nocobase-revision

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the shell tool to interact with the NocoBase CLI (nb) for environment validation and revision management.\n
  • It performs plugin discovery using nb api pm list to ensure the version control plugin is active.\n
  • It executes nb revision create to finalize and save application milestones.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied milestone descriptions for inclusion in shell commands, which is a potential injection surface.\n
  • Ingestion points: The <description> argument defined in the frontmatter and used throughout the workflow serves as the primary data ingestion point from the user.\n
  • Boundary markers: The instructions attempt to wrap the description in double quotes ("<DESCRIPTION>") when constructing the shell command to delimit the input.\n
  • Capability inventory: The agent is granted shell access, enabling it to execute commands that incorporate this untrusted data.\n
  • Sanitization: The skill provides guidelines to normalize input into professional summaries and enforces a 2000-character limit to reduce the likelihood of complex injection payloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 02:00 AM
Security Audit — agent-trust-hub — nocobase-revision