whit3rabbit

Front door for running the clean-room unattended loop as a Claude Code dynamic WORKFLOW using in-session subagents, so it never spends `claude -p` API tokens. Use when the user wants to run a clean-room / reverse-engineering / source-to-clean-implementation task hands-off but without paying per-token for `clean-room-skill run`, or says "clean-room as a workflow", "unattended clean-room without API cost", "/clean-room-loop". Runs a short discussion (authorization, end goal, target stack, policies, source roots, output roots, iterations) then launches the `clean-room-loop` workflow with the answers. NOT the enforced OS-level wall - that is `clean-room-skill run --agent-runtime claude`.

Use for authorized clean-room, reverse-engineering, source-to-implementation, compatibility rewrite, or migration tasks. Produces clean behavioral specs, implementation plans, clean code changes, verification reports, QC reports, open questions, and test plans without moving source expression into the clean implementation.

Starts the Clean Room startup wizard in bounded unattended controller mode for authorized source-to-implementation work with finite loop limits and safety stops.

Continues an existing Clean Room run from durable artifacts without relying on prior chat history.

Records Clean Room initialization preferences, separated artifact locations, model policy, schema profile, and clean-safe rule defaults before a clean-room run starts or resumes.

Creates or reviews the required Clean Room preflight goal contract before source discovery, decomposition, attended execution, or unattended execution.

Starts the Clean Room startup wizard in attended controller mode for authorized source-to-implementation work with human review pauses at clean-room gates.

Archives or quarantines current Clean Room artifacts and restarts from the scope gate with a fresh task id.

Use when a repository has synrepo context available.

Orchestrate coding agents with review loops, parallel races, repeat passes, and task-list progression. Use when: the user asks to "cook" something, wants iterative refinement, wants to race multiple approaches, or needs to work through a task list.

This skill queries the OpenAI OpenAPI spec (~26k lines, 148 endpoints, 477 schemas) without loading the full file into context. Use when implementing or modifying Anthropic-to-OpenAI translation logic, checking OpenAI API field names/types, verifying schema shapes, or comparing request/response structures. Triggers on questions about OpenAI API structure, field types, endpoint parameters, or schema definitions.

Zig 0.16.0 async I/O programming with the new std.Io interface. Covers async/concurrent primitives, Future handling, cancellation patterns, and I/O implementation strategies. Use when working with Zig 0.16.0+ async code, event loops, concurrent tasks, or non-blocking I/O.

Provides comprehensive Zig programming language expertise including syntax, standard library, build system, memory management, error handling, and C interoperability. Use this skill when working with Zig code, learning Zig concepts, debugging compilation errors, or building Zig applications across multiple versions (0.2.0 through master).

Use synrepo in repositories with a .synrepo/ directory. Prefer synrepo cards, compact search, and bounded task contexts before reading source files cold.

Use synrepo in repositories with a .synrepo/ directory. Prefer synrepo cards, compact search, and bounded task contexts before reading source files cold.

Use when the user asks to demonstrate a minimal Agent Skill end-to-end.

Use synrepo in repositories with a .synrepo/ directory. Prefer synrepo cards and search before reading source files cold.

Browser automation with Camoufox MCP. Use when an agent needs to browse a URL, extract page text or structure, fill or submit forms, click through a page, screenshot, run diagnostics, drive a multi-step interactive browser session, or tune privacy and anti-detection options through the camoufox-mcp-server MCP server. Also use when a fetch or HTTP request gets blocked, returns a bot wall, or needs a real browser fingerprint.