Guidance for Claude Code (and any other agent) working in this repository. Pooprusteek is a Rust TUI coding agent that talks to DeepSeek's reverse-engineered web API (no paid API key — cookie/token auth, local SHA-3 proof-of-work) instead of an official LLM API, with optional extra providers (OpenAI-compatible / Anthropic Messages / Gemini endpoints via /providers). Built on tokio + ratatui, it runs a single event-driven terminal UI supporting parallel conversations, sub-agents, an iterative GOAL worker/evaluator loop, MCP (stdio/HTTP/SSE tool servers…
Recipe for adding a new in-TUI slash command (/foo) to PoopRusteek — the Command trait, one-file-per-command layout, registration, and CommandResult routing. Use when adding or editing a slash command; not for agent tools (see pooprusteek-add-tool).
Recipe for adding a new agent-facing Tool (bash/skill/tool_search-style) to PoopRusteek — the Tool trait, the single registration line, and the hard tools-never-touch-app boundary. Use when adding an agent tool; not for slash commands (see pooprusteek-add-slash-command).
Read-me-first onboarding for any agent working ON the PoopRusteek codebase — what it is, the mandatory .memories read order, the core architecture, build/test commands, and all 11 contributor invariants. Use when starting cold; skip if you already know the layout.
How PoopRusteek's LLM-provider and MCP subsystems are structured, for anyone editing src/provider/ or src/mcp/ — the LLMProvider trait + fork(), the four protocols, MCP transports/discovery/OAuth, and the lock-across-await invariant. Use before touching provider or MCP code.
How to safely work in PoopRusteek's local RAG layer (src/semantic/) — the embedder + three corpora, hybrid RRF ranking, background init, deferred MCP schemas, the degrade-never-block invariant, and the MRR eval gate you must run and journal. Use before touching semantic ranking or corpus text.
Design, bootstrap, audit, and maintain a durable `.memories/` knowledge base so any agent can regain full project context across sessions. Use this whenever the user wants to set up, structure, clean up, convert, or improve an agent knowledge base / project memory — including phrases like ".memories", "memory bank", "agent context files", "AGENTS.md/CLAUDE.md structure", "onboarding docs for agents", "capture project knowledge", or "what goes in which memory file" — even if they don't say ".memories" verbatim. Applies to both new repos (bootstrapping from scratch) and existing repos (fixing a stale, scattered, or ad-hoc KB). Project-agnostic; the worked exemplar is this repo's own `.memories/`.
Cheat sheet of PoopRusteek's slash commands grouped by theme. Use when you need the exact syntax or purpose of an in-app command (/new, /goal, /mcp, /rag, /providers, etc.). Not for editing PoopRusteek's Rust source.
Install, authenticate, and run the PoopRusteek TUI coding agent. Use when setting up PoopRusteek, choosing a run mode, understanding DeepSeek auth, or locating config/session files. Not for editing PoopRusteek's Rust source.
Configure extra LLM providers (OpenAI-compat / Anthropic / Gemini) and MCP tool servers in PoopRusteek. Use when adding a provider endpoint, switching models, or wiring up MCP servers and OAuth. Not for editing PoopRusteek's Rust source.
Author/install markdown skills and drive PoopRusteek's local semantic (RAG) layer. Use when writing a SKILL.md, enabling/disabling skills, or controlling per-turn hints, tool_search, deferred schemas, and history search. Not for editing PoopRusteek's Rust source.
Master REST and GraphQL API design principles to build intuitive, scalable, and maintainable APIs that delight developers. Use when designing new APIs, reviewing API specifications, or establishing API design standards.
Master software architect specializing in modern architecture patterns, clean architecture, microservices, event-driven systems, and DDD. Reviews system designs and code changes for architectural integrity, scalability, and maintainability. Use PROACTIVELY for architectural decisions.
Implement proven backend architecture patterns including Clean Architecture, Hexagonal Architecture, and Domain-Driven Design. Use when architecting complex backend systems or refactoring existing applications for better maintainability.
This skill should be used when the user asks to "design ArkUI API", "add component property", "create Modifier method", "review ArkUI API", "deprecate API", "write JSDOC for ArkUI", or mentions OpenHarmony API design standards. Provides comprehensive guidance for ArkUI component API design following OpenHarmony coding guidelines, including static/dynamic interface synchronization, SDK compilation, and verification.
Master defensive Bash programming techniques for production-grade scripts. Use when writing robust shell scripts, CI/CD pipelines, or system utilities requiring fault tolerance and safety.
Accessibility guide for Capacitor apps covering screen readers, semantic HTML, focus management, and WCAG compliance. Use this skill when users need to make their app accessible.
Guides the agent through an Apple App Store preflight review for Capacitor apps before submission or after rejection. Covers guideline checklist selection, App Store metadata review, Capacitor and iOS project inspection, privacy manifests, Sign in with Apple, entitlements, and common rejection patterns. Do not use for Google Play review, generic store publishing only, or non-Apple mobile runtimes.
Complete guide to publishing Capacitor apps to Apple App Store and Google Play Store. Covers app preparation, screenshots, metadata, review guidelines, and submission process. Use this skill when users are ready to publish their app.
Plain text files in a repository that tell a coding agent how the project works: commands to run, conventions to follow and things to avoid. CLAUDE.md, AGENTS.md, cursor rules and skills are the common kinds.
CLAUDE.md or AGENTS.md?
CLAUDE.md is read by Claude Code. AGENTS.md is an open format that Codex, Cursor and other agents read. Many projects keep one and point the other at it.
What is a skill?
A folder with a SKILL.md that describes one capability, such as filling PDFs or reviewing code. The agent loads it only when the task calls for it.
Can I search my own team's files too?
Your agents already can, over MCP, limited to the files you're allowed to read. Searching them from this page is coming.
Aver005CLAUDE.md