SpecFlow

Spec-driven development orchestrator. Creates and executes spec-workflow specifications from an issue — Requirements → Design → Tasks → Implementation with dashboard approvals and subagent dispatch.

Run the i18n validation script to check for missing, extra, or misformatted translation keys before publishing

On-demand project health check — code quality (CGC), security (Codacy SRM), instruction file drift (CLAUDE.md vs Agents.md vs Gemini.md), issue landscape, index health. Triggers: "audit", "health check", "check project health", "scan codebase", "code health", "instruction drift", "check for drift".

Casual, freeform discovery mode (Phase 0 of the Spec Flow Lifecycle). Explore an idea without committing to code, issues, or specs. Dispatch subagents for research, surface past work from mem0, and build context until the idea is ready for an issue and formal spec. Triggers on "/chat", "/talk", "let's explore", "I have an idea", "what if we".

Structured brainstorm and research phase (Phase 1 of the Spec Flow Lifecycle). Takes an issue ID, dispatches research agents, runs codebase analysis, and produces a Discovery Brief with all open questions resolved. Bridges the gap between casual exploration (/chat) and formal spec creation (/spec). Triggers on "/discover", "/brainstorm", "research this issue", "explore this before we spec it".

Issue management — create, read, update, list. Backend is driven by `.specflow/config.json` `issue_backend` field: `plane` (Plane MCP, primary going forward) or `docvault` (legacy markdown — for projects not yet migrated). Triggers on: "create issue", "issue", "new issue", "new bug", "file a bug", "track this".

Migrate a user-level skill into the specflow plugin's skills directory. Use when porting a proven skill from ~/.claude/skills/ to the plugin for distribution. Covers generic vs project-specific classification, MCP prompt extraction, deprecation of user-level shims, and verification.

Universal session boot — indexes code, digests past sessions, checks code health, gathers project status, and presents a unified startup report. Works on any project with a .claude/project.json. Run at the start of every session. Triggers: "prime", "boot", "get me up to speed", "session status", "where are we", "what's the status", "catch me up", "project overview", "what's open". Also use proactively at session start if the user hasn't run it yet.

End-of-session retrospective that extracts prescriptive lessons from the current conversation and saves them to mem0 as structured retro-learning memories. Unlike digest-session (descriptive — what happened), retro is prescriptive — what to do differently next time. Triggers on "/retro", "session retro", "what did we learn".

Lightweight session reorientation — queries session-rag + mem0 for recent session context, checks git state, and surfaces 5 most recent open issues. Fast alternative to /prime for continuing same-day work or resuming after a short break. No indexing, no code search, no agent dispatch. Triggers: "start", "quick start", "resume", "orient me", "where were we".

End-of-session orchestrator — verifies session work, runs retro inline (if not already done), and saves a curated summary + retro lessons to mem0. No DocVault digest write — session-rag + mem0 are the durable session record. Run at end of any session, optionally after /pr-cleanup. Triggers: "wrap", "wrap up", "close session", "end of session", "done for the night", "finish up", "closing time".

Promote a battle-tested user-level skill to the specflow shipped skills directory (specflow/skills/)

Post-merge branch and worktree cleanup. Prunes remote-tracking refs, deletes merged local branches and their worktrees, pulls main fast-forward, and reports the result. Run after a PR is merged, before /retro and /wrap. Triggers: "pr-cleanup", "clean up branches", "prune worktrees", "clean after merge".

Publish a battle-tested specflow template by writing it to the canonical source (`src/markdown/templates/{name}.md`), regenerating the matching DocVault KB snapshot guide page, building, testing, committing, pushing, and stopping before `npm publish` for manual passkey auth. The canonical source IS `src/markdown/templates/`; the DocVault guide pages are derived KB snapshots, not editable sources. Mandatory backup-before-edit. Triggers on "publish templates", "ship template changes", "promote template", "sync templates to npm", "publish a template".