ai-create-plan

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands like git rev-parse and ls via a Bash interface to identify project roots and manage files within the local Obsidian vault. These operations are necessary for the skill's core functionality.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses the $OBSIDIAN_AI_VAULT environment variable and the user's $HOME directory to locate the Obsidian vault for writing plan documents. While this involves reading local file paths, it does not exfiltrate data to external domains.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to perform a deep analysis of a repository's source code and documentation folders. This exposes the agent to untrusted data that could contain malicious instructions intended to influence the generation of the implementation plan.
  • Ingestion points: Source code files and documentation within the analyzed repository (SKILL.md).
  • Boundary markers: The skill does not define specific delimiters or isolation instructions to prevent the agent from obeying prompts found within the analyzed files.
  • Capability inventory: The agent can read and write to the local filesystem and execute directory listing and git commands.
  • Sanitization: There is no explicit mechanism described to sanitize or filter out instructions from the ingested repository content.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 08:55 PM
Security Audit — agent-trust-hub — ai-create-plan