OpenClaw
Connect OpenClaw to PostMantis through MCP, and optionally add thin skills on top.
Recommended OpenClaw integration
Use PostMantis as a remote MCP server first. Add OpenClaw skills only when you want reusable workflow instructions on top.
What OpenClaw is
OpenClaw is an open-source personal AI agent. It can use MCP servers, install AgentSkills-compatible skills, and run automations through tools you already trust.
For PostMantis, the clean model is:
- PostMantis stays the publishing system of record
- MCP is the main integration path
- OpenClaw skills are optional wrappers for repeatable workflows
Use MCP first
PostMantis already exposes a standard remote MCP server:
https://postmantis.com/mcpThat gives OpenClaw direct access to the same PostMantis actions and readbacks documented elsewhere:
- profiles
- posts
- upload sessions
Register PostMantis as an MCP server in OpenClaw
If your OpenClaw runtime uses saved MCP server definitions, register PostMantis like this:
openclaw mcp set postmantis '{
"url": "https://postmantis.com/mcp",
"transport": "streamable-http",
"headers": {
"Authorization": "Bearer <POSTMANTIS_API_KEY>"
}
}'Then inspect the saved definition:
openclaw mcp show postmantisUse a dedicated PostMantis API key for this client. Keep it in secret storage or environment-backed configuration instead of hardcoding it into prompts.
Keys are still PostMantis keys
OpenClaw does not replace the PostMantis auth model. The same PostMantis API key scope, provider restrictions, and profile restrictions still apply.
Follow the normal PostMantis flow
List profiles
Start with postmantis.list_profiles if the destination profile IDs are not already known.
Create a draft or scheduled post
Use postmantis.create_post for draft-first flows, immediate publish requests, or future scheduling.
Publish, cancel, or delete as needed
- use
postmantis.publish_draftto publish or schedule a draft - use
postmantis.cancel_postfor scheduled posts that should not go out - use
postmantis.delete_draftfor drafts you want to discard
Read the post back
Use postmantis.get_post to inspect deliveries[] and confirm the real result per profile.
Inspect delivery outcomes later
Use postmantis.get_post after the workflow runs and inspect deliveries[] before reporting the
final result.
Add skills only when they help
OpenClaw supports AgentSkills-compatible SKILL.md folders and can install published skills from
ClawHub.
A good PostMantis skill should help OpenClaw use MCP correctly. It should not replace the real contract.
Good uses for a skill
Draft-first assistant
Default to drafts, then publish only after explicit approval.
Scheduling helper
Turn natural-language scheduling requests into exact future publish times.
Delivery follow-up
Summarize completed and failed posts, and call out anything that still needs attention.
Example local skill
Store a skill in one of OpenClaw's normal skill locations, such as <workspace>/skills/postmantis/SKILL.md:
---
name: postmantis
description: Use PostMantis MCP tools for drafting, scheduling, publishing, and delivery follow-up.
---
When profile IDs are unknown, call postmantis.list_profiles first.
Default to creating drafts unless the person explicitly asks to publish now or schedule.
Use postmantis.delete_draft for drafts that should be discarded.
Use postmantis.cancel_post for scheduled posts that should not be sent.
After publish actions, call postmantis.get_post and inspect deliveries[] before reporting success.Useful OpenClaw commands
openclaw skills list
openclaw skills list --eligible
openclaw skills install <skill-slug>
openclaw skills info <name>Start a new OpenClaw session after installing or changing a skill so the updated skill set is loaded.
Keep the integration clean
Use this layering:
- PostMantis HTTP, OpenAPI, MCP, define the real product behavior
- OpenClaw consumes PostMantis through MCP
- OpenClaw skills add small workflow hints on top
That gives you a clean setup with one publishing contract instead of one contract per AI client.
Notes that matter
- keep PostMantis API keys out of prompts and skill files
- prefer one dedicated key per OpenClaw automation or workspace
- use read-only keys unless the agent really needs write access
- do not treat accepted write calls as final publish success
- read
deliveries[]