agent-skill

Bootstrap CLAUDE.md, .claude/agents/, hooks, and plugin wiring with /agent-init

Cost-unrestricted patterns starting with /visual-qa (visual regression + LLM analysis via Playwright MCP)

Run agent-init for Codex CLI projects — emits AGENTS.md + .codex/skills/

Run agent-init for GitHub Copilot CLI — emits .github/copilot-instructions.md + AGENTS.md

Run agent-init for Gemini CLI (a.k.a. 'antigravity') — emits GEMINI.md + .gemini/skills/

Run agent-init for Cursor — emits .cursor/rules + .cursor/agents/

visual-qa scaffold for Codex CLI — emits .visual-qa.json + Playwright MCP entry for config.toml

visual-qa scaffold for GitHub Copilot CLI — emits .visual-qa.json + Playwright MCP entry

visual-qa scaffold for Gemini CLI — emits .visual-qa.json + Playwright MCP entry for settings.json

visual-qa scaffold for Cursor — emits .visual-qa.json + Playwright MCP entry for .cursor/mcp.json

Theme B — cost-conscious long-session optimisation: context-mode aggressive integration, prompt cache priming, automatic summariser hooks, end-of-session audit

Theme B for Cursor — advisory-only via .cursor/rules/thrift.mdc; no programmatic hooks, no cache prime, no metrics

Theme B for Copilot CLI — YAML/JSON hooks under .github/hooks/, store_memory summariser scratch, OpenAI-flavoured cost estimator, Phase 4 cache prime disabled by default (Copilot intermediates the model layer)

Theme B for Codex CLI — TOML-hook-based thrift patterns (~/.codex/config.toml [hooks] block), OpenAI rate table, session-priming variant of Phase 4. Port of harness-thrift.

Theme B for Gemini CLI — cost-conscious long-session optimisation: BeforeTool/AfterTool telemetry hooks, Vertex prompt-cache priming with minimum-token gating + storage-time cost, gemini-flash summariser, end-of-session audit

Theme D — exploration-focused codebase mapping. Parallel subagent scan, git-SHA-keyed cache, structural where/deps query slash commands, dependency graph for typed languages.

Theme D — debugging-focused disciplined investigation: reproduce → isolate → hypothesize → verify with structured error parsing, hypothesis state persistence, git/input bisection, and durable debug-log artifacts. Wraps superpowers:systematic-debugging when available.