monorepo

Prove whether a skill-pack change made agents better at writing code. Use when a single passing eval doesn't prove anything and you need multi-run pass@k evidence, when grading coding tasks with hidden tests the agent cannot see, or when comparing outcomes across skill-set versions.

Create design documents (WHICH/WHERE) for approved specs. A design is a max-200-line architectural sketch — components, interfaces, data flow, and key decisions with trade-offs — that gives reviewers a high-leverage point to redirect architecture before the full plan is written. Design is approved when `wiki/STATUS.md` shows the spec row at `design approved` — written there by a human signal that `kata-dispatch` or the active agent propagates.

Write implementation plans (HOW/WHEN) for approved designs. Translate an approved design into concrete steps, file changes, sequencing, and risks for a trusted agent to execute. Plan is approved when `wiki/STATUS.md` shows the spec row at `plan approved` — `staff-engineer` writes this row after a clean panel review (plans may be approved by agents).

Grade a single artifact (spec, design, plan, or implementation diff) against quality criteria and return findings by severity. Use when another skill spawns a fresh sub-agent for an independent review of its work. This skill never spawns sub-agents — it produces findings only — which structurally prevents the spec/design/plan/implement review loop from recursing.

Write specifications (WHAT/WHY) for features, changes, and improvements. Spec is approved when `wiki/STATUS.md` shows the spec row at `spec approved` — written there by a human signal (label, comment, APPROVED review, or in-session message) that `kata-dispatch` or the active agent propagates. Use when proposing changes, capturing findings as actionable specs, or evaluating spec quality. Pair with the `kata-plan` skill for the HOW side.

Merge gate for open pull requests. Verify contributor trust, classify PR type, rebase on main, fix mechanical CI failures, gate on `wiki/STATUS.md` approval state, and merge passing PRs. Sole external merge point.

Detect time-bounded artifacts whose retention window has passed, confirm each one's durable signal is preserved elsewhere, then retire them — past-period wiki files directly and terminal spec directories through a retention PR. Use on a scheduled archivist shift, or when the specs tree and weekly logs grow faster than they retire and repository search signal drops.

Perform a deep-dive codebase-health review — dead code, duplication, inconsistency, and accumulating debt — one area per run against a coverage map. Use on a scheduled developer-experience shift, when reviewing a change for maintainability, or when debt in one area has started slowing every agent invocation.

Generate JavaScript types, service base classes, typed clients, gRPC definitions, and MCP field metadata from proto contracts, or download a pre-generated bundle. Use when a proto definition changed and the generated code must be regenerated to match, when setting up a gRPC product for the first time, or when a runtime image needs to fetch the generated bundle.

Get career guidance and output review grounded in your organization's engineering standard. Use when a promotion conversation ended without specifics and you need evidence of what to improve, when reviewing agent output and you want a second opinion against the standard, or when asking about skills, levels, and career expectations. Also covers setting up the Guide service stack and ingesting knowledge content.

Build the knowledge indexes from HTML sources: process resources, then graphs and vectors. Use when populating the retrieval indexes an agent queries, as a build-pipeline step run in order before search or graph queries work.

Query the knowledge indexes: search by meaning, answer relationship questions with triple patterns, or list graph subjects by type. Use when you need ranked semantic results or graph relationships without standing up a vector database or SPARQL endpoint.

Produce a complete eval dataset from a single DSL file so you can prove agent changes with reproducible evidence, and run substrate identity verbs against any Supabase stack implementing the Substrate Contract. Use when setting up an eval, bootstrapping a realistic environment, regenerating a dataset after a schema change, or provisioning, picking, and issuing personas for an interview run.

Leave the current worktree and clean it up. Use when the work in a worktree is merged or abandoned and the session should return to the original directory.

Query recorded OpenTelemetry spans with JMESPath and render them as Mermaid sequence diagrams. Use when you need to read spans back from the span index, filter by trace or resource id, and see the call flow without wiring up a tracing UI.

Conduct a JTBD switching interview to test one of the repository's products. Build a persona grounded in the installation's synthetic content with the situation drawn from the chosen JTBD entry, hand the job to the agent at the public website in two Ask calls, and capture findings as GitHub issues classified against the job.

Bootstrap the Co-Aligned instruction architecture in a repository. Use when a repo has no layered agent instructions yet, when adopting the Co-Aligned standard, or when wiring the `coaligned` checks into the repository so instruction layers, jobs, and invariants stay enforced.

Author or repair an instruction layer to the Co-Aligned standard — an agent profile, agent reference, SKILL.md, skill reference, or checklist. Use when adding an agent or skill, when `coaligned instructions` flags a length breach, or when one layer has started restating another.

Run the full Co-Aligned check suite and act on what it finds. Use for a periodic co-alignment health check, when CI reports a `coaligned` failure, or before a release — to triage every finding and route it to the fix that owns it.

Author a declarative invariant rule module for `coaligned invariants`. Use when a repository needs to enforce its own architectural rule — a forbidden import, a value that must agree across files, a directory shape — as a `.coaligned/invariants/*.rules.mjs` module the CLI discovers and runs.

Author and maintain Jobs To Be Done entries for the Co-Aligned standard. Use when writing a Big Hire or Little Hire, when adding a `<job>` tag, when `package.json .jobs` blocks are stale, or when `coaligned jtbd` reports a schema or freshness failure.

Consolidate a sprawling backlog of overlapping issues and PRs into a focused set. Partition into clusters, synthesize each cluster's root cause via grounded coding into one spec + design, then close the redundant issues and superseded PRs as duplicates. Use when ad-hoc per-item handling keeps reinventing the same moves. Improvement-coach scope extension.

Write and review documentation in the websites/ folder. Scheduled runs review one topic in depth for accuracy, audience purity, and staleness. Interactive runs write or update pages following documentation standards. Use when writing, editing, auditing, or reviewing documentation, or running scheduled documentation review.

Implement a spec by studying its spec.md and plan, then executing the plan step by step. Use when a spec and plan are approved and ready for implementation. Triggers: "implement spec NNN", "implement the plan for spec NNN", "execute plan NNN", "build spec NNN", "start implementation of NNN".

Triage open GitHub issues against the product vision. Classify each as mechanical fix, product-aligned spec, or out-of-scope, and produce a report the agent acts on. Operates on issues only — PR mergeability is kata-release-merge.

Cut new versions of packages with unreleased changes on main. Determine version bumps, update package.json files, tag releases, push tags, and verify publish workflows. Canonical source for the release procedure.

Perform a holistic security review of the repository. Assess GitHub Actions supply chain, dependency hygiene, credential leak controls, CI audit gates, and application-level vulnerabilities. Use when reviewing PRs for security impact, auditing the repo posture, or investigating a reported vulnerability.

Apply security updates to the repository. Triage open Dependabot PRs against repository policies, review npm audit findings, and action dependency vulnerabilities. Merge PRs that pass all checks, fix minor issues on a new branch, or close PRs that violate policy.

Toyota Kata coaching protocol for facilitated sessions. Used by the improvement coach (facilitator) and by domain agents who participate via the Ask/Answer/Announce orchestration tools. Same five coaching kata questions across team storyboard meetings and 1-on-1 coaching sessions; mode-specific guidance lives in references/team-storyboard.md and references/one-on-one.md.

Set up the Kata Agent Team in your repository. Walks through GitHub App creation, secret configuration, agent selection, and generates workflow files. Use when setting up a new Kata installation or adding agents to an existing one.

Curate the wiki (agent memory) for cross-team collaboration. Run `fit-wiki audit` to fix every contract violation, clear expired claims, verify summary accuracy against weekly logs, follow up on stale teammate observations, and keep MEMORY.md current. Use when running scheduled wiki curation, auditing agent memory health, or checking cross-agent communication.

Stand up a new Monorepo-standard repository end to end — bootstrap the skeleton, install the skill packs, run coaligned-setup then kata-setup, and fill the seams neither owns (root package.json, directory tree, agent profiles, CI, remote creation). Use when creating a Forward Impact-style repo from nothing, or when one exists but the cross-cutting wiring is missing.

Define what good engineering means so roles have clear, defensible expectations, and provision activity-database substrates. Use when defining or updating skills, capabilities, behaviours, disciplines, tracks, levels, or questions; when pushing rosters, syncing GetDX snapshots, or ingesting GitHub artifacts; or when staging a seeded substrate.

Follow a live Anarlog session in real-time, coaching the user through a meeting or interview. Understands context from the session title, knowledge base, and candidate pipeline. Provides talking points, flags gaps in coverage, and suggests follow-up questions as the conversation unfolds. Use when the user asks to follow, shadow, or coach them through a live meeting.

Add a lightweight text-highlight review overlay to an HTML deck. Lets you highlight text on slides and capture feedback as a sidecar JSON (with source line/column + context) that an agent can act on in small iterations. Use when the user asks to add review/annotation/highlight/comment capability to a deck, make a deck "reviewable", or wants to mark up slides for revision. Pairs with the deck-create skill.

Process synced email/calendar files from ~/.cache/fit/outpost/ and ad-hoc document files (e.g. from ~/Desktop/ or ~/Downloads/) to extract structured knowledge into Knowledge/ as Obsidian-compatible markdown notes. Use on a schedule, when the user asks to process/extract entities, or when invoked by another skill (e.g. organize-files). Builds the core knowledge graph from raw data.

Prepare for meetings by gathering context from the knowledge base and calendar. Use when the user asks to prep for a meeting or wants a briefing on upcoming meetings. Creates personalized briefings with attendee history, open items, and suggested talking points.

Synthesize all evidence (CV screening, interview assessments, transcripts) into a final hiring recommendation. Produces a comprehensive recommendation document with full evidence trail, level/track/discipline confirmation, and a clear hire/no-hire decision. Use when all interview stages are complete.

Screen candidate CVs against the agent-aligned engineering standard to decide whether to invest interview time. Produces a structured screening assessment with interview/pass recommendation and suggested interview focus areas. Use when the user asks to evaluate a CV or when a new CV is detected.

Scan synced email threads for recruitment candidates, extract structured profiles, and create/update notes in Knowledge/Candidates/. Use when the user asks to track candidates, process recruitment emails, or update the hiring pipeline.