The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates code generation by reading existing local files via the --reference argument. This creates a surface where data processed by the tool could potentially influence the agent's output if the reference files contain embedded instructions. • Ingestion points: Reference files provided to the code-write script as seen in SKILL.md. • Boundary markers: None mentioned in the usage documentation. • Capability inventory: File system writing via the --target argument. • Sanitization: No sanitization or filtering of reference file content is described in the skill instructions.
[SAFE]: The skill utilizes a local utility script defined within the plugin root directory and does not initiate unauthorized network connections, remote downloads, or credential access.

code-writer