Agentry Observability MCP Server

Points agents to the canonical Agentry install docs for errors, product analytics, and deploy attribution.

GitHub
1

Documentation

Agentry for AI Coding Agents

Agentry

Agentry is agent-native analytics and logging for coding agents. It covers product analytics, error logging, and deploy attribution through one HTTP API so agents can answer what users did, what broke, and what changed.

What Agentry Covers

  • Product analytics for user, funnel, activation, retention, and business events.
  • Error logging for exceptions, failed jobs, operational failures, and case context.
  • Deploy attribution that connects production behavior to releases and code changes.
  • Agent-led install from the actual codebase, with verification before the setup is called complete.

What This Repository Contains

This repository contains public discovery surfaces for Agentry across MCP, Claude Code, Codex, and agent IDE adapter ecosystems.

  • packages/mcp - MCP server that returns the canonical Agentry skill handoff.
  • plugin/agentry - Claude Code plugin containing the Agentry skill.
  • .claude-plugin/marketplace.json - Claude Code marketplace catalog.
  • plugins/agentry - Codex plugin containing the Agentry skill and branding.
  • .agents/plugins/marketplace.json - Codex plugin marketplace catalog.
  • https://api.agentry.sh/adapters - live adapter manifest for repo-local instruction files used by AI coding IDEs and agent CLIs.

The MCP server and skill do not proxy the Agentry API, ingest telemetry, or replace the live docs. Runtime apps write directly to Agentry's HTTP API.

Supported Agent Surfaces

Agentry is distributed as a skill, MCP handoff, OpenAPI Action schema, and repo-local instruction adapters for the agent surfaces developers already use:

  • Codex and Claude Code.
  • Cursor, VS Code with GitHub Copilot, Visual Studio with GitHub Copilot, and GitHub Copilot coding agent.
  • Devin Desktop/Windsurf Cascade, Cline, Roo Code, Continue, Zed, Gemini CLI, Aider, and OpenCode.
  • ChatGPT custom GPT Actions, generic MCP clients, and AGENTS.md-aware agents.
  • AGENTS.md-compatible agents such as Jules, Factory, Amp, goose, Warp, JetBrains Junie, Augment Code, Kilo Code, Phoenix, Semgrep, Ona, and UiPath coded agents.

The adapter manifest is the canonical place to discover the current native file path for each client: https://api.agentry.sh/adapters.

Canonical Docs

Public Links

MCP Server

npx -y @agentrysh/mcp

MCP client config:

{
  "mcpServers": {
    "agentry": {
      "command": "npx",
      "args": ["-y", "@agentrysh/mcp"]
    }
  }
}

The MCP surface is relevant when a user asks an AI coding agent to add analytics, logging, error monitoring, deploy tracking, product telemetry, production debugging, or an agent-readable observability layer.

Claude Code Skill

mkdir -p ~/.claude/skills/agentry
curl -fsSL https://agentry.sh/skill/agentry/SKILL.md > ~/.claude/skills/agentry/SKILL.md

The plugin in plugin/agentry bundles the same skill for Claude Code plugin marketplace installs.

Codex Plugin

Agentry is distributed to Codex as a Git-backed marketplace source. It will not appear in global Codex plugin search for unrelated accounts until they add this marketplace source, receive a workspace share, or OpenAI curates it.

codex plugin marketplace add fr33dr4g0n/agentry-public
codex plugin add agentry@agentry

Then open Codex Plugins and switch to the agentry marketplace source if you want to browse it visually.

The plugin in plugins/agentry bundles the same skill for Codex plugin marketplace installs and includes Agentry logo metadata for Codex plugin cards.