View installed skills
In the desktop app, connect to the gateway and click the Skills icon in the bottom bar. For each skill, review:- name
- version
- status
- source
- enabled/disabled state
- implicit/not implicit state
- dependency diagnostics
- validation issues
- security findings
- audit timeline
Enable, disable, or change implicit mode
Disable skills by default until you know what they do. Newly installed skills are not implicit by default; after a skill is trusted, decide whether it should stay explicit-only or become implicit. An implicit skill can be picked up automatically when Pioneer decides it is relevant. A not implicit skill stays available, but agents should use it only when you explicitly select it in the composer or ask for that skill in the prompt. Recommended flow:1
Install the skill
Install and wait for validation.
2
Inspect the skill
Read its description, dependencies, and security findings.
3
Test on a sandbox gateway
Use a sandbox gateway first.
4
Run a safe test
Ask the agent to explain the skill before letting it call tools.
5
Choose implicit behavior
Keep the skill not implicit unless it is trusted and broadly useful.
6
Use in real work
Select explicit-only skills from the composer when a turn needs them, and use the skill only on gateways where it is needed.
Update a skill
When a new version is available:- Review the source and changelog if available.
- Update in a sandbox first.
- Recheck dependencies and security findings.
- Run a safe test.
- Use the updated skill on production gateways only after testing.
Uninstall a skill
Uninstall a skill when:- you no longer use it
- it has risky permissions
- dependencies are broken
- you want to replace it with a newer package
- it was installed on the wrong gateway
Audit history
Use the audit timeline to answer:- Who installed or updated the skill?
- When did policy change?
- Was the skill blocked?
- Did dependencies change?
- Were there validation warnings?
Good management habits
- Keep a sandbox gateway for testing skills.
- Avoid enabling skills globally unless needed.
- Keep specialized skills not implicit.
- Review skill tools before allowing write access.
- Remove unused skills.
- Recheck skill health after OS or dependency changes.
Further reading
- Skills Architecture explains skill policy, resolver behavior, runtime tools, and health snapshots.
- Agent Loop explains how explicit and implicit skills affect a turn.