The Agent Skills Directory

[SAFE]: The skill performs legitimate marketing planning tasks, reading from the brand/ directory and writing to marketing/launch/. These actions are well-defined and restricted to the project's workspace.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes untrusted data from the local brand/ directory to generate its output.\n
Ingestion points: Files brand/voice-profile.md, brand/positioning.md, and brand/audience.md are read to provide context for the launch plan.\n
Boundary markers: None are present; the instructions do not include specific delimiters or warnings to ignore potentially malicious content within the brand files.\n
Capability inventory: The skill's capabilities are restricted to generating and writing text files (marketing/launch/plan.md, marketing/launch/checklist.md). It lacks access to subprocess execution, network requests, or sensitive system files.\n
Sanitization: No sanitization or validation is applied to the data ingested from the brand files.\n- [SAFE]: No instances of remote code execution, hardcoded credentials, persistence mechanisms, or obfuscated content were detected in the skill's instructions or reference files.

launch-strategy