broken after applying the
snippet (standalone smoke would pass but live in-session wouldn't fire).
Added explicit restart instruction + verification recipe.
Also: dist/session-handoff.skill now tracked (was missing since promotion —
inconsistent with other dist/*.skill artifacts that ship in repo).
PATCH bump 0.3.1 → 0.3.2 (docs-only, no behavioral change in hook or skill).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CLAUDE.md.template gains `session handoff: read on start, write on end`
between `pull remote before work` and `follow project discipline`
(session-lifecycle clustering). Step 5.6 trigger->fulfiller table gains
the corresponding row (template ↔ table source-of-truth invariant).
Greenfield-bootstrap'ed projects now ship handoff trigger из коробки.
Existing projects unaffected — CLAUDE.md merge in Step 5 is idempotent
and respects user removals.
Bump 1.11.0 -> 1.12.0 MINOR (new canonical trigger = new capability,
backward-compatible). dist/project-bootstrap.skill rebuilt.
Closes [session-handoff-bootstrap-template-extend].
Unblocks [session-handoff-existing-projects-upgrade] Path B.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends Karpathy LLM Wiki canon with two new artifact types alongside
existing entities/concepts/packages/sources. Inspired by community
discussion that highlighted explicit tracking of surfaced tensions and
unanswered questions as missing aggregation points in the canonical
layout — they currently get scattered into concepts/ or lost in log.md.
Scope is schema-level only — no new ingest behaviour prescribed. Policy
on when to escalate an inline `> **Противоречие:**` flag into a
contradictions/<slug>.md page (and the analogous flow for
open-questions) stays the user's call.
setup-wiki [v1.0.0 → v1.1.0, MINOR — additive page types]:
- Discovery (Phase 1) now requires 6 content dirs for `noop` mode
- Phase 2 plan blocks list new dirs in greenfield + migrate
- CLAUDE.md schema template gains two page-type entries with status
enums (contradictions: open|resolved|accepted-divergence;
open-questions: open|answered|obsolete)
- index.md template gains two empty sections
- Phase 4a .gitkeep list, Phase 4b mkdir + touch, Phase 5 verify count
(four → six dirs), Phase 6 report count all updated
- README.md layout tree + content-dirs sentence
using-wiki [v1.0.0 → v1.1.0, MINOR — additive type values]:
- Prerequisites: four → six content directories
- Page frontmatter type enum: + contradiction | open-question
- Per-type frontmatter extensions documented (status + affects/touches)
- File naming patterns: + contradictions/<slug>.md, open-questions/<slug>.md
- index.md sections-by-type list updated
- README.md mirrors SKILL.md changes
dist/: setup-wiki.skill + using-wiki.skill rebuilt.
project-bootstrap inline reference block intentionally untouched — it's
labelled "Reference (for context only — setup-wiki is the source of
truth)" and drift-tolerant by design.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The interns-mcp source lives inside the `OpeItcLoc03/common` monorepo
(at ~/projects/.common/), not as a separate `interns-mcp` repo on Gitea.
The previous commit incorrectly wrote `git clone .../interns-mcp`.
Now the clone-fallback correctly:
- If ~/projects/.common/.git exists but lib/interns-mcp/ is missing:
git -C ~/projects/.common pull --ff-only (stale clone, needs fresh subdirs)
- If ~/projects/.common/ is absent:
git clone .../common.git ~/projects/.common (fresh clone of monorepo)
- Added "wrong repo" to Common mistakes section
- Added monorepo scope note to Out of scope section
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Replace all `<project-root>/.common/...` with `~/projects/.common/...`
(POSIX-absolute paths, mirroring setup-projects-meta after the
using-projects-meta-fix-paths fix). The cwd-relative form broke
when Claude was launched from System32 or another non-project dir.
- Add git clone fallback: when `interns-mcp` source is absent, clone
from `https://git.kzntsv.site/OpeItcLoc03/interns-mcp` instead of
stopping with "initialize first". Mirrors setup-projects-meta Phase 4.
- Remove "source must be in place" from Out of scope (now handled).
- Fix MCP registration cwd to `~/projects` (base dir containing .common/).
- Bump version 0.2.0 → 0.3.0 (MINOR — new capability: clone fallback).
Closes [setup-interns-fix-paths] + [setup-interns-clone-fallback].
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Bug: v1.10.0 bump + Step 5.6 map added these 2 canonical triggers, but the
CLAUDE.md template literal (lines 339-342) and corresponding prose paragraphs
were not updated. Symptoms: upgrade-mode on repos missing these 2 triggers
detected "already canon" and skipped insertion, leaving gaps.
Fix:
- Template: added `follow tdd-criteria` after `follow project discipline`,
added `recommend, don't menu` after `delegate to interns when allowed`
- Prose: added 2 new paragraphs describing each trigger with install hints
- Version: 1.10.0 → 1.10.1 (PATCH — completing what 1.10.0 claimed)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Part A — Pre-close coverage gate. ### Task completion now lists acceptance
criteria from the per-task file before setting 🟢; missing evidence → ask
the user. New section ### Post-commit task closure prompt: after a feat:/fix:
commit the agent asks "эта работа закрывает таску <slug>?" so shipped code
doesn't sit under stale ⚪ blocks (cf. extend-project-discipline-brainstorm-
workspaces / project-creation-lifecycle-skill, both shipped before close).
Part B — Local-first recommendations at session-start / "what next" triggers.
cwd-project board (🔴 → 🟡 → ⚪) leads; cross-project urgents are at most one
footnote line. Explicit "по всем проектам" flips the order. Pairs with
using-projects-meta's local-first read rule.
Bump 1.0.0 → 1.1.0 (MINOR — adds two new operation types; task block
suggested PATCH but Rule 3 grades these as capability additions). Closes
[using-tasks-close-coverage-gate] (next commit).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three modes now: greenfield-full (new + remote), add-remote, upgrade.
Step 0: detect git/remote/empty state for mode selection.
Step 1.5: Gitea API repo create + push.
Step 8: projects-meta sync for visibility.
Version 1.7.0 → 1.8.0 (MINOR).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
skills/compress/ was a byte-identical dupe of skills/caveman-compress/
(SHA256 match across all 7 scripts/ files; SKILL.md diff was name: +
Process step 2 only; descriptions textually identical = arbitrary
harness tie-break + double-counted listing budget).
Kept caveman-compress canonical: it carries README.md (benchmarks +
caveman-toolkit branding) and SECURITY.md (Snyk false-positive writeup),
and matches the caveman-* prefix invariant.
Ported the better Process-step wording from compress
(`cd <directory_containing_this_SKILL.md>`) into caveman-compress;
the prior `cd caveman-compress` form assumed cwd was the parent dir
and broke when invoked from ~/.claude/skills/caveman-compress/.
Added version: 1.0.0 to caveman-compress frontmatter
(first versioned release; aligns with skill-versioning concept).
Removed: skills/compress/, dist/compress.skill,
~/.claude/skills/compress/ (manual prune; install.sh has no prune step
yet -- future [install-ps1] task should add --prune).
Slash-command impact: /compress gone; /caveman-compress and
/caveman:compress (toolkit-canonical) remain.
Wiki: .wiki/concepts/compress-dedup.md (rationale + rejected
alternatives), index.md + log.md updated.
Tasks: [compress-dedup] flipped active -> done in same commit
(per-task file .tasks/compress-dedup.md created with full decisions
log).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Collapse Step 5.6 from a single-skill detector (only the
superpowers@claude-plugins-official plugin) into a generic
`trigger -> fulfiller` table walker that scales to every canonical
CLAUDE.md trigger.
Inline 9-row map in SKILL.md covers: caveman, superpowers (kind: plugin),
using-wiki, using-tasks, using-projects-meta, pulling-before-work,
project-discipline, using-interns, active-platform. The `kind` flag
(skill | plugin) drives which install command is emitted in the chat-only
recommendation block. Detection paths: ~/.claude/skills/<name>/SKILL.md
for skills, plugins.<id> in ~/.claude/plugins/installed_plugins.json
for plugins.
Algorithm: read project's CLAUDE.md -> match each non-comment line vs
map (substring + tolower, mirrors Step 5 idempotent merge) -> for each
canonical match check disk -> print one chat block listing every missing
fulfiller with copy-pasteable install commands, OR a single
"all dependencies satisfied" line. User-custom lines and removed
canonical lines are silently skipped. Hard rule "never auto-install"
preserved verbatim.
version: 1.6.0 -> 1.7.0 (MINOR per project-discipline Rule 3 -- adds
capability, absorbs prior superpowers-only detector cleanly).
Closes [bootstrap-skill-deps-check] (.tasks/STATUS.md done).
Closes [bootstrap-recommend-projects-meta] by absorption -- the deferred
mirror task was the seed of this generalization; the generic walker now
handles using-projects-meta along with everything else, no per-skill
mirror needed.
Design rationale at .wiki/concepts/bootstrap-skill-deps-check.md:
- why generic over per-skill mirrors (5x mirror explosion)
- skill vs plugin kind distinction
- MCP-server-backed skills (only check using-X policy skill;
setup-X self-fires via Prerequisites pointer)
- source-of-truth invariant: SKILL map + assets/CLAUDE.md.template
must stay in sync (a future CI lint could enforce)
SKILL.md Phase 5 platform-path table (lines 152-154) had stale
~/.local/projects-meta-mcp; replace with ~/projects/.common/lib/
projects-meta-mcp to match real install canon (body Phase 4 clone
target + ~/.claude.json mcpServers.projects-meta.args[0]).
version: 1.0.0 -> 1.0.1 (PATCH, doc consistency).
Closes [using-projects-meta-fix-paths] (.tasks/STATUS.md done).
Sibling task [migrate-to-common-lib] in projects-meta-mcp closed
the same day with matching path migration.
- Replace ~/.local/projects-meta-mcp → ~/projects/.common/lib/projects-meta-mcp
- Update setup-projects-meta and using-projects-meta skills
- Rebuild dist/ with new paths
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
setup-interns (v0.1.0) — 8-phase install of the local `interns` MCP server:
detect `.common/lib/interns-mcp/` source, `pip install -e`, write
`.common/secrets/interns.env` with endpoint API keys (gitignored), register
`mcpServers.interns` in `~/.claude.json` with absolute Python interpreter
path + `cwd` so the runtime resolves config relative to project root.
Mirrors setup-projects-meta / setup-context7 confirmation-gate shape.
Description 899 chars (under the 900 budget per
feedback_skill_description_length_limit).
using-interns (v0.1.0) — runtime policy for `mcp__interns__*`. Per-session
permission grant mirrors project-discipline Rule 4: ask-mode default,
conversational grant ("разреши интернов" / "allow interns"), conversational
revoke, always-ask paths for `**/.env`, `**/secrets/**`, `**/*.key`,
`**/.ssh/**`, `**/.aws/credentials` etc with transitive rule (Claude can't
bypass by reading the file with the local Read tool and forwarding content),
cost-cap >$0.10 always asks, session-end automatic reset. Routing hints for
`bulk_text_read` (3+ files or one file >400 lines) and `transcript_distill`
(before .wiki/log.md updates / session summaries). Description 814 chars.
project-bootstrap (1.5.0 -> 1.6.0, MINOR — capability added):
* assets/CLAUDE.md.template: insert `delegate to interns when allowed`
between `follow project discipline` and `we're on Windows`.
* SKILL.md Step 5: same insertion in inline template + new commentary
paragraph explaining the trigger, no-op semantics, install pointer.
* SKILL.md Step 5.5: bootstrap-manifest table extended with
`setup-interns`, `using-interns`, `project-discipline` rows.
* README.md: Workflow Step 5.5 description + See also section both pick up
the new skills.
Root CLAUDE.md dogfood: `delegate to interns when allowed` line added.
.wiki/log.md decision entry. .tasks/STATUS.md task moved to 🟢 done.
.tasks/interns-skills-mvp.md per-task file with goal / key-files /
decisions log / open questions.
Build + install verified: `dist/{setup-interns,using-interns,
project-bootstrap}.skill` rebuilt; `bash scripts/install.sh ...` succeeded;
harness skill listing shows full descriptions for all three (no H1
fallback) — confirms the description budget held.
Versioning per project-discipline Rule 3:
* setup-interns: 0.1.0 (first edit of unversioned artifact, MAJOR=0).
* using-interns: 0.1.0 (first edit of unversioned artifact, MAJOR=0).
* project-bootstrap: 1.5.0 -> 1.6.0 (MINOR — adds capability without
breaking existing CLAUDE.md merge or manifest consumers).
Server runtime itself (`.common/lib/interns-mcp/`) is out of scope for this
task — the skills land lifecycle, policy, and bootstrap integration so any
machine where the MCP server is later installed already has Claude's
policy in place.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Step 5 was binary on upgrade (append whole template / leave alone), so
projects bootstrapped before v1.2.0 silently missed new canonical triggers
(`check across all projects`, `we're on Windows`) on re-run. Now upgrade
reads existing CLAUDE.md, substring-diffs vs template, preserves a
deliberately-pinned platform line, and appends only missing lines after
explicit confirm. Re-runs are no-ops.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CLAUDE.md template gains a fifth trigger line, `check across all projects`
(verbatim phrase from `using-projects-meta` description). Every new or
upgraded project now auto-loads cross-project task aggregation and shared
projects-wiki access without an explicit verbal trigger.
No Step 5.7 mirror of Step 5.6 (superpowers plugin recommendation):
`using-projects-meta` Prerequisites already self-corrects to
`setup-projects-meta` when MCP tools are missing.
Synced: assets/CLAUDE.md.template, inline copy in SKILL.md (Step 5),
README.md + README.ru.md trigger lists, this repo's local CLAUDE.md,
dist/project-bootstrap.skill, .wiki/concepts/projects-meta-skills.md
(Bootstrap trigger section), .wiki/log.md.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
