project-retrospective

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious commands or safety violations were detected in the skill instructions or metadata.
  • [REMOTE_CODE_EXECUTION]: The skill instructs the agent to suggest installing an external tool from 'github:mindverse/skillhub' using the platform's '/install' command. This is a legitimate feature for skill discovery and discovery-based workflows.
  • [PROMPT_INJECTION]: The skill processes external data provided by the user (sprint metrics, incident timelines, and feedback), which represents a surface for indirect prompt injection. The risk is assessed as negligible because the skill lacks the ability to execute system commands, access the network, or write to the file system.
  • Ingestion points: User-provided inputs collected during 'Step 2: 信息收集' (Information collection).
  • Boundary markers: None identified; user data is directly interpolated into Markdown templates.
  • Capability inventory: Limited to text processing; no 'allowed-tools' are requested in the frontmatter, and no high-privilege subprocess or network operations are present.
  • Sanitization: None detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 03:44 PM