The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection by processing data from repository history and pull requests.
Ingestion points: Data is retrieved from .changelog/CHANGELOG.md, git logs, and Pull Request descriptions (SKILL.md).
Boundary markers: No explicit delimiters are used to separate ingested content from the narrative generation instructions.
Capability inventory: The skill utilizes the Bash tool for metadata collection and file opening, and the Write tool for saving outputs (SKILL.md).
Sanitization: No validation or escaping is applied to untrusted repository data before it is incorporated into the HTML narrative.
[COMMAND_EXECUTION]: Uses the local shell to inspect project history and display the final report.
Evidence: Executes git log to identify changes and open to launch the generated HTML file in the default browser (SKILL.md).
[EXTERNAL_DOWNLOADS]: Fetches the Mermaid.js library from a well-known CDN to render diagrams in the generated presentation.
Evidence: The instructions specify including a Mermaid CDN link within the generated HTML file for architecture visualization (SKILL.md).

atlas-present