The Agent Skills Directory

[SAFE]: The skill's primary function is text processing and template generation based on local project data. It follows a structured methodology (Demo2Win) to generate documentation.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses project-specific files such as discovery.json and opportunity-profile.json. However, these accesses are restricted to the local workspace context for the purpose of script generation. No network tools (e.g., curl, wget) are requested in allowed-tools, and no network exfiltration patterns were found in the instructions.
[REMOTE_CODE_EXECUTION]: There is no evidence of remote code downloads, package installations, or dynamic execution of untrusted scripts. The skill relies entirely on static markdown templates and structured JSON data.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple JSON sources and interpolates it into the output. While this creates a potential surface for indirect injection if the input JSON contains malicious instructions, the risk is mitigated as the skill's output is purely documentation (Markdown) and the agent has no dangerous capabilities (like shell execution or network access) enabled within this skill's scope.

weave-script