agentleFS
moasqmoasq/production-saas-starter/.claude/skills/service-connections/SKILL.md

service-connections

Skill532 starsChanged 52 days agoMITView on GitHub

This skill applies when enabling an optional provider MCP, completing host authorization, verifying a service connection, or diagnosing Resend, PostHog, Linear or component-provider access in this starter.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 8 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    Read AGENTS.md, .agents/tools.json and docs/AI_TOOLS.md.

    line 10
  2. Distinctive:

    Keep developer access separate from the application's auth, SMTP and Polar setup.

    line 12
  3. Very specific:

    Run node scripts/harness.mjs tools to inspect repository selections.

    line 18
  4. Distinctive:

    Keep credentials in the host's supported secret storage or private environment.

    line 26
  5. Distinctive:

    Never put tokens in repository files, prompts, generated adapters or evidence.

    line 27
  6. Distinctive:

    After authorization, perform a narrow read appropriate to the task and confirm the intended workspace.

    line 28
  7. Very specific:

    Use node scripts/mcp-probe.mjs <id> for explicit local MCP initialization and tool-list checks.

    line 35
  8. Distinctive:

    Use public documentation queries without private source, customer records, credentials or cookies.

    line 41
  9. Distinctive:

    Treat resource creation, email sending, publication, billing and production configuration changes as separate actions requiring the user's …

    line 46
  10. Distinctive:

    Use existing authorization when it covers the concrete action.

    line 48
  11. Distinctive:

    Keep provider administration in the main authorized session.

    line 52

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
<!-- Generated from .agents/skills/service-connections/SKILL.md; do not edit. -->

Service connection workflow

Read AGENTS.md, .agents/tools.json and docs/AI_TOOLS.md. Identify the requested provider, intended account/workspace and operation before changing configuration. Keep developer access separate from the application's auth, SMTP and Polar setup.

Establish the actual state

  1. Inspect callable host tools first. Prefer an already authorized host plugin when it serves the task; do not configure a second connection unnecessarily.
  2. Run node scripts/harness.mjs tools to inspect repository selections. Treat an enabled entry as configured only. Do not report it as connected, authorized or tested from its presence in .mcp.json or a generated host adapter.
  3. For a requested optional selection, edit .agents/tools.json, then run node scripts/harness.mjs sync and node scripts/harness.mjs check. Preserve catalog endpoints, including PostHog's readonly=true query and Linear's /mcp/readonly path. Do not silently broaden their permissions.
  4. Complete any OAuth or account selection in the coding host's interactive flow. Keep credentials in the host's supported secret storage or private environment. Never put tokens in repository files, prompts, generated adapters or evidence.
  5. After authorization, perform a narrow read appropriate to the task and confirm the intended workspace. Mark the connection read-only verified only for that operation. Report denied scopes, wrong accounts or unavailable tools precisely; retain successful earlier work while resolving the specific blocker.

Keep verification bounded

Use node scripts/mcp-probe.mjs <id> for explicit local MCP initialization and tool-list checks; it does not call tools or prove provider authorization. Use node scripts/harness.mjs check-mcp only for Better Auth's public documentation contract. Neither check establishes working application authentication, SMTP delivery, analytics collection or billing.

Use public documentation queries without private source, customer records, credentials or cookies. Read only data needed for the assigned provider task and keep returned sensitive content out of committed reports. Record verification time, provider/workspace, operation and outcome without secrets.

Treat resource creation, email sending, publication, billing and production configuration changes as separate actions requiring the user's task authorization. Use existing authorization when it covers the concrete action; do not ask again merely because this skill was loaded. Configure or verify a connection without creating test provider resources or sending messages as an incidental check.

Keep provider administration in the main authorized session. Reviewer roles use documentation only, and local QA roles do not gain provider access from a test assignment. Adding a developer MCP does not install a runtime SDK, enable analytics, replace SMTP, switch Polar billing or migrate Better Auth.

Return configured, authorized and read-only verified states separately, with the exact remaining user interaction or technical blocker. Do not call a connection complete when only configuration generation or MCP initialization succeeded.

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.