php-pro
Fail
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [CREDENTIALS_UNSAFE]: Hardcoded database credentials found in code examples within reference files. File 'references/async-patterns.md' contains 'password' => 'password' and 'root:password@localhost/database' in connection examples.
- [COMMAND_EXECUTION]: The skill requires executing local binaries and shell commands. 'SKILL.md' instructs the agent to run 'vendor/bin/phpstan analyse --level=9' and 'vendor/bin/phpunit', as well as 'mkdir -p .aix'.
- [PROMPT_INJECTION]: Vulnerability to indirect prompt injection through processing untrusted data with executable capabilities. Ingestion points: local project files referenced in 'SKILL.md'. Boundary markers: absent. Capability inventory: shell execution of local binaries. Sanitization: absent.
- [DATA_EXFILTRATION]: Performs network operations to non-whitelisted domains. File 'references/async-patterns.md' includes code examples that fetch data from 'api.example.com' using Swoole coroutine clients.
Recommendations
- AI detected serious security threats
Audit Metadata