RLM (Recursive Language Model) — large-context analysis via REPL pattern. Based on MIT's RLM paper (arXiv:2512.24601). Automatically detects large files and suggests the optimal RLM pattern.
★not rated 24 6mo agoA
tokens not measured
originalMIT
Model-selection economist agent for Claude Code: decides which Claude model (and effort/token budget) each task in a flow should run on by weighing effectiveness need against efficiency pressure and difficulty, bookended around a build as a right-sizing blueprint and a closing usage report. Ships with companion skills.
★not rated 24▲
+1
changed todayA
tokens not measured
A markdown brain that gives your AI coding tools judgment and rigor: sharp defaults, real decisions, and the discipline to verify before saying done. For Claude Code, Cursor and Codex.
★not rated 24 11d agoA
tokens not measured
originalMIT
EN: Reflect on the current conversation and save reusable learnings (memories, anti-patterns / Reflexion-style lessons, skill stubs, project-doc updates) with explicit confirmation. ES: Reflexiona sobre la conversación actual y guarda aprendizajes reusables (memorias, anti-patrones estilo Reflexion, stubs de skills…
★not rated 23 1mo agoA
tokens not measured
originalMIT
Explains unfamiliar operations before Claude Code executes them. Claude generates inline explanations for shell commands and file edits, no external LLM needed.
Camada compilada de contexto para qualquer AI coding tool. Compila projetos em estrutura .first-plan/ com 15 camadas (discovery, conventions, reuse, features, quality, contracts, evolution, runtime) e gera arquivos de instrução tool-specific (AGENTS.md para Codex, .cursorrules para Cursor, .github/copilot-instructions.
★not rated 23 16d agoA
tokens not measured
originalMIT
Graph, categorize and search your installed Claude Code skill collection: what is loadable, what is disabled, what points at things that no longer exist — plus a zero-token searchable tier for dormant skills.
★not rated 23 24d agoA
tokens not measured
originalApache-2.0
Detects correction or long-term note-taking scenarios, runs lightweight scholar preflight on fresh tasks, keeps the agent in an explicit closed loop until the user confirms completion, escalates into Ascended Mode after repeated failed corrections or manual trigger, archives detailed mistake / note entries into pro.
★not rated 23 4mo agoA
tokens not measured
originalMIT