sk Skill · JuliusBrussee
caveman-manage
Inspect Caveman Cloud's experiment lifecycle and block unsafe execution. Use when asked to start, approve, cancel, promote or roll back a Caveman experiment.
Open on skills.sh ↗read 2026-09-17
- installs 8w
- 47,491
- 30-day movement
- starts with the next reading
- Related entries
- 1
- Connections
- 0
bashGo
An agent skill that inspects the lifecycle of Caveman Cloud experiments and blocks unsafe lifecycle changes. It reads experiment state and evidence, then either recommends one supported action (start, approve, cancel, promote, roll back) or refuses to execute.
Lifecycle changes to experiments are production control actions, and this skill enforces evidence and state gates before any change is proposed.
Use it to
- List experiments when no id is given
- Evaluate evidence before approving an experiment
- Block unsafe start, approve, promote or rollback commands
- Re-read experiment state after an operator runs a command
For Operators managing Caveman Cloud experiments via an AI agent
- Host repository
- JuliusBrussee/caveman
- Installs, lifetime
- 55k
- Installs, 8 weeks
- 47k
- Host stars
- 106k
- Host language
- Go
topicscavemanexperimentslifecyclesafety-gatesagent-skillevals