Works with the AI Agents you already use
CookieGuard AI — Browser Agent Authentication Security Audit
Audit session-cookie security in Browser Use, Playwright, and CI without exposing cookie values to your agent.
Free
CookieGuard AI — Browser Agent Authentication Security Audit
Example session with this skill installed
Check whether the session cookie our Browser Use agent gets when it logs in to our app at http://127.0.0.1:3000 is configured securely, and give me a report with any fixes.
- Read your context and instructions
- Compiled the cookieguard ai
Summary
Your Browser Use agent's login sets the session cookie session_id
without HttpOnly, so CookieGuard reports
FAIL (exit code 1). One confirmed high-severity finding, one-line fix. After the fix, the same agent login re-audited as
PASS. No cookie value ever reached the agent.
Scope: origin http://127.0.0.1:3000 · auth cookie session_id · policy failOn=high, SameSite lax/strict, max lifetime 30 days · workflow: the agent's form login · evidence: Set-Cookie headers + browser cookie store
Findings
| Rule | Severity | Status | Cookie | Evidence | Fix |
|---|---|---|---|---|---|
| CG001 Missing HttpOnly | high | confirmed | session_id | Set-Cookie without HttpOnly (Secure ✓, SameSite=Lax ✓) | Add HttpOnly at the server |
JavaScript on the page can read a session cookie without HttpOnly. That raises token exposure if script execution is ever compromised. It does not by itself show an XSS exists.
CookieGuard output (before)
Target: http://127.0.0.1:3000
Mode: browser; classification: rule
Cookies inspected: 1 (1 observations)
Authentication cookies: 1 confirmed, 0 candidate, 0 unknown
Findings:
HIGH CG001 — "session_id": Missing HttpOnly
Fix: Set HttpOnly on the authentication cookie at the server.
Result: FAILED — 1 confirmed violations at or above high
CookieGuard for Browser Use: FAIL (exit code 1); evidence: server-declarations-and-browser-store
The fix
- Set-Cookie: session_id=<value>; Path=/; Secure; SameSite=Lax
+ Set-Cookie: session_id=<value>; Path=/; Secure; HttpOnly; SameSite=Lax
Enforcement added to your agent run
audit = CookieGuardAudit(allowed_origins=['http://127.0.0.1:3000'], auth_cookies=['session_id'])
await audit.start(session) # before the login
await agent.run()
result = await audit.finish() # decided by CookieGuard, not the agent
result.write_json('cookieguard-result.json')
result.assert_policy() # fails the run on FAIL or INCOMPLETE
Re-audit after the fix
Findings:
No observed policy violations.
Result: PASSED — 0 confirmed violations at or above high
CookieGuard for Browser Use: PASS (exit code 0); evidence: server-declarations-and-browser-store
Limitations: cookie-configuration audit only. XSS, CSRF, and server-side session expiry or revocation were not tested.
Next steps
- Ship the
HttpOnlychange in your login handler. - Keep
assert_policy()in the agent run or CI, so a regression fails the build. - Rerun the audit against staging before release.
Connects securely to your tools. The creator never sees your data.
About this skill
Audit authentication-cookie security in AI browser automation without exposing cookie values to the agent.
CookieGuard AI checks the session cookies your login sets: missing HttpOnly or Secure, SameSite policy, Domain scope, lifetime, and __Host-/__Secure- prefix problems. Its local engine returns PASS, FAIL, or INCOMPLETE with rule IDs and concrete fixes, and the agent can't override the verdict. Works with Browser Use agents (PyPI: cookieguard-browser-use), Playwright tests and CI pipelines (npm: @aiscihub/cookieguard-playwright, cookieguard-ci), and MCP clients (@aiscihub/cookieguard-mcp).
The skill tells your agent when CookieGuard applies and when it doesn't, installs it only with your agreement, runs the audit through your existing setup, and writes a report with findings, fixes, and limitations. Cookie values, tokens, and passwords never appear in output or reports.
Scope: configuration audit only, not XSS, CSRF, or server-side session testing. Requires Node.js 24+; Browser Use also needs Python 3.11+ and Chrome or Chromium. Use only on applications you own or are authorized to test. Open source (MIT): https://github.com/aiscihub/cookieguard-ai
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse All Claude Skills.
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 buying a skill.
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 to use
No account required to browse
Trust & safety
Security scan
Security scan passed 10 days ago
- Free with an Agensi account
- Software
Python
Node.jscookieguard-ci@aiscihub/cookieguard-mcp