ingest-context-skills

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill acquires its runtime environment and importer logic from the vendor's official GitHub repositories.
  • Evidence: The launcher script references https://github.com/getcolors/context-sql.git and https://github.com/getcolors/blue.git.
  • [COMMAND_EXECUTION]: The skill uses a bash wrapper to execute the uv package manager and Python code to perform database and Git operations.
  • Evidence: The context-skills script uses uv run to execute the package_context_sql_blue.ingest module.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides a surface for processing external data by ingesting other agent skills from remote repositories.
  • Ingestion points: Upstream skill repositories (e.g., getcolors/skills) imported via the ingest command as described in SKILL.md.
  • Boundary markers: The documentation explicitly states that selection does not certify the source follows standards and mandates that scripts in imported payloads must not be executed.
  • Capability inventory: The skill performs database writes to PostgreSQL, reads/writes local configuration files in ~/.config/context-sql/, and interacts with the Git CLI.
  • Sanitization: The importer implements a verification process that independently compares every selected path and byte with the source Git repository to detect unauthorized modifications.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 10:53 AM