The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Node.js script (scripts/scaffold-plan.mjs) to generate plan artifacts. Technical analysis of the script confirms it implements robust security controls, including:
Strict validation of the slug argument using a whitelist regex.
Path confinement logic that ensures writes are restricted to the .omo/ directory.
Checks to prevent directory traversal and the use of symbolic links during file operations.
[PROMPT_INJECTION]: The skill includes advanced instructions for identifying and mitigating indirect prompt injection when processing data from untrusted sources like Discord threads or external repositories. It specifically directs the agent to treat such content as "claims" rather than instructions and to verify them against the codebase.
[DATA_EXFILTRATION]: No evidence of unauthorized data exfiltration or sensitive file access was found. The skill operates within a defined workflow and restricts its durable output to the .omo/ folder within the project workspace.
[REMOTE_CODE_EXECUTION]: The skill does not perform remote downloads or execute code from untrusted external sources. The multi-agent spawning logic utilizes platform-standard protocols for subagent coordination.

ulw-plan