dataset

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local TypeScript files (dataset/search.ts and dataset/validate.ts) using the bun run command. These executions are scoped to the skill's own internal logic for data retrieval and schema validation.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests project records from external JSON files located in dataset/entries/.
  • Ingestion points: Reads project data from dataset/entries/<id>.json.
  • Boundary markers: None identified; the agent is instructed to treat lessons as evidence.
  • Capability inventory: The skill can execute shell commands via bun.
  • Sanitization: Includes a validation script (dataset/validate.ts) that checks entries against a JSON schema (dataset/schema.json) before they are accepted.
  • [SAFE]: The skill does not perform any unauthorized network requests, credential harvesting, or obfuscation. Its behavior is consistent with its stated purpose of knowledge management.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 06:03 AM
Security Audit — agent-trust-hub — dataset