lov-daily-post

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill incorporates a local profile management system via scripts/profile_store.py that enforces restricted access to sensitive data fields. It explicitly blocks the storage of credentials, tokens, and secrets in user-owned profile files by checking for sensitive keywords.
  • [INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface for indirect prompt injection through its ingestion of project-related data.
  • Ingestion points: The skill reads project CHANGELOGs, commit histories, and release records from the target workspace as specified in SKILL.md.
  • Boundary markers: Explicit execution boundaries are defined to limit the agent's actions to the user's specific request and prevent unauthorized system changes or bypassing of host protections.
  • Capability inventory: The skill uses local file read/write operations for configuration persistence and relies on host-provided CLI and browser tools for data gathering.
  • Sanitization: The references/authorship-integrity.md contract requires a discourse audit to verify that all output is traced to specific project evidence, acting as a secondary validation layer against malicious instructions embedded in project logs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 09:46 AM
Security Audit — agent-trust-hub — lov-daily-post