agent-supply-chain

Pass

Audited by Gen Agent Trust Hub on Sep 28, 2026

Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill provides Python code templates in references/integrity-patterns.md and a GitHub Action manifest in references/ci-integration.md that uses python -c to execute multiline strings. This is used for generating and verifying SHA-256 integrity manifests.
  • [INDIRECT_PROMPT_INJECTION]: The manifest generation and verification workflows involve reading content from files within user-specified directories. This creates an ingestion surface for potentially untrusted data that the agent processes to compute hashes, satisfying the evidence chain for this category.
  • Ingestion points: Reads all files in the plugin directory in references/integrity-patterns.md.
  • Boundary markers: None specific to the data being read; files are treated as binary for hashing.
  • Capability inventory: File reading, file writing (INTEGRITY.json), and CI script execution.
  • Sanitization: None; the skill reads raw file content to generate SHA-256 digests.
  • [COMMAND_EXECUTION]: The CI integration documentation recommends executing Python logic via shell commands (python -c) in a CI pipeline environment to gate deployments based on integrity checks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 28, 2026, 05:29 PM