The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data, such as meeting transcripts or 'rough ideas', which could contain malicious instructions designed to influence the agent's behavior. * Ingestion points: Intake phase in SKILL.md accepts arbitrary text or output from the external meeting-notes skill. * Boundary markers: The skill uses structured markdown templates (templates/ticket.md, templates/project-brief.md) to isolate processed content. * Capability inventory: The skill has write access to a Linear workspace via MCP tools (save_project, save_milestone, save_issue). * Sanitization: The skill mandates a 'Preview & approve' gate (Phase 4) in SKILL.md requiring explicit user confirmation before any tool execution, providing a strong defense against automated prompt injection attacks.
[COMMAND_EXECUTION]: The skill invokes several Linear MCP tools to create and manage project data. * Evidence: Scripts in rules/linear-mcp-contract.md and rules/dependencies-and-search.md utilize tools like list_teams, save_project, and save_issue. * Context: These commands are core to the skill's functionality and are protected by a strict create-order protocol and mandatory user review.

linear-planning