The Agent Skills Directory

[SAFE]: The skill defines orchestration logic for managing multiple local project instances ('squads'). It instructs the agent on how to bootstrap new environments and delegate tasks using established protocols.
[SAFE]: Disambiguation instructions for terminology ('squad' vs. 'group of agents') are standard operational guidance for the agent and do not attempt to bypass safety filters or ignore ethical constraints.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data by reading .squad/manifest.json files from other repository paths (e.g., ../friend-repo).
Ingestion points: The skill reads manifest files via squad registry add and squad discover commands in SKILL.md.
Boundary markers: The skill does not explicitly mention sanitization of manifest fields, but treats them as structured JSON data for capability discovery.
Capability inventory: The skill uses git init, squad init, gh issue create, and gh issue view (documented in SKILL.md).
Sanitization: The skill provides explicit warnings against sharing authentication credentials or secrets in cross-squad communication, which serves as a manual sanitization guideline for the agent.
[COMMAND_EXECUTION]: The skill utilizes legitimate developer tools including git, gh (GitHub CLI), and the squad CLI tool (a vendor resource from 'sbroenne') to perform repository management and issue tracking. These operations are within the expected scope of a project orchestration skill.

cross-squad