The Agent Skills Directory

[SAFE]: The skill provides structured guidance for SaaS product launches. No patterns related to prompt injection, data exfiltration, or malicious execution were found in SKILL.md or the evaluation suite.
[INDIRECT_PROMPT_INJECTION]: The skill utilizes local project context files to refine its output, which is a common pattern for tailored AI assistance. While this introduces a theoretical surface for indirect prompt injection, the risk is negligible due to the advisory nature of the skill and the expected user control over the context files.
Ingestion points: Reads .agents/product-marketing.md, .claude/product-marketing.md, or product-marketing-context.md as described in SKILL.md.
Boundary markers: None explicitly defined in the instructions.
Capability inventory: Generative advice only; no high-risk tools or command execution capabilities were identified in the provided skill scripts.
Sanitization: None explicitly defined for the ingested file content.

launch