Agent Claude Code
Use this agent to audit documentation accuracy and git hygiene on any project. Cleans up plan files committed by mistake, verifies .gitignore coverage, checks README and CLAUDE.md accuracy against actual project state, and optionally verifies URLs match canonical domains. Examples: "Run docs-checker" — full audit "Run…
Agent Claude Code
A naive bilingual FR/EN visitor who discovers a website with zero context, writes a first-impression report, then reads the project's README to react to the gap between perception and intent. Examples: "Run dummy-visitor https://budget.example.com /Dev/my-budget-app" — full visit + informed reaction.
Agent Claude Code
Executes well-scoped implementation tasks with clear specifications. Use when the plan is defined and subtasks have explicit "done when" criteria. Not for architecture decisions or ambiguous tasks.
Agent Claude Code
Audits a project against the global portfolio standards (author signature, gitignore, no secrets, no dead code, test file placement). Use before releases or during compliance sweeps.
Agent Claude Code
L3 of the one-way cascade: diagnoses structural problems after a direct fix (L1) and systematic-debugging (L2) have both failed. Evaluates root cause, recommends solutions including rearchitecture, stack changes, or rewrites. Does NOT write production code — produces a plan for the implementer.
Command Claude Code
Run docs-checker and portfolio-audit in parallel on the current project. Optionally specify a canonical URL.
Command Claude Code needs its repo
Sweep /.claude disk hygiene to stated retentions, clean up stale plans from repos, audit plugin health, and compact agent memory files.
Command Claude Code
Scaffold a new app in /Dev with full portfolio compliance from day one.
Command Claude Code needs its repo
Sync the claude-code-setup repo from live /.claude/ config. Copies, anonymizes and redacts private regions, cleans stale files, updates README counts, updates workflow guide, audits for leaks. Refuses to overwrite hand-edited destinations.
Command Claude Code
Monthly technical health review in two phases. Phase 1: fast triage across ALL portfolio apps — scores each app on commit activity, portfolio prominence, time since last scan, and detected issues. Proposes a priority order. Phase 2: deep review on the apps the owner selects. Run without args for the full flow, or skip…
Skill Claude Code
How CI gates and GitHub branch protection actually behave on this solo free-tier account — the zero-coverage hole in aggregate jobs, the 403 on private repos, and the three traps that permanently deadlock a solo merge. Load before writing or changing a CI gate, adding a pr-gate caller, or setting/checking branch…
Skill Claude Code
How to cut a CLAUDE.md without losing a fact — the four-bucket taxonomy, the trigger-line shape that makes an extracted skill actually load, the gitignore prerequisite, and the line-coverage check that must be made to fail once. Load before editing, restructuring or pruning any CLAUDE.md, or when claude-md-weight.sh…
Skill Claude Code
How the two memory systems on this machine resolve on disk, why per-agent stores are project-scoped and must never be consolidated upward or committed, and why claude-mem was removed. Load before creating, moving, consolidating or gitignoring any memory store, or before proposing a third place to keep session…
Skill Claude Code
Portfolio three-layer data system (inventory, README pitch frontmatter, editorial.ts) and coherence rules. Load before changing how any project is described, listed, ordered or given a hub tile.
Skill Claude Code
Why uv is the only Python manager on this machine, where the enforcement actually lives, and which plausible "fixes" are wrong. Load before writing any Python here, before touching a hook or scheduled job that runs Python, and before diagnosing a wrong-interpreter symptom.
Skill Claude Code
Why each launchd LaunchAgent on this machine exists, scheduled at the hour it is, and which plausible "fixes" are wrong. Load before creating, editing, moving or diagnosing any scheduled job. Current state comes from jobs-inventory.sh, never from this file.
Skill Claude Code
Test conventions across this portfolio — regression-test comment format, property-test setup for TS and Python, the financial-math arbitrary settings that were tuned by measurement, and why a failing check may be pinning a defect rather than reporting one. Load before writing tests, adding a regression test to a bug…
Command Claude Code
New-model setup review — derive every model pin and headless job from the tree, add the new model's effort entry, re-test prompts and design scaffolding tuned for the previous model. Proposes; commits nothing without the owner.
Command Claude Code
Take a plan from review to reviewed code — review, fold findings, stop for owner approval, execute, code-review, stop before push. Owns the sequencing and the approval stop only.
Skill Claude Code
Load before any frontend or visual design work in a project with no stated direction.
Command Claude Code
Audit the rendered design of live sites. Evidence capture first, then one critique per site. Argument is a URL, a repo name, or all.
Skill Claude Code
Load before reviewing any rendered UI — a live site, a local build, or a page you just changed — and before claiming a visual change is done.
Skill Claude Code
Load before designing a new site or page, or redesigning one, when no DESIGN.md exists yet.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: