claude-code-handbook

Slash commands for Claude Code. Copy to `~/.claude/commands/` (global) or `.claude/commands/` (project).

Use when cleaning up project clutter — temporary files, backups, build artifacts, abandoned worktrees. Identifies and removes irrelevant files safely.

Use when starting a new task and need to gather the right context first. Assembles a context package without starting the task.

Use during long sessions when the context window fills up. Manages token budget, suggests compaction, identifies what to drop.

Use when creating or updating the daily task log. Tracks accomplishments, blockers, and plans for cross-day continuity.

Use BEFORE committing to an implementation approach. Generates and compares 2-3 solution candidates with trade-offs.

Use when syncing JIRA ticket data into local docs for sprint planning. Creates structured user-stories/ files from JIRA issues.

Use when reviewing decomposed user story subtasks. Suggests consolidation opportunities for more efficient development.

Use when reviewing large/complex PRs (>300 lines, multi-concern). Dispatches specialized reviewer agents in parallel. For standard PRs, use /review-and-comment instead.

Use when promoting code between environments (develop → SIT → UAT → PROD). Executes the comprehensive environment promotion PR workflow.

Use when reviewing and updating quarterly goals and OKRs progress. Generates quarterly engineering metrics summary.

Use when analyzing HRP review effectiveness metrics and outcomes data. Identifies improvement areas across review periods.

Use when debugging a bug to find its origin. Traces backward through call chains to find the original trigger. Never fixes just where the error appears.

Use at the end of a session to capture context for the next session. Generates session handoff document with state, decisions, and next steps.

Use BEFORE proposing solutions. Transforms ambiguous requirements into structured, actionable problem definitions. Does NOT propose solutions.

Use when creating a new Claude Code skill or command. Guides through type selection, structure, frontmatter, and testing.

Use when checking current sprint progress. Shows JIRA board status, velocity metrics, and team blockers.

Use when syncing Claude Code config between machines. Subcommands push, pull, status, bootstrap. Templates paths, redacts tokens, propagates memory.

Use when investigating a bug systematically. Apply structured debugging methodology — NO FIXES WITHOUT ROOT CAUSE INVESTIGATION FIRST.

Use when implementing a feature or bug fix with TDD. Red-green-refactor cycle. If you didn't watch the test fail, you don't know if it tests the right thing.

Use when generating a weekly engineering summary. Aggregates accomplishments, learnings, and plans from daily logs.

Custom subagents and Agent Teams for Claude Code. See [`agents/README.md`](agents/README.md) for the full guide.

Code review specialist. Dispatched by /orchestrate-review for parallel review. Can also be invoked directly for focused PR analysis.

Sprint planning and JIRA coordination specialist. Use for sprint reviews, capacity planning, story assignment, and blocker analysis.

Hill-climbing optimizer for the headless review prompt. Runs test-score-mutate loops against a real PR. Use when someone says "autoresearch", "optimize the review prompt", or "improve the headless review".

Use this skill when you need documentation for a third-party library, SDK, or API before writing code that uses it — for example, "use the OpenAI API", "call the Stripe API", "use the Anthropic SDK", "query Pinecone", or any time the user asks you to write code against an external service and you need current API reference. Fetch the docs with chub before answering, rather than relying on training knowledge.

Single-shot feature implementation from ticket to merged PR. Orchestrates brainstorming, API research, pattern scanning, TDD, code review, and verification — designed to deliver production-grade code in one pass.

Security code review using OWASP Top 10:2025 standards. Use when reviewing auth code, payment logic, user input handling, or when someone says "security review", "check for vulnerabilities", "OWASP check", or "is this code secure?".

Follow-up on PR review comment replies. Use when someone says "check if they addressed my comments", "follow up on PR feedback", "did they fix what I asked?", or when evaluating developer responses to review feedback.

Comprehensive PR code review with subagent isolation. Use when someone says "review this PR", "code review", "check this code for issues", or needs to analyze changes for correctness, security, and performance.

Use when setting up SpacetimeDB projects, running the dev server, publishing modules, or managing databases via CLI. Triggers on: "spacetime init", "start a SpacetimeDB project", "publish module", "spacetime logs", "spacetime sql", "spacetime generate", or any CLI workflow question.

Use when learning SpacetimeDB architecture, evaluating it for a project, or needing conceptual grounding before writing modules. Triggers on: "what is SpacetimeDB", "should I use SpacetimeDB", "how does SpacetimeDB work", "explain SpacetimeDB", or any architectural discussion about SpacetimeDB.

Use when writing SpacetimeDB server modules in Rust — table definitions, reducers, lifecycle hooks, scheduling, views, procedures, and migrations. Triggers on: "create a SpacetimeDB module", "add a table", "write a reducer", "SpacetimeDB Rust", server-side SpacetimeDB work, or Rust module code generation.

Use when building TypeScript or React clients for SpacetimeDB — connection setup, subscriptions, calling reducers, authentication, and React hooks. Triggers on: "connect to SpacetimeDB", "SpacetimeDB client", "useTable", "subscribe to SpacetimeDB", "SpacetimeDB React", "SpacetimeDB TypeScript", or client-side SpacetimeDB work.

Use when building React apps with TanStack Query, Router, Table, or Form — especially with SpacetimeDB as the backend. Triggers on: "TanStack Query", "useQuery", "useMutation", "TanStack Router", "TanStack Table", "TanStack Form", "React data fetching", or building frontends with SpacetimeDB.

Run the full PR verification checklist — code organization, build, test, mocks, squash, push, update PR description, and validate all review comments.

6 hooks across 4 events