colab-attach — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited colab-attach (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Connects the OpenInterp MCP server to the researcher's compute session (Colab Pro / Free / vast.ai / runpod / local).
The researcher pastes a URL like https://abc123.ngrok-free.app and asks you to attach, OR you see no default session active in ~/.openinterp/sessions.json and a later openinterp skill is about to run.
Call the MCP tool `colab_attach` with:
endpoint: the HTTPS URL the researcher providedsession_name: defaults to default; only override if user explicitly names a sessionIf the tool returns {"ok": true, "health": {...}}, report the loaded model_id and number of probes available. If it returns an error, the URL is unreachable or the backend isn't running — ask the researcher to verify their Colab cell completed.
Compact one-liner:
"Attached to {model_id} at {endpoint}. {probes_loaded} probes available."
If the researcher wants details, call colab_status afterward.
~/.openinterp/sessions.json).~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.