The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill's bootstrap script in references/full-workflow.md dynamically searches for and executes a JavaScript file (cli.js) located within a plugin cache directory ($HOME/.codex/plugins/cache/sisyphuslabs/omo/...). This represents dynamic loading and execution from computed paths.
[COMMAND_EXECUTION]: The skill frequently executes shell commands via the omo CLI and provides a wrapper function in the bootstrap script to facilitate this execution.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install dependencies using npx lazycodex-ai install if the required CLI is not found. While this is a user-facing instruction, it references a third-party package.
[INDIRECT_PROMPT_INJECTION]: The skill ingests arbitrary user-provided 'briefs' which are then used to generate instructions for subagents.
Ingestion points: The create-goals command accepts a --brief or --brief-file argument (documented in references/full-workflow.md).
Boundary markers: The skill uses structural markers like TASK:, DELIVERABLE, SCOPE, and VERIFY to delineate subagent tasks, though these are not cryptographic boundaries.
Capability inventory: The skill has the ability to spawn subagents, execute shell commands, and perform git operations via git-master.
Sanitization: There is no evidence of explicit sanitization or filtering of the input brief before it is processed and passed to subagents.

ulw-loop