The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes user-provided research artifacts from docs/research/ and roadmap files from docs/roadmap/, which constitutes a standard indirect prompt injection surface. However, the skill implements a 'Lead-as-Skeptic' protocol where the orchestrator must challenge and verify all agent-generated ideas against research evidence before publication, mitigating the risk of malicious data influence.
Ingestion points: docs/research/ (research-findings artifacts) and docs/roadmap/ files.
Boundary markers: The skill enforces output conformance to the docs/templates/artifacts/product-ideas.md template.
Capability inventory: The skill can spawn sub-agents (Agent tool), create tasks (TaskCreate), and write to restricted progress and artifact directories.
Sanitization: Uses a skeptic review mechanism and a max-iteration limit for rejections before human escalation.

ideate-product