The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or high-risk behaviors were found. The skill is entirely instructional.
[DATA_EXPOSURE]: The skill prompts the agent to read context from .agents/product-marketing-context.md. This is a standard project-specific metadata file and does not constitute unauthorized access to sensitive system resources.
[INDIRECT_PROMPT_INJECTION]: The skill has an ingestion point in SKILL.md where it reads external context from a local file. However, because the skill has no destructive capabilities (no subprocess calls, file-writing, or network operations), the injection surface is functionally inert. Evidence chain: 1. Ingestion points: .agents/product-marketing-context.md in SKILL.md; 2. Boundary markers: absent; 3. Capability inventory: none; 4. Sanitization: absent.

launch-strategy