All authors
Claude Skills by CritosAI
github.com/CritosAI7 skills0 installs7 views
- Crit Agent OnboardTurn a fresh chat into one of the agents rostered in THIS repo. Use at the START of a new session when you were NOT handed a RESTART block: it reads the roster, lets you pick, runs the boot order from the session contract including the agent's own definition, and gives a short in-role report. It REFUSES rather than improvise: if the roster or the chosen agent's definition cannot be found, it stops and names what is missing — and it refuses any agent the roster does not list. The "open" booken...Votes: 0GitHub stars: 3
- Crit CheckpointClose a session cleanly and emit a paste-ready RESTART BLOCK to resume in a fresh chat. Use ONLY when winding down a session you intend to pick up later. It reconciles (/crit-status-check), saves STATUS/DECISIONS/LESSONS to truth, tidies docs (/crit-doc-lint), routes cross-domain ripples to HANDOFFS, and re-emits your onboarding pattern refreshed with the current state. This is NOT the per-round close-out — that lighter rule (commit + HANDOFFS + /crit-doc-lint + /crit-status-check) lives in t...Votes: 0GitHub stars: 3
- Crit Doc LintRun the deterministic documentation-hygiene battery — twenty-one checks over links, orphan docs, the inbox, the roadmap, the memory logs, the pinned system tier and the agent definitions. The steward's cheap mechanical audit pass; use before a governance audit, at a session checkpoint, or after the doc tree was reorganised. The interpretive audits (redundancy, real drift, taxonomy, the cross-domain rollup) stay with the steward's own reasoning — this only runs the deterministic layer.Votes: 0GitHub stars: 3
- Crit Memory IndexGenerate MEMORY-INDEX.md — the deterministic, read-only ATLAS of a project's memory: every ADR ID, every lesson ID, each memory folder with its writer and size, and the registers, each as a file:line POINTER. Use when a project's memory has grown past what an agent can read at boot, when you need to find which log owns a fact, or in a round-close to keep the atlas fresh (--check tells you if it is stale). It holds no facts, only addresses — so it cannot drift into a second source of truth. Ne...Votes: 0GitHub stars: 3
- Crit Scaffold ProjectStand up the agentic layer in a repo: PIN the versioned system tier (the five contracts) into .critos/system/, install the A1 definition into .claude/agents/, and create the project tier (settings + roster) and the work tier (memory, registers) as skeletons. Use ONCE when adopting the method in a new repo. Every step is idempotent — what already exists is reported and left untouched, never overwritten — so an interrupted run is resumed by re-running it. It does NOT configure the project: that...Votes: 0GitHub stars: 3
- Crit Status Check"Fai il punto" — before you tell the user (or yourself) what's left on your plan, reconcile your STATUS against reality on BOTH axes: git (what landed that STATUS doesn't know) and the shared inbox (what is OWED BY YOU that STATUS may call "blocked on others"). Runs a deterministic pass so you answer from truth, not from a possibly stale STATUS. Use whenever asked "what's left / where are we / dammi lo stato del piano", at the start of planning, or when a STATUS feels old. The script is read-...Votes: 0GitHub stars: 3
- Crit TeamThe steward's guided work on a project's TEAM — which agents exist, and which DOMAIN each one owns. Use it (1) right after /crit-scaffold-project, to decide which agents the project should have — a short interview, a MEASURED reading of the repo, a proposed partition the user rules on; (2) at any time, to check whether the partition still fits — which agent has outgrown its domain, which border is in the wrong place, what should split or merge; (3) to conduct a split or a merge step by step. ...Votes: 0GitHub stars: 3