claude-agent-pack

Orchestrates the full agent-pack pipeline for a task: git-engineer → [tech-lead] → engineer(s) → code-reviewer → [security-reviewer] → [performance-reviewer] → smell-reviewer → test-engineer → merge-reviewer → git-engineer (push/PR). Use when implementing a feature, fix, or change end-to-end. Trigger this when someone says: implement this, build this feature, make this change, add this functionality, code this up, I need this feature built, ship this. Do NOT use for targeted bug fixes with a known root cause — use /hotfix or /debug instead. Do NOT use for pure restructuring with no behavior change — use /refactor instead.

Review changed files or an open PR with code-reviewer, smell-reviewer, security-reviewer, and performance-reviewer. Use when the user wants a quality check on a diff, a PR number, or the current working-tree changes without running the full implement pipeline. Trigger this when someone says: review my PR, check my code, look at this diff, quality check, code review, does this look good, review PR #123, give me feedback on these changes, smell check, structural review. Do NOT use when the user wants to implement new changes — use /implement instead.

Decompose a complex task using tech-lead and optionally pressure-test the plan using devils-advocate before implementation begins. Use when the task is ambiguous, spans multiple concerns, or when the user wants to stress-test an existing proposal before committing. Trigger this when someone says: how should I approach this, plan this out, I am not sure how to do this, think through this with me, is this a good approach, pressure-test my plan, help me design this. Do NOT use when the task is already well-scoped — go straight to /implement instead.

Refactor existing code with impact analysis first. Routes to tech-lead for blast-radius assessment, then engineers, with heavy emphasis on test coverage. Use when restructuring code with no intended behavior change. Trigger this when someone says: refactor this, clean up this code, extract this pattern, restructure this, this code is messy, improve the code structure, rename this across the codebase, remove duplication. Do NOT use when the change modifies behavior — use /implement instead. Do NOT use for single-file cleanup — just edit the file directly.

Scaffold a new feature end-to-end: API contract → database schema → backend service/controller → frontend. Use when building something new from scratch across multiple layers. More opinionated than /implement — always runs the full vertical slice in the correct dependency order. Trigger this when someone says: build this feature from scratch, scaffold a new endpoint, I need a full-stack feature, new feature end-to-end, build the whole thing. Do NOT use when the task is already partially built or only one layer needs changes — use /implement instead.

Read and create Azure DevOps work items and PRs via the az CLI (devops extension) for the org/projects configured in AZURE_DEVOPS_ORG/AZURE_DEVOPS_PROJECTS. Use for one-off Azure Boards/Repos operations outside the code pipeline: viewing or creating work items, viewing PRs, commenting, updating state, linking work items to PRs/commits. Trigger this when someone says: check my work items, list open PRs in Azure DevOps, create a work item, comment on PR #N, close this work item, link this commit to a work item, connect to azure devops, smoke test the az cli. Do NOT use for GitHub PRs/issues -- that's skills/devops-github. Do NOT use for a full code review or implementation pipeline -- use /review-pr or /implement instead.

Read and create GitHub PRs and issues via the gh CLI for repos configured in GITHUB_ORG/GITHUB_REPOS. Use for one-off GitHub Boards/Repos operations outside the code pipeline: viewing PRs or issues, commenting, changing state, linking commits. Trigger this when someone says: check my PRs, list open issues, create a GitHub issue, comment on PR #N, close this issue, link this commit to an issue, connect to github, smoke test the gh cli. Do NOT use for Azure DevOps boards/work items -- that's skills/devops-azure. Do NOT use for a full code review or implementation pipeline -- use /review-pr or /implement instead.

Use when unsure whether a request is a GitHub or Azure DevOps operation, or when someone says "devops help", "connect to devops", "check my work items or PRs", "devops?". Routes to devops-github or devops-azure based on intent. Also use as a fallback when a request doesn't clearly map to either one.

Fast-track fix pipeline for production incidents. Skips worktree isolation and planning ceremony. Still requires code-reviewer and merge-reviewer as a safety gate. Use only when the root cause is already known and the fix is well-understood. Trigger this when someone says: emergency fix, production is broken, quick patch, urgent fix, hotfix, critical bug in prod, I know the fix I just need to apply it. Do NOT use when diagnosis is still needed — use /debug first. Do NOT use when the fix touches more than 3 files, adds dependencies, or changes the schema — use /implement instead.

Validates required frontmatter fields and body content in every agent and skill file in the pack by running scripts/lint-agents.sh. Interprets failures with specific fix instructions. Use before committing new or modified agents/skills, or as a CI gate. Trigger this when someone says: lint agents, validate agents, check agent files, run lint-agents, validate the pack, check if my agent is valid, did I write the skill correctly.

Synthesize a daily recap of Claude activity for the current project and write it to the Obsidian vault. Reads the day's auto-logged session notes, git history, and current-state note, then produces a narrative "what happened today" rollup that the link-only daily note lacks. Use when someone says "recap today", "daily recap", "write a recap", "summarize today's work", "obsidian recap", "wrap up the day", "end-of-day summary". Defaults to today; accepts a date argument (YYYY-MM-DD). Requires OBSIDIAN_VAULT_PATH to be set.

Use when unsure which Obsidian command to use, or when someone says "obsidian help", "log something to Obsidian", "what obsidian commands are there", "obsidian?". Routes to the correct skill based on intent. Also use as a fallback when an Obsidian operation does not clearly map to capture, log, daily, search, or brief.

Runs a memory hygiene pass on the project memory/ directory. Reviews all active memory files against the current codebase and archives or supersedes stale entries. Use before major feature work or when memory files feel out of date. Trigger this when someone says: clean up memory, audit memory files, memory is out of date, stale decisions, memory hygiene, are our memory files current, check if decisions are still valid. Do NOT use when searching for a specific memory entry — use /memory-query instead.

Generate a structured orientation for a codebase. Reads the project, memory, and conventions to produce a mental model: architecture, key entry points, data flow, and known gotchas. Use when joining a project, returning after time away, or onboarding a new contributor. Trigger this when someone says: orient me, explain this codebase, how does this project work, I am new here, give me an overview, help me understand the architecture, catch me up. Do NOT use to document conventions — use /conventions instead. Do NOT use to plan a task — use /plan instead.

Generate or refresh a durable, directory-level map of the codebase — what each directory is for and its key entry-point files — stored as memory/architecture/repo-map.md and stamped with the git commit it was verified against. Use to create the map, refresh it after the tree drifts, or check whether it is stale. Trigger this when someone says: map the repo, build a repo map, what does each directory do, refresh the repo map, is the repo map stale, generate the file manifest, update the codebase map. Do NOT use to produce a full onboarding orientation — use /onboard instead (it reads this map). Do NOT use to search memory — use /memory-query instead.

Scaffolds a project with the Claude Agent Pack structure: copies CLAUDE.md, docs/CONVENTIONS.md (from template), docs/MEMORY-WRITING.md, and creates the memory/ subdirectories. Runs scripts/setup-project.sh and guides the user through next steps. Trigger this when someone says: set up this project, initialize the agent pack, scaffold the project structure, add Claude Agent Pack to this repo, new project setup, onboard this repo. Do NOT use to verify an existing installation — use /system-check instead.

Synthesize a context brief from recent Obsidian session logs and captures for the current project. Use before starting work to load project context without a specific search query. Trigger when someone says "brief me", "what have we been working on", "load project context", "catch me up", "obsidian brief", "what's the recent context", "summarize recent sessions". Does NOT require a search term — it proactively summarizes what's relevant. Read-only.

Show or open today's Claude activity note in the Obsidian vault. Use when someone says "show today's Obsidian note", "open daily note", "what did I do today", "obsidian daily", "today's activity", "what did Claude do today". Read-only — does not modify any files.

Search past Claude session logs and captures in the Obsidian vault. Use when someone says "find my notes about X", "search obsidian", "what did I log about X", "find that capture", "look up session from last week", "search my Claude notes". Returns matching entries with excerpts and wikilink references. Read-only.

Scaffold or update docs/CONVENTIONS.md by reading the actual codebase and interviewing the user. Use when setting up a new project or when conventions have drifted and need to be documented. Trigger this when someone says: document our team standards, set up conventions, what are our coding standards, conventions are out of date, update our style guide, write our CONVENTIONS.md. Do NOT use to read or apply existing conventions — those are loaded automatically. Do NOT use for architectural decisions — use /memory-audit instead.

Diagnose and fix a failing test, error, or unexpected behavior. Routes to the appropriate engineer based on the error context. Lighter than /implement — no worktree isolation, no full pipeline ceremony. Trigger this when someone says: something is broken, fix this error, my test is failing, why is this not working, I'm getting an exception, this is throwing an error, debug this, why does this crash. Do NOT use when the root cause is already known and the fix is small — use /hotfix instead. Do NOT use for architectural changes — use /implement instead.

Interview the user about a plan or decision through structured one-at-a-time questioning until reaching consensus. Produces a design brief artifact or hands off when the user signals readiness. Use when an idea is vague and needs shaping before planning or implementation begins. Trigger this when someone says: interview me, help me think through this, I have an idea but I don't know how to approach it, ask me questions about this, what should I be thinking about here. Do NOT use when the task is already well-scoped — use /plan instead.

Save a note, decision, or thought to the Obsidian vault right now. Use when someone says "capture this", "save this to Obsidian", "note this down", "log this decision", "obsidian note", "save this for later", "capture that idea". Requires OBSIDIAN_VAULT_PATH to be set.

Run smell-reviewer against a file, directory, or the current working-tree changes. Use when someone says: check for code smells, smell review, find structural issues, review for smells, smell check, look for god classes, find long methods, check for dead code, smell this file, smell-review PR #N. Also use after /implement or /review-pr when the user wants a dedicated structural pass. Do NOT use when the user wants a full review pipeline -- use /review-pr instead.

Search project memory files for a specific topic, decision, or constraint and return matching entries with citations. Use when you need to find what was decided about X, why Y was chosen, or whether Z is a known issue. Trigger this when someone says: what do we know about X, was there a decision about Y, check memory for Z, find the memory entry for X, what did we decide about Y, look up memory for X. Do NOT use for a full memory hygiene pass — use /memory-audit instead.

Verify the Claude Agent Pack installation and current project scaffolding in one pass. Runs check-readiness.sh and check-updates.sh, reports all failures and outdated items together, and offers to run remediation. Use before starting work on a project or when something seems off after a git pull. Trigger this when someone says: check my setup, is everything installed, something is not working, verify the pack, is my installation up to date, check if Claude Agent Pack is ready, run a health check. Do NOT use to initialize a new project — use /setup-project instead.