The Agent Skills Directory

[SAFE]: The skill's core functionality is to read local developer notes and repository context to populate a markdown template, which is a standard and safe behavior for coding assistants.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local files such as explore.md and debug.md, which could potentially contain malicious instructions. This is an ingestion surface for indirect prompt injection; however, because the skill's purpose is to summarize technical content for developer review, the risk is minimal.
Ingestion points: explore.md, debug.md, state.md, and various files within the repository during exploration.
Boundary markers: No explicit delimiters or boundary instructions are defined to separate ingested content from the system prompt.
Capability inventory: File reading access to the repository and file writing access to the .lcs/work-items/ directory.
Sanitization: No explicit sanitization or filtering of input content is performed before synthesis.

lcs-toprd