23,553 tagged Building agents, measured the same way as everything else here.
Browse within: agentic-ai 234LLM 216agent 200agents 151skills 144agentic-workflow 111agent-orchestration 87openai 86antigravity 79agentic-framework 73cli 73ai-skills 70Autonomous Agents 67agent-framework 63
Skill Claude CodeCodex
Join the Cortex Agent Center to collaborate with other agents. Use when a task needs "find an agent that knows a domain better", "delegate a subtask", or "let others discover and call you". Covers self-onboarding (device authorization, one user confirmation), declaring capabilities, discovering agents, and delegation…
Skill Claude CodeCodex
Use the Claude Codex leaf MCP for consent-gated Anthropic chat, model listing, and provider status.
Skill Claude CodeCodex
Use the Grok Codex leaf MCP for consent-gated xAI chat, model listing, and provider status.
Skill Codex
Plan and orchestrate coding, review, testing, or documentation work through the DeepSeek Worker MCP with Flash as the mandatory default model, then have the primary GPT/Sol model inspect exact result revisions, validate them, and safely apply accepted changes locally. Use when the user asks to use, delegate to…
Skill Claude CodeCodex
Use when the user explicitly requests Claude, Sonnet, Fable, or Opus, or when an independent Claude perspective would materially improve a coding task, review, or diagnosis.
jean-baptiste-tiple/mcp-template
Skill Claude Code
MCP : tools serveur MCP, widgets MCP Apps, resources ui://, OAuth 2.1 connecteur, mcp-handler, inspector, pattern prepare/save (zéro IA serveur), éditions en deltas (ops), golden queries, AX. FR : tool MCP, widget dans Claude/ChatGPT, serveur MCP, canal MCP, bridge host, anonymisation/adaptation par le modèle…
AgRenting/agrenting-codex-plugin
Skill Claude CodeCodex
Discover and hire one or more remote Agrenting agents, then track the independent hirings and collect their results. Use when the user asks Codex to delegate work, find remote agents, hire agents, or outsource tasks through Agrenting.
Skill Claude CodeCodex
REST API server and MCP protocol integration.
kilroyblockchain/free2pa-devtool
Skill Codex
Add Free2PA provenance and local trust checks to an agentic application. Use when a developer asks to protect an agent, OpenClaw project, ChatGPT app, MCP server, Agents SDK workflow, skills, prompts, SOUL.md, AGENTS.md, policies, or other control files from unnoticed changes or outside-group publishers.
Skill Claude CodeCodex
This guide teaches an LLM how to use the Kitchen MCP tools effectively.
234194027-cpu/aion-memory-nexus
Skill Codex
Set up an external AI agent to connect to the Life Memory System through MCP, including verified installation, retrieval, RawEvent writes, sync, diagnostics, and Working-Agent governance. Use when an agent needs to install or verify the Life Memory MCP bridge, onboard Codex/Claude/Cursor/Windsurf style clients, or…
Skill Claude CodeCodex
Connect cloud agents to Cloud2Local (local MCP). Use when configuring remote MCP URL/headers/token, cloud2local access, auto public expose status, or troubleshooting cloud agent connectivity to a local MCP server.
Skill Claude CodeCodex
Interview yourself into a thinqee persona knowledge base. Use when the user wants to build or extend their own persona (voice, values, methods, positions) for the thinqee engine, get "grilled" about how they work, or turn a declined question into a signed note. Runs the installed thinqee CLI's interview; it never…
Skill Claude CodeCodex
Chat and coordinate with other live Claude Code sessions on this machine via the interclaude tools. Use when asked to "message the other session", "ask the other Claude", "send this to the other worktree", "broadcast to my sessions", "who's online", "coordinate the agents", or whenever a message arrives and needs a…
Skill Claude CodeCodex
A skill that converts structured knowledge lists and proposed capabilities into importable TFOnto files. TFOnto is a file format for describing a robot's concepts, data fields, relationships, and capabilities.
Skill Claude CodeCodex
Open, catch up, send to, or manage a Huddora room safely with plugin seat identity, xd://huddoramessagesend, persona get/set, and plugin-owned roomwatch delivery.
Skill Claude CodeCodex
Mandatory lookup skill — read the reference file for any MCP tool before calling it to get correct arguments, usage patterns, and output handling.
Skill Claude CodeCodex
A runtime guide for using outside workers alongside Codex, while keeping Codex responsible for execution, approvals, and final decisions.
sachin0610-srm/contract-sentinel
Skill Claude CodeCodex
Guidelines and patterns for defining Tools, Resources, and Prompts in a NitroStack application with schema validation via Zod, including caching, rate-limiting, and base64 file uploads.
zetsubo-dev/codex-claude-subagent
Skill Claude Code
Use when the user asks Codex to involve Claude as a subagent, Claude Code, Claude reviewers, second opinions, or interactive Claude Code sessions from Codex. Provides guidance for the local codexclaudesubagent MCP tools.
princeruhulofficial/mcp-tool-chain-optimizer
Skill Claude CodeCodex
Analyze multi-step AI agent tool chains for reliability, bottlenecks, cost and better ordering. Use when the user wants to measure or improve the success probability of a sequence of tools.
Skill Codex
Use for non-trivial tasks that may need shared VPS Skills, shared MCP tools, prior context from Obsidian/MemOS, or durable Work state through CodexPro Bridge. Skip ordinary chat and short calculations.
Skill Claude CodeCodex
Run many subagents over one list of work without them colliding or each inventing its own idea of the task. Use whenever the user asks to process, extract, audit, translate, review, or convert a set of files, pages, records, tickets or rows and it would go faster with several agents at once. Also use when they say "in…
Skill Codex
Use CatHub Nyaanes for session check-in and explicit end-of-work handoff. Use when beginning or resuming work, when the user asks for unread updates or current status, or when the user explicitly declares that the current work session is ending. Never infer work completion from silence, a finished job, a final answer…
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: