Skill Claude CodeCodex
Use when working in a repository that uses bd or Beads for durable project task tracking, issue dependencies, blocker management, multi-session handoff, or shared work memory. Trigger when the user asks to find ready work, claim or close tasks, create follow-up work, inspect blockers, recover project context, or…
Agent Claude Code
Runs the ci-settle loop (formula convergence, step X3) — watches CI checks, triages bot comments, fixes real findings or declines with a stated reason, regenerates generated artifacts. No merge authority.
Agent Claude Code
Weekly standing patrol for docs drift vs code, broken links, index consistency, and i18n lag reporting (which locales are behind English and by how much). Patrols for progressive-disclosure violations doc-writer may have missed.
Agent Claude Code
Weekly standing patrol that mines telemetry — prunes never-consulted assets, recalibrates time-boxes, reclassifies "other" failures, and mines primitive candidates (duplicated code shapes, repeated raw hackage imports) into triager-fed issues.
Agent Claude Code
The project's strategic pairing agent (Yegge's Seneschal). Nick invokes it inside this repo, one task at a time, for design/strategy work — docs IA, website design system, architecture decisions, roadmap shaping. It does not implement; its deliverables are specs/ADRs/issues fed into the queue.
Agent Claude Code
Per-release patrol that runs every user-facing skill against the new release; API drift becomes issues fed to the triager. Empirical verifier for the skill-author persona's work.
Agent Claude Code
Designs and maintains the user-facing skills NeoHaskell ships. Every skill change is verified by actually running the skill against the current release (its area verifier). Persona skill-author.
Agent Claude Code
Visual QA against the design spec for website changes (verify phase) — cross-locale layout sanity, accessibility basics. Every visual change attaches before/after Playwright screenshots to the PR; the pr-ready checker blocks without them.
Agent Claude Code
For website changes, owns the "spec" phase as a design spec — information architecture, wireframe/description of the change, interaction notes. Gate 1 approves design, not prose. Persona frontend-ux.
Hook Claude Code
Runs after a tool call finishes for Edit, Write, MultiEdit, Edit, Write, MultiEdit and Bash tool calls, executing bd-token-tracking.py via python3 with --mode (3 commands). From neohaskell/NeoHaskell.
Hook Claude Code
Runs before the agent uses a tool for Edit, Write and MultiEdit tool calls, executing dialect-guard.py and expectation-guard.py via python3 with --show-current (3 commands). From neohaskell/NeoHaskell.
Hook Claude Code
Runs when a session starts, executing session-start.sh with --hook-json (2 commands). From neohaskell/NeoHaskell.
Hook Claude Code
Runs when a session ends, executing bd-dolt-sync.sh via bash. From neohaskell/NeoHaskell.
Settings file Claude Code
Agent settings declaring 4 hook events (PostToolUse, PreToolUse, SessionStart, SessionEnd) and 15 allowed tools.
Skill Claude CodeCodex
Work on the Neo CLI's bundled browser IDE under neo/ - the Vite/React frontend at neo/assets/ide/, its embedded dist/ synchronization into the Rust binary, and visual critique of the rendered event-modeling canvas. Use when touching assets/ide/, the neo ide Rust backend (src/ide/), or when asked to screenshot/critique…
Skill Claude CodeCodex
Write or modify Rust code in the Neo CLI under neo/ - crate conventions, the error-as-LLM-repair contract, output prefixes, and interactive-vs-CI (--ci) behavior. Use when editing any neo/ .rs file, adding a subcommand or JSON field, wrapping a subprocess, or writing any user-facing error. This is Rust, not NeoHaskell…
Skill Claude CodeCodex
Route a request about the Rust Neo CLI under neo/ to the exact command, subsystem and module before editing. Use at PLAN time for any neo/ change: locate by command surface (new/build/run/test/lock/ide/inspect/validate/skills), by subsystem (reconcile, subprocess, interpret, ide, tui), or by error/output contract.…
Skill Claude CodeCodex
Run and reason about the Neo CLI test layers under neo/ - unit, integration, e2e, and the generated-project / Neo-on-Neo smoke verification. Use before declaring any neo/ change done, or on "run the neo tests / cargo test / integration / e2e". Wraps the right command per layer inside nix develop, states network/Nix…
Skill Claude CodeCodex
Stage playbook for executing steps of the change formula v2 (ADR-0075) — the 5-step molecule that takes a request from spec to merged PR. Use when claiming or executing any bead poured from the change formula (spec, spec-approval, build, verify, pr), when parking a stuck step, or when recording a spec approval. Covers…
Skill Claude CodeCodex
Derive NeoHaskell concepts (Query, Command, Event, Outbound integration) through their canonical Template Haskell marker instead of hand-writing the instances the marker generates. Use whenever creating or editing a query, command, event, entity, or any concept that has a TH marker — reach for deriveQuery / command /…
Skill Claude CodeCodex
Add, modify, or debug NeoHaskell dialect enforcement rules (hlint law + edit-hook teacher). Use when a new dialect rule is needed, a rule misfires, or an exception should be registered or promoted.
Skill Claude CodeCodex
Use when Nick asks to enqueue work into the change process — "encola GH issue 123", "enqueue issue.
Skill Claude CodeCodex
Write NeoHaskell (not vanilla Haskell) code in this repo — copy-adapt discipline, the fast repair loop, and the dialect escape hatch. Use when implementing or modifying any .hs file.
Skill Claude CodeCodex
Route a request to exact code locations using the codemap — capability lookup, extension points for new code, signature grep, edit-set expansion. Use at PLAN time, before any implementation; execution must never search.