weekly-composer-dependency-audit

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands (composer outdated, composer audit) to perform its primary task. These commands are restricted to read-only auditing operations and do not include package installation or system modification.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing and summarizing untrusted data from project configuration files.
  • Ingestion points: The skill reads contents from composer.json and composer.lock in the local project directory (SKILL.md).
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard potential instructions embedded within the ingested metadata.
  • Capability inventory: The agent has the capability to execute shell commands and generate natural language reports based on the findings.
  • Sanitization: No evidence of metadata validation, filtering, or escaping is present before the data is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 12:04 PM
Security Audit — agent-trust-hub — weekly-composer-dependency-audit