claude_skills

Teaches an agent graph engineering — both halves. Knowledge graphs (ontology design, entity/relation/event extraction, fusion, GraphRAG/memory serving; distilled and translated from Southeast University's graduate Knowledge Graph course, npubird/KnowledgeGraphCourse, 4.4K stars) and task graphs (agent orchestration — parallel fan-out, verifier separation, the stop rule, human gates). Use when asked to build a knowledge graph, extract entities/relations from text, design an ontology, dedupe/merge entities, add graph memory or GraphRAG to an agent, orchestrate multi-agent workflows as a graph, or LEARN graph engineering — in teaching mode the agent explains each stage with worked examples and generates visual diagram artifacts.

When the user wants to edit, review, or improve existing marketing copy. Also use when the user mentions 'edit this copy,' 'review my copy,' 'copy feedback,' 'proofread,' 'polish this,' 'make this better,' 'copy sweep,' 'tighten this up,' 'this reads awkwardly,' 'clean up this text,' 'too wordy,' or 'sharpen the messaging.' Use this when the user already has copy and wants it improved rather than rewritten from scratch. For writing new copy, see copywriting.

This skill should be used when code search is needed (whether explicitly requested or as part of completing a task), when indexing the codebase after changes, or when the user asks about ccc, cocoindex-code, or the codebase index. Trigger phrases include 'search the codebase', 'find code related to', 'update the index', 'ccc', 'cocoindex-code'.

Use when working in a repository that uses bd or Beads for durable project task tracking, issue dependencies, blocker management, multi-session handoff, or shared work memory. Trigger when the user asks to find ready work, claim or close tasks, create follow-up work, inspect blockers, recover project context, or choose between local planning and persistent project tracking.

Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.

Verifies claims in backlog items, skill documentation, or plugin content against primary sources using web lookups. Spawns parallel verification agents that must use WebFetch/WebSearch/gh — training data recall is explicitly rejected as evidence. Produces VERIFIED/REFUTED/INCONCLUSIVE verdicts with citations. Use when items are marked UNVERIFIED or when verifying tool API claims, CLI flags, or documented software behavior.

Same reverse-thinking method as dh:rt-ica (see dh:rt-ica or dh-glossary for the definition), but localizes any MISSING input to the affected task only, as a non-blocking information-completeness pre-pass before task decomposition and plan generation. Use when grooming backlog items, generating plans, decomposing tasks under uncertainty, or working in brownfield and refactor scenarios. Produces completeness summary (APPROVED-FOR-PLANNING, APPROVED-WITH-GAPS, or BLOCKED-FOR-PLANNING), missing input report with dependency mapping, required unblock actions, and planning annotations for downstream tasks. Use dh:rt-ica instead at the S2 implementation gate where missing inputs must halt execution.

Wraps investigation requests with evidence-chain discipline. Use when asked to find out why something happens, research a root cause, debug an issue, or investigate unexpected behavior. Transforms vague investigation requests into reproducible-proof investigations with a 5-step protocol — disambiguate, reproduce, read source, build evidence chain, present findings. Invoke with /find-cause followed by a description of what to investigate.

Use when orchestration or planning agents are producing task plans, task prompts, or TASK.md instructions that must be unambiguous, verifiable, and resistant to hallucination. Applies CLEAR (Concise, Logical, Explicit, Adaptive, Reflective) to structure and write agent task prompts, then adds CoVe (Chain of Verification) checks where accuracy risk is meaningful. Activates on draft task prompts, swarm plans, migration tasks, and multi-step plans requiring independently executable steps.

Create high-quality Claude Code agents from scratch or by adapting existing agents as templates. Use when the user wants to create a new agent, modify agent configurations, build specialized subagents, or design agent architectures. Guides through requirements gathering, template selection, and agent file generation following Anthropic best practices (v2.1.63+).

Scientific delegation framework for orchestrators — provide observations and success criteria while preserving agent autonomy. Use when assigning work to sub-agents, before invoking the Agent tool, or when preparing delegation prompts for specialist agents.

Builds comprehensive Claude Code skills using parallel research agents — categorization, parallel documentation gathering, anti-hallucination checkpoints, and final validation. Use when building a skill from official docs, when "research for skill" or "create comprehensive skill" is requested, or when extensive multi-source documentation gathering is needed before skill creation.

Configure clang-format code formatting. Use when: user mentions clang-format or .clang-format, analyzing code style/patterns, creating/modifying formatting config, troubleshooting formatting, brace styles/indentation/spacing/alignment/pointer alignment, or codifying conventions.

When writing a git commit message. When task completes and changes need committing. When project uses semantic-release, commitizen, git-cliff. When choosing between feat/fix/chore/docs types. When indicating breaking changes. When generating changelogs from commit history.

Add automated documentation updater to any Claude skill. Creates a Python sync script that downloads upstream docs, processes markdown for AI consumption, and maintains local cache with configurable refresh. Collects template variables, then delegates implementation through 5-phase workflow. Use when adding auto-updating reference documentation to plugins or skills.

Runs the description-drift experiment — spawns all Claude Code agents simultaneously to collect self-reported capabilities, then compares them against static frontmatter descriptions to reveal how reliable orchestrator routing based on descriptions actually is. Use when measuring description drift across the agent fleet, re-running the capability collection experiment, analyzing a specific agent's self-reported capabilities, or auditing whether frontmatter descriptions accurately reflect agent behavior.

When setting up commit message validation for a project. When project has commitlint.config.js or .commitlintrc files. When configuring CI/CD to enforce commit format. When extracting commit rules for LLM prompt generation. When debugging commit message rejection errors.

Orchestrate research entry lifecycle in ./research/ — create, batch-import, refresh stale entries, and validate structure. Use when asked to add a tool, research a URL, document a library, refresh research, validate entries, or given any tool or library URL. Supports --batch (parallel multi-URL), --rerun (refresh one or all entries), and --validate (structural check with auto-fix of error-severity issues).

Orchestrator context window discipline enforcement. Prevents the orchestrator from reading source files it will not edit, running diagnostic commands that waste context, and rationalizing delegation bypasses. Use when setting up orchestrator guardrails, reviewing delegation discipline, or diagnosing context window waste in multi-agent workflows. Activates PreToolUse hooks that surface decision points before source file reads and diagnostic command execution.

You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or when users request help with ideation, marketing, and strategic planning. Explores user intent, requirements, and design before implementation using research-validated prompt patterns.

Write and evaluate effective Python tests using pytest. Use when writing tests, reviewing test code, debugging test failures, or improving test coverage. Covers test design, fixtures, parameterization, mocking, and async testing.

GitLab CI/CD pipeline configuration and GLFM documentation expertise. Use when modifying .gitlab-ci.yml, optimizing pipelines, testing with gitlab-ci-local, writing GitLab README/Wiki content, configuring Docker-in-Docker workflows, or implementing CI Steps composition.

Agent Skills Open Standard reference (agentskills.io). Use when creating portable skills for Claude Code, Cursor, Gemini CLI, OpenAI Codex, VS Code, Roo Code, and 20+ compatible agents. Covers frontmatter schema, naming rules, directory structure, progressive disclosure, validation, and authoring. Load before creating cross-agent skills.

Use when building, extending, or debugging FastMCP v4 (or v3) Python MCP servers. Activates on FastMCP tool/resource/prompt creation, provider and transform implementation (CodeMode, Tool Search), auth setup (MultiAuth, PropelAuth, KeycloakProvider), client SDK usage, FastMCPApp and Generative UI server building, fastmcp-slim client-only installs, nginx reverse proxy deployment, Prefab Apps, OTEL observability, testing, and v3→v4 migration. Grounded in official FastMCP v4 docs plus locally-verified v3 gotchas — zero speculation.

Look up prior Claude Code sessions when context is lost or forgotten. Use when asked what was done before, what happened in the last session, or any request to recall past conversation history, prior decisions, experiments, or outcomes. Indexes and searches raw JSONL transcripts from ~/.claude/projects/ via DuckDB. Returns verbatim user messages, summarizes AI actions and sub-agent outcomes, detects tool errors and user frustration signals, and reports tool usage statistics. Summaries cached at ~/.claude/kaizen/session-summaries/.