devcontainer-exec

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard CLI tools including docker and git to resolve container identities, inspect mount points, and execute project-specific commands (build, test, lint) within isolated environments.
  • [EXTERNAL_DOWNLOADS]: The skill invokes npx --yes @devcontainers/cli to download and run the official Development Containers command-line interface from the npm registry for container lifecycle management.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data including devcontainer.json and container labels to determine execution contexts.
  • Ingestion points: Metadata labels and configuration files accessed via docker inspect (SKILL.md).
  • Boundary markers: Explicit instructions to "ask, never guess" and verify container identity before execution.
  • Capability inventory: File system access through bind mounts and command execution via docker exec and docker start/stop (lifecycle.md).
  • Sanitization: None mentioned for metadata parsing, but the skill mandates user approval for all destructive or lifecycle-altering actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 03:56 PM
Security Audit — agent-trust-hub — devcontainer-exec