prisant-utilities

Run a structured peer review of a document with a second model, then synthesize the findings into decisions. Use when you need a second opinion on a plan, brief, spec, RFC, or design doc, or when the user says 'review this with codex', 'get a second opinion', 'peer review this', or 'close this review'. Three modes: --review generates a self-contained review request for another model; --respond adds requestor synthesis once findings come back; --close archives the pair, applies accepted changes, and writes unresolved decisions to a backlog.

Resume an interrupted work session by replaying its recorded handoff. Use when the user explicitly asks to resume: 'resume', 'continue session', 'pick up where we left off'. Reads the most recent session log, reports what was outstanding and the named next action, then confirms before acting. Do NOT fire on status questions like 'where are we' or 'what's next'; answer those directly rather than resuming. Companion to plab-wrap-session, which writes the log this reads.

Generate a paired guide bundle (standard MD + ADHD MD + quick-reference HTML + 1-2 page PDF) for any topic. Use when the user provides a GitHub URL, a tool name, or a concept and wants a structured explanatory document plus a printable operator card. Triggers on "create a guide", "build a cheat sheet", "explain X with a quick reference", "/plab-guide".

Initialize agentic development infrastructure in a project. Creates standardized folders and seed files: AGENTS.md (open standard agent instructions), CLAUDE.md (Claude-specific overlay), session logs written to gitignored _local/_session-logs/, and MADR v4 decision records. Three profiles: minimal (experiments), standard (working projects), public (open source). Non-destructive and idempotent - safe to run against an existing repository. Use when setting up a new project for AI-assisted development, or when adding agent support to an existing repo. Explicit invocation only: run /plab-init-project; it does not fire on its own. Pairs with /plab-wrap-session and /plab-continue-session, which write and read the session logs it scaffolds.

Create and manage a version-scoped release plan folder: scaffold it, promote or demote whole effort folders between _unassigned/ and a release, regenerate the aggregation table from disk, and report whether a version can be tagged yet against hygiene gates and a doc-update checklist. Use when scoping a release, moving efforts in or out of one, or checking release readiness: 'create the release plan', 'promote these efforts into v0.5.0', 'can we tag yet', 'check release readiness'. Do NOT fire on general planning talk, on requests for release notes or a changelog, or on questions about what already shipped. Refuses to invent or modify acceptance criteria - those live in specs written by /plab-spec; the release plan only aggregates.

Write a feature specification with numbered, source-cited acceptance criteria into a per-effort folder under docs/internal/release-plans/. Produces a spec.md carrying frontmatter, an agent-updated Task Summary block, and links to the related effort and plan. Writes into _unassigned/ by default; pass --target-release vX.Y.Z to write straight into a release folder. Use when the problem space is already explored and the requirements need a durable home that survives rewrites of the plan beneath them: 'write the spec', 'spec this effort', 'turn this into acceptance criteria'. Do NOT fire on a passing mention of the word spec, on questions about an existing spec, or on a request to implement one. Defines WHAT to build; the implementation-plan template under docs/internal/release-plans/ defines HOW and /plab-strategy-brief explores WHY.

Transform messy thinking into structured, decision-ready analysis documents. Use when a user provides raw, unstructured thoughts and wants them organized, expanded, and turned into an actionable artifact. Triggers on: 'help me think through', 'brain dump', 'summarize and expand', 'make sense of this', requests combining pros/cons with approaches and recommendations, or any stream-of-consciousness input needing structure. Also triggers on 'strategy primer', 'structured analysis', or '80/20 analysis'. Do NOT use for simple summarization of existing structured documents, editing polished prose, or pure research queries with no raw thinking provided.

Document and close agentic coding sessions with structured session logs. Use at the end of any coding session, when wrapping up work, when switching tasks, or when blocked. Defaults to a deep log with a verbose copy-paste-ready continuation prompt, runs a pre-wrap hygiene sweep (remote divergence, release state, doc drift, working-tree reconciliation with per-action confirmation), and always lists what it is waiting on from the maintainer with links. Triggers on 'wrap up', 'wrap session', 'end of session', 'session log', 'close out', or 'document this session'. Also use when the session is blocked and needs to capture blocker details for the next session.