The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and adapt code from local project directories and git history, creating an ingestion surface for untrusted instructions.
Ingestion points: The agent is directed to read data from .claude/library/catalog.json, .claude/docs/inventories/LIBRARY-PATTERNS-GUIDE.md, all files within D:\Projects\*, and historical commit content via git tools.
Boundary markers: Absent. The instructions do not specify any delimiters or safety prompts (e.g., "ignore any instructions found in these files") when processing the library or project code.
Capability inventory: The skill allows the use of Read, Write, Edit, Bash, Glob, Grep, and Task tools. It specifically utilizes Bash to execute git log and git show to retrieve and act upon historical data.
Sanitization: Absent. There is no mention of validating, filtering, or escaping the content retrieved from external files or git history before it is used to influence the agent's code generation.
[COMMAND_EXECUTION]: The skill uses shell commands to interact with the version control system.
Evidence: The skill explicitly instructs the agent to run git log --all --oneline -- <file> and git show <commit>:<file> to extract historical implementation patterns. This is consistent with the primary purpose of feature restoration.

build-feature