Coding & Development

Agent skill 'subagent-driven-development' from DanhDue/ai-agent-tools.

Agent skill 'debugging-investigator' from alphasafal/production-agent-skills.

Agent skill 'product-manager' from alphasafal/production-agent-skills.

Invoke Google's Antigravity CLI (agy) from inside Claude's own sandbox, authenticated via the user's personal Antigravity subscription sign-in (never an API key). Use this ONLY when the user explicitly asks to use Antigravity CLI, agy, or "this bridge" for a task — never invoke it on your own judgment the way you might reach for a general-purpose tool. Typical explicit phrasings: "Antigravityに聞いて", "agyに投げて", "use antigravity for this", "run this through the antigravity bridge".

Agent skill 'verification-before-completion' from DanhDue/ai-agent-tools.

Deep project review skill for thorough codebase and architecture audits.

Make CLIs agent-friendly: structured help, machine-readable output, and safe automation hooks.

Pre-flight validation and quality gates for batch LLM operations. ACTUALLY tests samples through LLM before burning tokens. Uses SPARTA contracts for DuckDB validation queries. Integrates with task-monitor for enforced quality gates.

Perform a read-only, defect-first review of a specified code change and return every actionable finding. Use when another agent delegates review of uncommitted changes, a base-branch diff, a commit, or custom review instructions.

Install Codex skills into $CODEX_HOME/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).

Standardized compliance QRA benchmarks against candidate LLMs

从已完成的代码框架、配置和工程资料建立、新增、补充或重构技术文档的信息架构、内容边界与表达方式。

基于项目证据创建或更新指定作用域的 AGENTS.md；发现既有匹配指南时，先取得文件选择、合并或覆盖和最终写入确认。

将已收敛的 map、ADR、spec 或 implementation plan 编译为 Gate 计划、状态账本和跨轮复用的固定执行提示词。

Shared GPT + classifier inference gateway for persona monitor tasks. Routes validation and classification through a 4-tier cascade: heuristic → classifier → local GPT → scillm.

Self-improvement workbench for /assistant. All the tools needed to diagnose, train, evaluate, and promote models in a continuous loop. The "warm pond" where /assistant evolves its own inference stack.

Step back and critically reassess project state. Use when asked to "assess", "step back", "fresh eyes", "check alignment", "sanity check", "health check", "prune documentation", or "evaluate what's working". Offers documentation pruning and doc-code alignment analysis. Offer to run after major changes (don't auto-run).

Generate and query the centralized agent identity registry. Scans .pi/agents/*/AGENTS.md, parses frontmatter, outputs agents-registry.json and optionally syncs to /memory for semantic search.

File-based inter-agent messaging with headless dispatch. Check inbox, send bugs/requests to other projects, automatically spawn headless agents to fix bugs, and track progress via task-monitor.

Canonical map and shared contracts for the agent-governance ecosystem: the pi.receipt_envelope.v1 boundary envelope, the component graph, and the rules for which component owns which schema. Use when wiring a skill or extension into the shared receipt world, when asking how shame, triage-error, tau, ask, project-watchdog, ops-herdr, ponytail, and Memory fit together, or when validating an envelope.

Artifact-driven status surfaces for long-running project-agent work. Maintains status.json, events.jsonl, proof manifests, and a stale-aware STATUS.html so humans can tell where the agent is, what passed, what is still unproven, and what decision or action is next — without dashboard theater.

The `mote` skill teaches AI agents **when and how** to publish Markdown with Mote — it contains no upload implementation of its own (it drives the CLI or the MCP tools).

Track multi-session plans on a locally-running kanban board (github.com/hechtch/kanban) via its agent REST API. Claim a plan, move it through todo/doing/blocked/awaiting_merge/done, leave notes, and set the git branch. Use at the start of any non-trivial multi-session plan, when resuming such a plan, or when the user asks what's on the board or where a plan stands.

Generates secure, memorable passwords using 3 random dictionary words with leet-speak substitutions. Use when the user asks to generate a password, create a new password, or needs a secure passphrase.