The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to generate and update project documentation, creating a vulnerability surface where instructions embedded in source data could influence agent behavior.
Ingestion points: Files in Source-Documents/ (including Contracts/, Emails/, and Vendor Docs), Meeting-Notes/, and Research/ directories are ingested to update root and derived documents.
Boundary markers: The provided templates for meeting notes and project memory do not include delimiters or specific instructions to the agent to ignore potentially malicious embedded content in the source data.
Capability inventory: The skill has the capability to create and modify a wide range of project files and trigger other operational slash commands (e.g., /prd-generator, /arb-review).
Sanitization: There is no evidence of validation or filtering for instructions hidden within the input data processed during document synchronization and scaffolding.

project-docs