Complete AI Training
Start for Free

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

Skill · DevOps

Manifest

Install and configure the Manifest observability plugin for your agents. Use when setting up telemetry, configuring API keys, or troubleshooting the plugin.

Agentic Awesome SkillsAdded Sep 5, 2026
Use it in my AI

How to use it

  1. Start your plan and connect your AI once
  2. Add the skill to your AI, one time only: press Copy skill and paste it into your AI's instructions: a project in ChatGPT, Claude or Grok, a Gem in Gemini, an agent in Microsoft 365 Copilot, or a rule in Cursor. It stays there for every new chat. In Kimi, paste it at the start of a new chat.
  3. Ask for the task in your own words, or say it directly:
Use the Manifest skill to help me with this.
SKILL.md

Manifest Setup

Follow these steps in order. Do not skip ahead.

Use this skill when

  • User wants to set up observability or telemetry for their agent
  • User wants to connect their agent to Manifest for monitoring
  • User needs to configure a Manifest API key or custom endpoint
  • User is troubleshooting Manifest plugin connection issues
  • User wants to verify the Manifest plugin is running

Do not use this skill when

  • User needs general observability design (use observability-engineer instead)
  • User wants to build custom dashboards or alerting rules
  • User is not using the Manifest platform

Instructions

Step 1 — Stop the gateway

Stop the gateway first to avoid hot-reload issues during configuration.

claude gateway stop

Step 2 — Install the plugin

claude plugins install manifest

If it fails, check that the CLI is installed and available in the PATH.

Step 3 — Get an API key

Ask the user:

To connect your agent, you need a Manifest API key. Here's how to get one:
1. Go to https://app.manifest.build and create an account (or sign in)
2. Once logged in, click "Connect Agent" to create a new agent
3. Copy the API key that starts with mnfst_
4. Paste it here

Wait for a key starting with mnfst_. If the key doesn't match, tell the user the format looks incorrect and ask them to try again.

Step 4 — Configure the plugin

claude config set plugins.entries.manifest.config.apiKey "USER_API_KEY"

Replace USER_API_KEY with the actual key the user provided.

Ask the user if they have a custom endpoint. If not, the default (https://app.manifest.build/api/v1/otlp) is used automatically. If they do:

claude config set plugins.entries.manifest.config.endpoint "USER_ENDPOINT"

Step 5 — Start the gateway

claude gateway install

Step 6 — Verify

Wait 3 seconds for the gateway to fully start, then check the logs:

grep "manifest" ~/.claude/logs/gateway.log | tail -5

Look for:

[manifest] Observability pipeline active

If it appears, tell the user setup is complete. If not, check the error messages and troubleshoot.

Safety

  • Never log or echo the API key in plain text after configuration
  • Verify the key format (mnfst_ prefix) before writing to config

Troubleshooting

ErrorFix
Missing apiKeyRe-run step 4
Invalid apiKey formatThe key must start with mnfst_
Connection refusedThe endpoint is unreachable. Check the URL or ask if they self-host
Duplicate OTel registrationDisable the conflicting built-in plugin: claude plugins disable diagnostics-otel

Examples

Example 1: Basic setup

Use @manifest to set up observability for my agent.

Example 2: Custom endpoint

Use @manifest to connect my agent to my self-hosted Manifest instance at https://manifest.internal.company.com/api/v1/otlp

Best Practices

  • Always stop the gateway before making configuration changes
  • The default endpoint works for most users — only change it if self-hosting
  • API keys always start with mnfst_ — any other format is invalid
  • Check gateway logs first when debugging any plugin issue

Limitations

  • Use this skill only when the task clearly matches the scope described above.
  • Do not treat the output as a substitute for environment-specific validation, testing, or expert review.
  • Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.