ingest-content

Pass

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection when processing course content.
  • Ingestion points: The skill reads course-manifest.json and markdown files (module-*.md, week-*.md) from various local directories or user-provided paths.
  • Boundary markers: Absent. There are no instructions for the agent to use delimiters or warnings to ignore potential instructions embedded within the ingested course files.
  • Capability inventory: The skill generates a TypeScript script ({{SCRIPTS_DIR}}/ingest-[slug]-course.ts) and executes it using pnpm tsx to perform database upserts.
  • Sanitization: Absent. The markdown content is parsed and converted to HTML without explicit validation or sanitization of the input text against malicious instructions.
  • [COMMAND_EXECUTION]: The skill generates and executes local code using the shell command pnpm tsx {{SCRIPTS_DIR}}/ingest-[slug]-course.ts based on external content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 20, 2026, 04:22 PM
Security Audit — agent-trust-hub — ingest-content