tech-debt

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): Indirect Prompt Injection Surface. The skill is designed to ingest data from external sources such as 'Code reviews and PR comments' and 'Incident post-mortems'. If this data contains malicious instructions, they could influence the agent's reasoning or summary generation.
  • Ingestion points: Reads data/engineering/tech_debt.json which aggregates external content.
  • Boundary markers: No boundary markers or instructions to ignore embedded commands are present when processing external text.
  • Capability inventory: File system read (multiple JSON files) and file system write (data/engineering/tech_debt.json).
  • Sanitization: No sanitization or validation logic is specified for the ingested technical debt descriptions.
  • [DATA_EXPOSURE] (LOW): Sensitive Information Access. The skill reads data/cfo/latest_forecast.json. While no network exfiltration is present in the skill instructions, the inclusion of financial data in the context of a tool that processes untrusted external comments creates a risk of accidental data disclosure through the agent's response.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 01:17 PM