context-distiller

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local shell commands to map codebases and aggregate context files.
  • [REMOTE_CODE_EXECUTION]: The skill recommends using package runners to execute context management tools. Evidence: Recommends bun x repomix in references/automated-context-packing.md to fetch and run the Repomix utility.
  • [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes external files. 1. Ingestion points: Aggregates context from docs/AGENTS.md, .gemini/GEMINI.md, and project source files. 2. Boundary markers: The skill does not define specific isolation markers for ingested content. 3. Capability inventory: The agent has capabilities for shell command execution and file operations. 4. Sanitization: While it provides guidance on scrubbing secrets, it does not include specific sanitization for potential instructions embedded in the project files.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 08:27 PM