Vuln Scan
Scan a Python (uv) and Docker repo for known CVEs, judge real exposure, and get a decision-ready patch list.
$9
Works with the AI Agents you already use
vuln-scan
Example session with this skill installed
Any vulnerabilities we should patch?
- Read your context and instructions
- Compiled the vuln-scan
23 findings, 1 critical. The two that matter: urllib3 2.2.1 → 2.5.0 (affects every outbound call) and starlette 0.37 → 0.47.2 (served to untrusted input; the fastapi<0.116 cap has no recorded reason and the bump resolves cleanly). 9 low-value fixes ride along in the same lock refresh; 11 Debian base-image items have no fix yet. Recommended: one PR with uv lock --upgrade-package for urllib3, starlette, h11. Raw files: ~/.claude/vuln-scans/my-service/2026-11-16/
Connects securely to your tools. The creator never sees your data.
About this skill
Most vulnerability scanners hand you a wall of CVEs. This one tells you which ones matter for how your service actually runs, and what to bump.
What it scans
- Python dependencies in
uv.lock, withpip-audit(run throughuvx, nothing installed) and severities from the OSV database. - The Dockerfile's final-stage base image with Trivy, locally. No Docker Hub or Docker Scout account needed.
- Optionally, the real built image (
--image), built from committed files only, as CI sees them. - GitHub Dependabot alerts, cross-checked so you learn why Dependabot reports things the scan doesn't (stale manifests, newer advisories, accepted risks).
How it triages
For each fixable finding it checks how the package arrives, whether it's prod or dev only, whether the library serves untrusted input or is only a client, whether the vulnerable feature is used, and the platform. It then confirms that each bump actually resolves with uv lock --dry-run, so "major bump" is never a lazy reason to skip a fix.
What you get
A headline, then tables: Patch now, Low value (same lock refresh), Not applicable, and Waiting on Debian (base image). Plus what's new or resolved since your last scan, a "What I didn't cover" list, and one recommended action. Raw TSV results are kept per day for comparison.
It's read-only on the repo: no lockfile edits, branches or commits.
Python support covers uv.lock only (other manifests are surfaced through the Dependabot cross-check). Packages from private indexes have no advisory data. Investigation only: it recommends fixes but doesn't apply them unless you ask.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed today
Requires
Permissions
Allowed Hosts
File Scopes
With --image it builds a local Docker image and, only if the Dockerfile declares a gcp_token build secret, reads a gcloud access token into an env var for the build. Tokens are never printed.