Honeycomb observability skills for AI coding agents. Adds query patterns, production investigation workflows, SLOs & triggers, OpenTelemetry instrumentation, and Beeline migration guidance. Designed to complement the Honeycomb MCP server.
honeycomb-investigator for autonomous multi-step production debugging, instrumentation-advisor for codebase-to-Honeycomb gap analysis (Claude Code and Cursor)/honeycomb-setup for interactive MCP server configuration (Claude Code and Cursor)Full Plugin Install:
| Tool | Install |
|---|---|
| Claude Code | claude plugin marketplace add honeycombio/agent-skill then claude plugin install honeycomb |
| OpenAI Codex | codex plugin marketplace add honeycombio/agent-skill then install Honeycomb from the plugin directory |
| Cursor | Install from cursor.directory — or Team Marketplace import (setup) |
| Augment (Auggie CLI) | auggie plugin marketplace add honeycombio/agent-skill then auggie plugin install honeycomb |
| GitHub Copilot CLI | copilot plugin install honeycombio/agent-skill:honeycomb |
Skills + MCP (manual setup):
| Tool | Skills Directory | MCP Config |
|---|---|---|
| VS Code Copilot | .github/skills/ | .vscode/mcp.json |
| OpenAI Codex CLI | honeycomb/skills/ | ~/.codex/config.toml |
| Cline | Rules system | cline_mcp_settings.json |
MCP Server Only: Windsurf, Amazon Q Developer, Continue, and Copilot Coding Agent can connect the Honeycomb MCP server directly. See Honeycomb Docs: MCP Configuration for setup instructions.
For detailed setup instructions for each tool, see Honeycomb Docs: Agent Skills Setup.
claude plugin marketplace add honeycombio/agent-skill
claude plugin install honeycomb
claude plugin add ./honeycomb
Requires the Honeycomb MCP server to be configured (the /honeycomb-setup command can help).
Honeycomb is a Cursor plugin (skills, agents, the /honeycomb-setup
command, hooks, and the MCP server). Install it from the Cursor plugin
directory:
→ cursor.directory/plugins/honeycomb
Open the listing and use its Install action. The plugin's MCP server config is shown for review before it's passed to Cursor — confirm it, then reload.
On Teams/Enterprise plans you can import this repo as a private Team
Marketplace: Cursor dashboard → Settings > Plugins > Team Marketplaces >
Import, then paste https://github.com/honeycombio/agent-skill and grant team
access. See Cursor: Plugins for team
marketplace setup.
For a manual / local install (e.g. plugin development), see
honeycomb/.cursor-plugin/INSTALL.md.
codex plugin marketplace add honeycombio/agent-skill
After adding the marketplace, restart Codex, open the plugin directory, select Honeycomb Plugins, and install Honeycomb.
auggie plugin marketplace add honeycombio/agent-skill
auggie plugin install honeycomb
copilot plugin install honeycombio/agent-skill:honeycomb
python -m venv .venv && source .venv/bin/activate
make install
make test
Structural tests validate plugin layout, frontmatter, and hook behavior — no API keys needed.
This plugin uses semantic versioning. Tags follow the format v{major}.{minor}.{patch}.
Marketplace users can pin to a specific version via git ref.
Python
83.9%
Shell
12.2%
JavaScript
3.1%
Honeycomb observability skills for AI coding agents. Adds query patterns, production investigation workflows, SLOs & triggers, OpenTelemetry instrumentation, and Beeline migration guidance. Designed to complement the Honeycomb MCP server.
honeycomb-investigator for autonomous multi-step production debugging, instrumentation-advisor for codebase-to-Honeycomb gap analysis (Claude Code and Cursor)/honeycomb-setup for interactive MCP server configuration (Claude Code and Cursor)Full Plugin Install:
| Tool | Install |
|---|---|
| Claude Code | claude plugin marketplace add honeycombio/agent-skill then claude plugin install honeycomb |
| OpenAI Codex | codex plugin marketplace add honeycombio/agent-skill then install Honeycomb from the plugin directory |
| Cursor | Install from cursor.directory — or Team Marketplace import (setup) |
| Augment (Auggie CLI) | auggie plugin marketplace add honeycombio/agent-skill then auggie plugin install honeycomb |
| GitHub Copilot CLI | copilot plugin install honeycombio/agent-skill:honeycomb |
Skills + MCP (manual setup):
| Tool | Skills Directory | MCP Config |
|---|---|---|
| VS Code Copilot | .github/skills/ | .vscode/mcp.json |
| OpenAI Codex CLI | honeycomb/skills/ | ~/.codex/config.toml |
| Cline | Rules system | cline_mcp_settings.json |
MCP Server Only: Windsurf, Amazon Q Developer, Continue, and Copilot Coding Agent can connect the Honeycomb MCP server directly. See Honeycomb Docs: MCP Configuration for setup instructions.
For detailed setup instructions for each tool, see Honeycomb Docs: Agent Skills Setup.
claude plugin marketplace add honeycombio/agent-skill
claude plugin install honeycomb
claude plugin add ./honeycomb
Requires the Honeycomb MCP server to be configured (the /honeycomb-setup command can help).
Honeycomb is a Cursor plugin (skills, agents, the /honeycomb-setup
command, hooks, and the MCP server). Install it from the Cursor plugin
directory:
→ cursor.directory/plugins/honeycomb
Open the listing and use its Install action. The plugin's MCP server config is shown for review before it's passed to Cursor — confirm it, then reload.
On Teams/Enterprise plans you can import this repo as a private Team
Marketplace: Cursor dashboard → Settings > Plugins > Team Marketplaces >
Import, then paste https://github.com/honeycombio/agent-skill and grant team
access. See Cursor: Plugins for team
marketplace setup.
For a manual / local install (e.g. plugin development), see
honeycomb/.cursor-plugin/INSTALL.md.
codex plugin marketplace add honeycombio/agent-skill
After adding the marketplace, restart Codex, open the plugin directory, select Honeycomb Plugins, and install Honeycomb.
auggie plugin marketplace add honeycombio/agent-skill
auggie plugin install honeycomb
copilot plugin install honeycombio/agent-skill:honeycomb
python -m venv .venv && source .venv/bin/activate
make install
make test
Structural tests validate plugin layout, frontmatter, and hook behavior — no API keys needed.
This plugin uses semantic versioning. Tags follow the format v{major}.{minor}.{patch}.
Marketplace users can pin to a specific version via git ref.
Python
83.9%
Shell
12.2%
JavaScript
3.1%