debt-ops-init

Pass

Audited by Gen Agent Trust Hub on Jul 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell and Python commands to identify the project root and read local cache files from the user's home directory.
  • Evidence: The script calculates a repository hash using git rev-parse --show-toplevel and python3 -c to access files in $HOME/.cache/debt-ops.
  • [PROMPT_INJECTION]: The skill incorporates external, unverified data from local cache files and project manifests into the agent's persistent instructions, which are then executed by the agent during normal workflows.
  • Ingestion points: Reads from $HOME/.cache/debt-ops/cache/$REPO_HASH/feedback.list and project manifests (e.g., package.json, pyproject.toml).
  • Boundary markers: Uses HTML comments <!-- debt-ops:feedback v1 --> to delimit the injected content in the AGENTS.md file.
  • Capability inventory: The skill modifies local files (AGENTS.md) and explicitly instructs the agent to execute the resulting 'Quality checks' block after future edits.
  • Sanitization: No validation or sanitization of the command strings is performed before they are interpolated into the prompt template.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 4, 2026, 08:49 AM
Security Audit — agent-trust-hub — debt-ops-init