What problem does it solve?
Agents managing PostHog skills through the MCP skill-* tools often waste context, make redundant round-trips, or corrupt skills by choosing the wrong write primitive. This Skill provides the decision tree, efficiency rules, and pitfall guidance needed to discover, read, create, update, and archive skills correctly—especially large skills with many bundled files.
Core Features & Use Cases
- Tool selection decision tree: Maps every task (discover, read, create, edit body, patch files, rename, archive) to the correct
skill-* tool with the smallest appropriate write primitive.
- Progressive disclosure workflow: Enforces lazy loading—list descriptions first, fetch one body per task, and pull bundled files only when the body references them.
- Concurrency-safe writes: Explains
base_version optimistic locking, chaining versions across sequential writes, and recovering from 409 conflicts.
- Use Case: An agent asked to fix a typo in a skill's bundled script uses
skill-get to capture the version, then a single skill-update with file_edits instead of round-tripping the entire bundle—producing one clean, reviewable version.
Quick Start
Ask the agent to update an existing PostHog skill's body or bundled files using the smallest appropriate skill-* tool with a valid base_version.
