ai-execute-task

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Executes shell commands for project discovery and verification.
  • Evidence includes running git rev-parse --show-toplevel to identify the repository base path.
  • Evidence includes running make lint && make test (or project equivalent) to verify that tasks are correctly implemented.
  • [INDIRECT_PROMPT_INJECTION]: Ingests and processes documentation files from a local Obsidian vault, which creates a vulnerability surface for injected instructions.
  • Ingestion points: Reads PRD (prd.md), Tech Spec (tech-spec.md), and Task lists (tasks.md) from the directory defined by the $OBSIDIAN_AI_VAULT environment variable.
  • Boundary markers: The skill instructions do not define explicit delimiters or 'ignore embedded instructions' warnings for content ingested from the vault files.
  • Capability inventory: The skill has permissions to write to the local filesystem (code repository and vault) and execute arbitrary shell commands via the build system (e.g., make).
  • Sanitization: No evidence of sanitization, filtering, or validation is applied to the content read from the vault files before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 02:27 PM
Security Audit — agent-trust-hub — ai-execute-task