Minutes

Detected

• Talks to external services11 places

  Makes outbound requests. The hosts are listed so you can see where data goes.

  api.anthropic.com, api.mistral.ai, api.openai.com

• Reads environment variables in bulk3 places

  Copies or enumerates the whole environment, which can include keys meant for other tools.

• Sends conversation content outward7 places

  Logs or forwards prompts, messages or tool results to a service outside your machine.

• Runs shell commands29 places

  Spawns processes on your machine. Expected for command, browser and build tools.

• Loads code at runtime21 places

  Uses eval, dynamic imports or encoded payloads, which static review cannot fully follow.

• Reads or writes other agents' configuration48 places

  Touches folders such as ~/.claude, ~/.codex or ~/.cursor, which hold settings and sometimes keys.

• Contains instructions aimed at AI agents12 places

  Text in code or agent-facing files that steers a model. Worth reading before you trust it.

• Runs unpinned packages at runtime15 places

  Executes npx or uvx without a version, so a compromised upstream release would run.

Not detected: accesses credential files, uses elevated privileges, installs itself to run later, matches a known malicious pattern.

Dependencies with known advisories

None found in the lockfiles and manifests checked against OSV.dev.

Left out of the profile

40 pattern matches in documentation, configuration or CI files were recorded but are not part of the profile, because they do not run when the server does. The owner dashboard lists them.