The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes external project artifacts such as PRDs, Epics, Stories, and Architecture documents to generate change proposals. This creates a surface where malicious instructions embedded in those documents could influence the agent's behavior during the impact analysis phase.
Ingestion points: Processes content from PRD, Epics/Stories, and Architecture files as specified in Step 2 of SKILL.md.
Boundary markers: No clear delimiters or instructions to ignore embedded commands are used when interpolating these artifacts into the analysis context.
Capability inventory: While this skill primarily generates text-based proposals, the output is intended to drive subsequent backlog reorganization and document updates by other agents.
Sanitization: Input content from external artifacts is not sanitized or validated before being analyzed and reflected in the proposal.
[DYNAMIC_EXECUTION]: The documentation includes a template for dynamic code execution using node -e to perform file system operations.
Evidence: references/document-change-log.md provides a bash snippet that executes an inline JavaScript block to read and write files using fs.readFileSync and fs.writeFileSync.
Impact: The use of node -e to execute dynamically generated script strings with interpolated arguments (like file paths and descriptions) is a known vector for potential command or script injection.
[EXTERNAL_DOWNLOADS]: The skill references several external resources and source files hosted on the author's GitHub repository.
Evidence: Links in references/document-change-log.md point to https://github.com/Gamaroff/agent-skills/blob/develop/shared/resources/change-log.js and other configuration documents.
Impact: These are documented as vendor-specific resources for the change log engine and originate from the author's official repository.

change-management