maintaining-project-lessons
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements security best practices by explicitly instructing the agent to redact credentials and sensitive information ("去除凭证等敏感原文") when gathering evidence from project files, issues, or commit logs.
- [SAFE]: External links are provided solely for methodology attribution and documentation purposes, pointing to well-known services (GitHub, X). There are no automated downloads, package installations, or remote code execution patterns.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests untrusted data from external logs and issue reports to generate documentation.
- Ingestion points: Reads project files (
LESSONS.md,AGENTS.md), issue reports, and commit history. - Boundary markers: Uses structured markdown templates and specific ID-based numbering to segregate extracted evidence.
- Capability inventory: Limited to reading and writing local project documentation files.
- Sanitization: Mandatory instructions are provided to remove sensitive identifiers and credentials from processed data.
Audit Metadata