ln-522-manual-tester

Pass

Audited by Gen Agent Trust Hub on Apr 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions specify a fallback mechanism to fetch reference documentation from a remote GitHub repository (https://raw.githubusercontent.com/levnikolaevich/claude-code-skills/master/skills/) using the WebFetch tool if local files are unavailable. This repository belongs to the skill's author.
  • [COMMAND_EXECUTION]: The skill's primary workflow involves creating bash scripts within the target project's tests/manual/ directory, modifying their permissions using chmod +x, and executing them via the Monitor or Bash tools to verify acceptance criteria.
  • [REMOTE_CODE_EXECUTION]: The skill dynamically generates shell scripts which are then executed. This behavior is intended for testing purposes but represents a capability for running arbitrary code generated by the agent.
  • [PROMPT_INJECTION]: There is an attack surface for indirect prompt injection as the skill processes data from untrusted sources, including Story descriptions (Acceptance Criteria), infrastructure.md, and runbook.md.
  • Ingestion points: Story AC, project infrastructure documentation, and project runbooks.
  • Boundary markers: Absent; the skill interpolates this data directly into the logic used to generate bash scripts.
  • Capability inventory: File system writes, permission modification (chmod), and shell command execution (Bash, Monitor).
  • Sanitization: No explicit validation or escaping of input data is defined before it is used in script generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 14, 2026, 06:40 PM