agentleFS
Sign inStart free
reefbarmanreefbarman/agentlink/.agents/skills/meridian-integration/SKILL.md

meridian-integration

Skill2 starsChanged 19 days agoMITView on GitHub

Maintain AgentLink's Meridian companion plugin when changing tools, provider transport, prompts, session identity, or Meridian integration; diagnose delayed tool handoffs through Meridian.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Before tuning tool disclosure, run the repository's tool, context and session telemetry reports as required by CLAUDE.md.

    line 30
  2. Distinctive:

    Use synthetic fixture files and avoid logging credentials or real prompt text.

    line 36
  3. Distinctive:

    Update plugin/package docs, this skill, and the bundled provider guide together when the contract changes.

    line 38

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

AgentLink owns tools and approvals; Meridian bridges requests through the Claude Agent SDK. Read integrations/meridian-plugin-agentlink/README.md and the plugin's index.js before changing this integration. The runtime user's setup guide is resources/builtin-skills/documentation/references/providers.md. For this machine's endpoint and installed-plugin workflow, read references/local-setup.md relative to this skill. Reinspect the running service before assuming that snapshot is still current.

The plugin loads the supplied catalog eagerly instead of copying tool names. Adding or renaming a tool normally needs no Meridian change. Check these seams:

  • The foreground/background system prompt must retain the leading You are AgentLink, identity used by the plugin. Update detection/tests if it changes.
  • meridianSessionAffinity must keep sending the live conversation ID as x-session-affinity. Never replace it with a static header shared by chats.
  • The OpenAI translator must preserve each tool's name, schema, arguments and matched result ID. New deferral flags need explicit compatibility review: deferred tools can lift Meridian's one-turn cap and restore hidden generation.
  • Native tools and MCP tools execute in AgentLink, not in Meridian. Preserve approvals, parallel call envelopes, real result continuation and final set_task_status delivery. Keep the plugin scoped to AgentLink requests.
  • Meridian pins deferral per session. Evaluate changed policy in fresh chats; do not attribute an old session's retained policy to plugin failure.

Before tuning tool disclosure, run the repository's tool, context and session telemetry reports as required by CLAUDE.md. After changes, run the plugin tests, the live probe.mjs command in the plugin README, and affected AgentLink provider tests. For behavior changes, use the implicated model and actual AgentLink OpenAI completion facade against a real Meridian/SDK instance; mocks cannot establish latency savings. Record SDK calls, hidden output, tool completion, resume, and cache behavior. Use synthetic fixture files and avoid logging credentials or real prompt text.

Update plugin/package docs, this skill, and the bundled provider guide together when the contract changes. Do not claim npm publication or activate/restart other users' services merely because code changed. Honor the user's requested scope.

The local installer copies the plugin, so source edits do not automatically reach the running service. When a requested local integration update requires a plugin change, reinstall the copy and reload plugins, then verify active status and a fresh conversation. Keep other plugins and clients' policies intact.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free