What Plugins is for
A plugin is useful when the missing piece is executable behavior: a tool that calls an internal service, a lifecycle hook that records an event, or an integration shared by a team. Official documentation describes a manifest and Python registration code, separate from modifications to the Hermes core. Choose this route only after checking whether an existing tool, skill, or MCP server already solves the problem.
What you need
- A maintained Hermes installation whose plugin API matches the plugin documentation.
- A trusted plugin source, declared dependencies, and an explicit description of data access and external effects.
Set it up
- 1
Define one narrow capability and its input/output contract. Decide which actions are read-only and which require additional authorization.
- 2
Develop against the official example in an isolated profile. Keep credentials outside source code and document an enable/disable and rollback procedure.
Check it works
- Confirm that the expected tool is registered, run one harmless example, and inspect both the returned data and any external side effects.
- Exercise a missing credential or invalid input and verify that it fails clearly without silently expanding permissions.
Limits and failure points
- A plugin can run code; a descriptive manifest is not a security boundary.
- Compatibility and dependency maintenance belong to the plugin operator. Documentation support does not mean every third-party plugin is verified or interchangeable.
Primary sources
Spotted something wrong or out of date? Tell us on X at @HermesWatcher (opens in a new tab) and we will fix it.