The Agent Skills Directory

[SAFE]: The skill scans the local wiki/ directory and its subfolders to collect statistics on project progress and skill usage. This data gathering is confined to project documentation and does not involve accessing sensitive system files or private user data.
[SAFE]: The skill creates retrospective reports in YAML format and knowledge base entries in Markdown format. These write operations are limited to the project's local file system and do not include any persistence mechanisms or unauthorized data exfiltration.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it ingests data from various project files that may contain untrusted content. 1. Ingestion points: Files in the wiki/, wiki/requirements/, and wiki/issues/ directories, as well as design change records. 2. Boundary markers: The skill does not use specific delimiters or instructions to ignore instructions embedded in the source data. 3. Capability inventory: The skill is capable of writing new Markdown and YAML files to the project structure. 4. Sanitization: No explicit validation or sanitization of the ingested text is performed before it is processed or written to reports.

cc-retro