MCP server
squidbrake
Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.
- Version
- main
- Analysed
- Last observed
What the code does
- Runs shell commands Code that starts system commands on the machine it runs on.
- Sends data to outside servers Code that posts data to network addresses.
- Brings outside content into the agent Tools that return web pages, search results, documentation or messages written by others. That text reaches the model, and instructions planted in it can steer the agent (indirect prompt injection). Treat what these tools return as untrusted input.
- Refers to credential files Code that names files such as SSH keys or cloud credentials.
- Downloads and runs code Code that fetches a script and pipes it into a shell.
- Runs hooks automatically Hooks that execute without being called explicitly.
- Runs an AI coding agent unsupervised Code that starts an AI coding agent with its approval prompts switched off.
- Reads AI and developer logins Code that refers to AI tool credentials or publishing tokens, such as .npmrc.
Having a capability is not wrongdoing; many tools need these to do their job. It tells you what to check before you install.
This is public metadata, not a safety certification or complete assessment. Static analysis cannot establish what a tool does at runtime.
Risk
Sign in to see this tool's overall risk level and what drives it. Free.
Coverage
Complete static analysis
Detailed report
16 findings from automated static analysis of the latest version, with severities and explanations.
Using this tool across your team? We can assess your whole inventory.
Request an inventory assessment