Audit History
agent-browser - 8 audits
Version comparison
Capability and finding changes across audited versions, newest first.
| Version | Date | Result | Review items | Change vs previous |
|---|---|---|---|---|
| v8 Latest | Jul 27, 2026, 11:19 AM | No confirmed findings | 0 | No capability change |
| v7 | Jul 21, 2026, 09:54 AM | 4 confirmed | 0 | No capability change |
| v6 | Jul 18, 2026, 09:12 AM | No confirmed findings | 0 | No capability change |
| v5 | Jul 17, 2026, 10:54 PM | 6 confirmed | 4 | No capability change |
| v4 | Jul 12, 2026, 09:36 AM | 4 confirmed | 4 | No capability change |
| v3 | Jul 8, 2026, 02:18 PM | 6 confirmed | 19 | No capability change |
| v2 | Jul 9, 2026, 11:09 AM | 5 confirmed | 72 | No capability change |
| v1 | Jul 8, 2026, 02:18 PM | 6 confirmed | 19 | Baseline |
Jul 27, 2026, 11:19 AM
All 265 static findings are false positives in documentation or transparent shell templates. The skill documents user-directed browser automation and contains no evidence of prompt injection, credential exfiltration, hidden network activity, or malicious system access.
Risk Factors
⚙️ External commands (50)
🌐 Network access (50)
📁 Filesystem access (17)
Jul 21, 2026, 09:54 AM
Static detections primarily match Markdown, shell examples, placeholders, and expected browser-automation documentation; they do not demonstrate hidden execution or embedded malware. The audit identified material capability risks: unrestricted belt CLI permission, third-party hosted processing of browser data, and rate-limit avoidance guidance through proxy rotation. Publication should require clear consent, narrowed permissions, and removal of evasion-oriented guidance.
Confirmed security concerns (4)
Risk Factors
⚙️ External commands (50)
🌐 Network access (50)
📁 Filesystem access (17)
Jul 18, 2026, 09:12 AM
Adjudicated 265 static findings against the agent-browser skill, which consists of Markdown documentation and example/template shell scripts wrapping the 'belt'/'infsh' CLI for Playwright browser automation. Every finding is a false positive: shell command substitutions capture CLI output into variables with hardcoded literal arguments, 'hardcoded URLs' are example.com/docs placeholders, 'path traversal' matches are relative Markdown links, and the credential-handling examples actually teach secure practices (secrets managers, no logging). No data exfiltration, command injection, or prompt injection was found.
Risk Factors
⚙️ External commands (50)
🌐 Network access (50)
📁 Filesystem access (17)
Jul 17, 2026, 10:54 PM
Most static alerts are false positives caused by Markdown syntax, example URLs, relative links, and ordinary shell substitutions. Confirmed risks include hosted screen capture, SOCKS proxies, rate-limit evasion, secret handling, external dependency trust, and unrestricted browser actions. Publication should require stronger consent, trust, and abuse-prevention guidance.
Confirmed security concerns (6)
Capability review items (4)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Risk Factors
⚙️ External commands (50)
🌐 Network access (50)
📁 Filesystem access (15)
Detected Patterns
Jul 12, 2026, 09:36 AM
Most static alerts are Markdown syntax, examples, relative links, or routine shell redirection. Confirmed concerns include external dependency installation and SOCKS proxy routing. Semantic review found remote credential handling, cookie logging, local file uploads, and browser-scoped JavaScript execution, with no prompt injection evidence.
Confirmed security concerns (4)
Capability review items (4)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Risk Factors
⚙️ External commands (50)
🌐 Network access (50)
📁 Filesystem access (17)
Jul 8, 2026, 02:18 PM
Most static hits are documentation examples or Markdown syntax, including path traversal and hardcoded example URLs. I confirmed the executable shell templates, SOCKS proxy support, and screen capture or video behavior as real security-sensitive capabilities. Semantic review also found sensitive risks around cookies, TOTP flows, file uploads, page JavaScript execution, and proxy-based rate limit avoidance.
Confirmed security concerns (6)
Capability review items (19)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Risk Factors
⚙️ External commands (184)
🌐 Network access (60)
📁 Filesystem access (17)
Detected Patterns
Jul 9, 2026, 11:09 AM
Many static findings are markdown, example URL, relative-link, or cleanup-pattern false positives. Confirmed risk remains because the skill enables remote browser automation with external CLI commands, screenshots, video, proxies, authenticated sessions, file uploads, and page JavaScript. No prompt injection attempt was found in the reviewed files.
Confirmed security concerns (5)
Capability review items (72)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.
Risk Factors
⚙️ External commands (50)
🌐 Network access (50)
📁 Filesystem access (17)
Detected Patterns
Jul 8, 2026, 02:18 PM
Most static hits are documentation examples or Markdown syntax, including path traversal and hardcoded example URLs. I confirmed the executable shell templates, SOCKS proxy support, and screen capture or video behavior as real security-sensitive capabilities. Semantic review also found sensitive risks around cookies, TOTP flows, file uploads, page JavaScript execution, and proxy-based rate limit avoidance.
Confirmed security concerns (6)
Capability review items (19)
These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.