Skills, prompts, AGENTS.md files and MCP servers that Codex reads. Skills are shared with Claude Code; the rest is Codex-native. 274,429 indexed, measured the same four ways as everything else here.
Use when coordinating with other AI agents via backchannel (bch) — checking the inbox at task start, sending DMs or channel messages to peer agents, requesting reviews, or reacting to incoming messages. Trigger phrases include "message the other agent", "check your inbox", "tell codex/claude/gemini to…", or any task…
Onboard a project into this governance layer: scan a repo that was pasted or cloned into Projects/, draft its bible from what the code actually shows, create its tracker and workspace pointer, register it in the composition manifest, and echo the reconstructed setup back for correction. Use this whenever the user…
MCP server "lightrag" as configured in Ricardo-Kaminski/local-rag. Launched with C:\Users\ricar\miniconda3\envs\local-rag\python.exe C:\workspace\pessoal\local-r.
★not rated 2 6mo agoA
tokens not measured
originalMIT
Capture a failed or blocked task before it disappears. Categorizes the failure mode, extracts what was learned, and generates either a retry packet with the root cause fixed or an escalation message with specific decision points. Use whenever an agent returns a blocker, error, or partial result.
MCP server "sqlite-analytics" as configured in AleBrito124356/mcp-server-cookbook. Runs C:\dev\mcp-server-cookbook\servers\sqlite_analytics\server.py with python. Needs 1 environment variable to run.
★not rated 2 2mo agoA
tokens not measured
originalMIT
R-Code can be available as a local MCP server named r-code. Use it only when the user asks you to delegate a focused investigation to R-Code's own Agent.
Use TypeCast to turn LLM output into validated, typed JVM values. Reach for it whenever code calls an LLM and then parses/validates the response by hand: extraction of case classes or Java records from text, classification/routing via sealed traits, constraint enforcement with retry, streaming partial JSON into a UI…
Delegate an explicitly approved local coding task from Codex to Kimi Code, monitor it, return review feedback to the same Kimi session, and independently verify the result. Use only when the user explicitly asks to try Kimi or has approved a Codex suggestion to use Kimi for the current task.
Local Codex-to-Figma workflow for inspecting, planning, creating, and refining the currently open Figma file. Use for natural-language requests about the current Figma canvas or when the user explicitly invokes $layntra.
Use when you need DOM-level precision (exact text/aria/placeholder matching → screen coordinates) while operating the host's real Chrome with human identity and zero automation traces. humandom adds a read-only DOM locator alongside humanbrowser; all actions remain OS-level (tap/typetext/presskey). When the DOM…
Browser automation via MCP using your real Chrome browser. Use when the user asks to navigate websites, scrape data, fill forms, take screenshots, or automate browser tasks. Unlike headless browsers, helm controls the user's actual Chrome via a lightweight extension.
Build and maintain Knowledge Graphs from webpages using WordLift APIs. Use when importing pages from sitemaps via WordLift Sitemap Import API, creating product catalogs with GS1 Digital Link identifiers (GTIN-based), generating slug-based entity IDs for organizations/people/webpages, creating JSON-LD markup…
Use when designing UI with OpenPencil — creating layouts via op CLI, batch design DSL, or MCP tools. Covers PenNode schema, semantic roles, typography, color, spacing, and common component patterns.
Find invoices that are past their due date, work out who owes what and for how long, and draft a chaser for each one. Use when asked about overdue payments, debtors, aged receivables, or "who hasn't paid us".
Run one bounded trajectory reflection when repeated same-cause failures, contradicted assumptions, real-use failure after local green, material scope drift, a materially unresolved phase boundary, an unapproved authority-sensitive next effect, or user doubt suggests the current path may be wrong. Use Context Canvas…
Adds Arize AX tracing to an LLM application for the first time. Follows a two-phase agent-assisted flow to analyze the codebase then implement instrumentation after user confirmation. Use when the user wants to instrument their app, add tracing from scratch, set up LLM observability, integrate OpenTelemetry or…
Responsive HTML email template generation using MJML 4.x framework. Use when the user asks to create, generate, design, or build an email template — including welcome emails, promotional blasts, transactional templates, newsletters, or any responsive email. Also use when the user asks to compile MJML to HTML, work…
MCP server "reviewed-server" as configured in Renga154/agentrisk. Runs ./vendor/reviewed-server/index.js with node. Needs 1 environment variable to run.
MCP server "yahoo-finance" as configured in haiiibin/claude-multi-agent-investing. Runs locally from the /ABSOLUTE/PATH/TO/yahoo-finance-mcp Python package.
★not rated 2 22d agoA
tokens not measured
originalMIT
Guides learners to discover knowledge through strategic Socratic questioning and progressive scaffolding removal. Combines question ladders, misconception detectors, Feynman explanations, and worked-example fading to build durable understanding. Use when teaching complex concepts, correcting misconceptions, onboarding…
Use when an AI agent needs a repeatable workflow to get a real answer or compare providers through agentic-mcp, with discovery, proof, and CLI fallback steps.
★not rated 2 4mo agoA40 tokens
originalMIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: