The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to capture and organize knowledge into persistent storage (Obsidian, Notion, local vaults) that is explicitly intended to be read by future AI agent sessions.
Ingestion points: User input during the 'Design Interview' and automated collection of session metadata (plan records, verification gates) as described in SKILL.md and references/handover-artifacts.md.
Boundary markers: While the skill advises marking stale knowledge and stripping credentials, it does not instruct the agent to use specific delimiters or 'ignore embedded instructions' warnings when reading from the resulting wiki pages.
Capability inventory: The skill facilitates writing markdown files to local or connected cloud storage and reading from runtime todo records.
Sanitization: The 'Safety rules' section in SKILL.md explicitly mandates stripping credentials and personal data before writing to shared stores.
[COMMAND_EXECUTION]: The skill uses platform-specific CLI commands to track and record its execution state.
Evidence: SKILL.md contains a runtime evidence block executing omh runtime record. references/handover-artifacts.md references omh runtime todo show and omh_todo actions.
Context: These commands are associated with the 'oh-my-hermes' (OMH) platform ecosystem, which aligns with the skill's author ('oh-my-hermes maintainers').

omh-wiki