The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes git blame via subprocess.run in scripts/lib/utils.py to identify the creator of session history files. This execution is restricted to the local filesystem and specific project files.
[PROMPT_INJECTION]: The skill processes untrusted text from .specstory/history markdown files, creating a surface for indirect prompt injection.
Ingestion points: Session messages are extracted from markdown files in scripts/lib/parser.py.
Boundary markers: None identified; the parsing logic does not isolate message content with delimiters.
Capability inventory: The skill performs file reads and executes the git command.
Sanitization: No content sanitization is applied to the extracted text before it is summarized by the LLM.
[EXTERNAL_DOWNLOADS]: Installation instructions in scripts/lib/utils.py point to the official SpecStory GitHub repository and documentation, which are trusted sources for this specific utility.

specstory-yak