grant-management

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. It is designed to ingest and analyze external codebase data, such as proposal records and budget structures, and incorporate findings into a generated report. There are no explicit boundary markers or sanitization steps mentioned to prevent malicious instructions within the analyzed data from influencing the agent's behavior.
  • Ingestion points: Codebase files containing grant and award data (Phase 1).
  • Boundary markers: Absent.
  • Capability inventory: Local file writing to docs/grant-management-analysis.md.
  • Sanitization: Absent.
  • [SAFE]: The skill includes a Self-Evolution Telemetry mechanism that appends execution metadata to a local file in ~/.claude/projects/. This is a local logging function used for skill performance tracking and does not involve unauthorized data exfiltration or external network access.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 10:57 AM
Security Audit — agent-trust-hub — grant-management