agentleFS
Sign inStart free
aadi-labsaadi-labs/agentmailer-plugins/agentmailer-mcp/SKILL.md

agentmailer-mcp

Skill1 starsChanged 3 months agoMITView on GitHub

Connect to AgentMailer's hosted MCP server, complete human-approved OAuth, or troubleshoot authentication in Claude Code, Codex, Cursor, and other MCP clients.

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 6 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use the universal Streamable HTTP endpoint:

    line 8
  2. Distinctive:

    Prefer the client's native remote-MCP configuration so it can discover OAuth from AgentMailer's protected-resource metadata.

    line 12
  3. Distinctive:

    Every first-time signup requires human approval.

    line 26
  4. Very specific:

    Call auth_me and require a trusted identity with inboxes:create before creating an identity inbox.

    line 28
  5. Very specific:

    Call list_inboxes as a low-risk functional check.

    line 29
  6. Very specific:

    If no existing inbox fits, call create_inbox with a stable idempotency key and an optional lowercase username.

    line 30
  7. Very specific:

    Verify the returned address ends in @agentmailer.ai.

    line 31
  8. Very specific:

    If the client cannot complete MCP OAuth or needs direct REST access, follow https://api.agentmailer.ai/auth.md exactly.

    line 33
  9. Very specific:

    Do not call create_inbox with an unverified credential.

    line 35
  10. Distinctive:

    Read references/troubleshooting.md when connection, OAuth, or permission checks fail.

    line 39
  11. Distinctive:

    Never claim the connection works until an authenticated tool call succeeds.

    line 46
  12. Distinctive:

    Use the human-approved signup example to understand the approval boundary, then the

    line 50

≈ “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

AgentMailer MCP

Use the universal Streamable HTTP endpoint:

https://api.agentmailer.ai/mcp

Prefer the client's native remote-MCP configuration so it can discover OAuth from AgentMailer's protected-resource metadata. Do not ask the user to paste OAuth tokens into prompts, source files, or shell history.

Connect

Choose the shortest path for the active client:

  • Claude Code: run claude mcp add --transport http agentmailer https://api.agentmailer.ai/mcp, then claude mcp login agentmailer.
  • Codex: run codex mcp add agentmailer --url https://api.agentmailer.ai/mcp, then codex mcp login agentmailer.
  • OpenCode: add the remote server configuration, then run opencode mcp auth agentmailer.
  • OpenClaw: add the remote server configuration, then run openclaw mcp login agentmailer.
  • Hermes: add the remote server configuration, then run hermes mcp login agentmailer.
  • Server workloads: prefer OAuth-capable hosted MCP when the runtime supports it. For direct REST access, follow https://api.agentmailer.ai/auth.md exactly.
  • Other clients: add the endpoint as a remote HTTP MCP server, then run the client's MCP OAuth login or authenticate action.

Every first-time signup requires human approval. After authorization:

  1. Call auth_me and require a trusted identity with inboxes:create before creating an identity inbox. Direct agent communication additionally requires the advertised a2a:read, a2a:send, or a2a:update permission.
  2. Call list_inboxes as a low-risk functional check.
  3. If no existing inbox fits, call create_inbox with a stable idempotency key and an optional lowercase username.
  4. Verify the returned address ends in @agentmailer.ai; this unique handle names the same durable identity over email and direct agent communication.

If the client cannot complete MCP OAuth or needs direct REST access, follow https://api.agentmailer.ai/auth.md exactly. Do not invent or partially reproduce its signup and token procedure.

Do not call create_inbox with an unverified credential. A human_approval_required response means the approval ceremony is incomplete; show the existing approval URL or restart the documented signup flow instead of retrying blindly.

Diagnose

Read references/troubleshooting.md when connection, OAuth, or permission checks fail. Preserve the distinction between:

  • transport failure;
  • OAuth discovery or callback failure;
  • authenticated but insufficient permissions;
  • a successful connection with no inboxes yet.

Never claim the connection works until an authenticated tool call succeeds.

Examples

Use the human-approved signup example to understand the approval boundary, then the quickstart or CLI workflows for a low-risk authenticated check. Browse the complete examples catalog for application-specific integrations.

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