Audit web app security with proof and fix real npm risk
an evidence-first security audit for your own web application.
Free
Works with the AI Agents you already use
Audit web app security with proof and fix real npm risk
Example session with this skill installed
Illustrative npm audit --json report (6 packages, made-up names) and matching package-lock.json.
Command: python3 scripts/npm_audit_triage.py npm_audit.json --lock package-lock.json
- Read your context and instructions
- Compiled the audit web app
package sev scope direct installed fix
sheet-parser high runtime True 0.18.5 none
web-framework high runtime True 4.2.1 same-major web-framework@4.2.2
bundler-dev high dev-only True 4.5.3 SEMVER-MAJOR bundler-dev@5.0.0
image-native high dev-only False 0.33.1 same-major npm audit fix
pwa-plugin moderate dev-only True 1.1.0 DOWNGRADE pwa-plugin@0.0.1
workbox-build moderate dev-only False 7.1.0 DOWNGRADE pwa-plugin@0.0.1
total=6 runtime high/critical=2 dev-only=4
DO NOT APPLY (fix is older than installed): pwa-plugin (via pwa-plugin@0.0.1); workbox-build (via pwa-plugin@0.0.1)
Needs changelog review (major): bundler-dev@5.0.0
No npm fix (check upstream/overrides): sheet-parser
Note: Real run of the triage script on an illustrative audit file with made-up package names.
Connects securely to your tools. The creator never sees your data.
About this skill
What it does: an evidence-first security audit for your own web application. Each checklist point gets one probe, the agent reads every hit instead of counting greps, proves access control with unauthenticated requests (expect 302/401/403, never 200 or 500), and reports per point with a tick or a risk plus the cost to close it. Includes proven recipes: secret scanning in CI with value-level allow-lists and a test that the allow-list is not too wide, magic-byte upload checks, dependency patching without breaking the build (scratch-copy simulation, downgrade detection, major-bump changelog checks), answering 'does this CVE list affect us?', and a safe CSP rollout. Who it is for: developers and small teams who must answer 'are we secure?' honestly before a release or after a scary headline. What is inside: SKILL.md, report template; scripts/npm_audit_triage.py (npm 7+ audit JSON + lockfile) and scripts/file_signature_check.py, both with offline self-tests. Limits: not a penetration test or a substitute for professional review. Only for systems you own or are authorized to test.
- Covers common web-app checks; it is not a full penetration test.
- Runtime vs dev scope comes from lockfile flags; bundlers can still ship dev packages, which the skill tells you to verify.
- npm ecosystem only for the triage script; other ecosystems follow the same method by hand.
- Live checks are unauthenticated only by design.
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
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed today
- Free with an Agensi account
Permissions
Allowed Hosts