SkillShield
Review AI skill repositories with layered security analysis
Squidbrake
Apply rules and human approvals to AI agent actions
Meet Squidbrake
Squidbrake checks AI agent actions against a set of rules before they run. Routine work can proceed automatically, while actions that need review wait for a person to approve or reject them. The approval view explains the proposed change and records the request, the decision, and the surrounding sequence of activity.
Policies are written in YAML and can cover shell commands, coding tools, and connected MCP operations. The system also considers chains of actions, such as removing safeguards before deleting a resource. Teams can inspect live decisions, revoke access, and export an audit record. Squidbrake offers open-source self-hosting and hosted team options; enforcement depends on connecting the paths agents actually use.
Discussion
Tried it? Share a useful thought or ask a question.
Good conversations start with a hello.
Sign in to commentProduct details
- Website
- squidbrake.com
- Status
- Launched
- Category
- Security & privacy
- Listed
Explore related topics
Similar products
Darkmoon
Coordinate authorized security assessments and review evidence-backed findings.
mcpgawk
Monitor MCP tools and stop unapproved changes
Arcjet
Add runtime security decisions to applications and AI agent actions
Cencurity
Inspect AI coding traffic through a local security gateway
Strix
Continuously test applications and deliver validated security findings and fixes