The Agent Skills Directory

[COMMAND_EXECUTION]: The workspace skills utilize standard CLI tools for development and synchronization tasks.
Evidence: skylar-start-here/SKILL.md executes pnpm and git commands; github-sync/SKILL.md uses the gh (GitHub) CLI.
[EXTERNAL_DOWNLOADS]: Development setup and synchronization skills perform remote downloads from trusted registries and repositories.
Evidence: skylar-start-here/SKILL.md runs pnpm install and git submodule update. These target the official pnpm registry and the project's own submodules.
[DATA_EXFILTRATION]: Various skills aggregate workspace data and transmit it to external productivity platforms. All destinations are well-known technology services.
Evidence: notion-sync/SKILL.md (Notion), linear-sync/SKILL.md (Linear), activity-reporter/SKILL.md (Slack, Gmail, HubSpot). These operations are neutral and serve the primary purpose of the workspace.
[PROMPT_INJECTION]: The system processes untrusted data from external sources, which constitutes an indirect prompt injection surface.
Ingestion points: activity-reporter/SKILL.md processes content from Slack, Gmail, GitHub PRs, and Linear issues; research-analyst/SKILL.md analyzes call transcripts and research notes.
Boundary markers: The skills primarily use structured JSON output formats and clear section headers to demarcate untrusted data from instructions.
Capability inventory: The agent can execute commands (gh, pnpm, git) and perform network writes to Slack, Notion, Linear, and HubSpot.
Sanitization: Explicit sanitization or filtering logic for external content is not visible in the provided instruction files.

visual-digest