The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access commands were detected within the skill's instructions or example files.
[INDIRECT_PROMPT_INJECTION]: The skill instructions define a process for ingesting and interpreting external data (prior recaps, stakeholder summaries, and project documents) which constitutes an indirect injection surface. However, the risk is negligible as the skill possesses no dangerous capabilities such as network access or shell execution.
Ingestion points: Reads meeting topics, attendees, prior recaps, stakeholder summaries, and project documents discovered via project or topics frontmatter.
Boundary markers: The skill does not implement specific delimiters or instructions to ignore embedded prompts within the ingested data.
Capability inventory: The skill is limited to reading files and generating text. It has no capabilities for network operations, subprocess execution, or writing to the filesystem (confirmed by the "Writes: nothing" clause in the Project Memory Contract).
Sanitization: No validation or sanitization of the content from ingested artifacts is performed prior to processing.
[DATA_EXPOSURE]: The skill accesses project-specific memory and local artifacts (.claude/pm-skills.local.md) to ground its analysis. This behavior is consistent with the skill's stated purpose of strategic preparation and does not involve exfiltration to external domains.

foundation-meeting-brief