ccdawn-feature-reuse-research

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is instructed to process content from external URLs (GitHub, documentation, package registries) and local project files. While the skill's actions are restricted to read-only tasks, content from these untrusted sources could attempt to manipulate the research conclusions or the agent's logic for the next implementation steps.
  • Ingestion points: External data from GitHub repositories, official documentation, and package registries (npm, PyPI, Cargo, etc.), along with local code modules as defined in SKILL.md.
  • Boundary markers: The instructions do not specify the use of delimiters or explicit boundary markers to isolate untrusted external content from system instructions.
  • Capability inventory: The skill is strictly limited to read-only search and local file reads. It explicitly forbids installing dependencies, writing code, or executing external scripts.
  • Sanitization: No specific sanitization, validation, or filtering logic is mentioned for the external content processed during the research phase.
  • [SAFE]: The skill incorporates strong security-conscious instructions, such as the explicit prohibition of installing dependencies, running external code, or copying external code directly into the project. It focuses exclusively on read-only evaluation and decision-making.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 09:31 AM
Security Audit — agent-trust-hub — ccdawn-feature-reuse-research