The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to inspect the git repository and search the knowledge graph.
Evidence: SKILL.md (Step 2) uses git diff --name-only and git diff main...HEAD --name-only to identify changes.
Evidence: SKILL.md (Steps 3, 4, 5, and 6) uses grep to parse the .understand-anything/knowledge-graph.json file.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from external sources (git diffs and JSON metadata) into the agent context.
Ingestion points: SKILL.md (Steps 2, 4, and 5) retrieve content from git diff output and node summaries in .understand-anything/knowledge-graph.json.
Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore' instructions to isolate untrusted data from the system prompt.
Capability inventory: The skill has the ability to execute shell commands (git, grep) and write files (.understand-anything/diff-overlay.json).
Sanitization: Absent. The skill does not validate or sanitize the content of file paths, commit messages, or node summaries before processing.

understand-diff