The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external information such as codebase facts, web research, and external research dossiers. This creates a surface for indirect prompt injection attacks where malicious instructions could be embedded in the data processed by the agent.
Ingestion points: The skill consumes codebase facts, source evidence, web research evidence, and research dossiers (SKILL.md).
Boundary markers: The instructions do not specify the use of delimiters or boundary markers to isolate external data from the agent's core instructions.
Capability inventory: The skill utilizes omh CLI tool calls to record plan artifacts, mark plans as accepted, and track runtime status (SKILL.md).
Sanitization: No explicit sanitization or validation of ingested external content is described before it is used to shape planning options and tradeoffs.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell-based commands using the omh tool suite to manage the planning lifecycle.
Evidence: Instructions include running omh runtime record, omh hermes plan --record, and omh hermes plan-accept to create and update planning artifacts within the repository structure.

ulw-plan