The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted data such as user-provided development requirements, existing source code, and system failure logs. This creates a potential surface for indirect prompt injection if the processed content contains malicious instructions intended to manipulate the agent's behavior.
Ingestion points: The skill instructions in SKILL.md and references/source-notes.md suggest analyzing 'existing code', 'interfaces', or 'failure records' as context.
Capability inventory: The skill itself does not include scripts or specific tool calls, but the agent's environment may provide file-system or network access.
Boundary markers: The skill does not explicitly define delimiters for external content but relies on the 'scope card' structure to organize data.
Sanitization: No explicit sanitization or escaping mechanisms are defined for the interpolated data.
[EXTERNAL_DOWNLOADS]: The skill provides informational links to the 'Shape Up' book hosted on Basecamp's official website. These references are used for educational purposes and do not involve executable code or automated downloads.

shape-up