The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it incorporates untrusted data from the local project environment into its reasoning process.\n
Ingestion points: The skill reads project context from git log, CLAUDE.md, package.json, and conversation history (SKILL.md, Step 1).\n
Boundary markers: Absent. There are no specific delimiters or instructions used to prevent the agent from following commands that might be embedded in the context data.\n
Capability inventory: The skill has access to Bash, Read, Grep, and Glob tools (SKILL.md YAML frontmatter).\n
Sanitization: Absent. The skill does not validate or filter the content of the project files it reads.\n- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard git operations (status, rev-parse, log, diff) for context gathering. It also executes npx windags to perform planning operations and write local history triples.\n- [EXTERNAL_DOWNLOADS]: The skill uses npx:windags as an allowed tool, which can result in the download of the windags package from the npm registry if it is not present in the local cache.

next-move