What problem does it solve?
When a user opens a browser tab whose response includes a SLICC-specific RFC 8288 Link header, the agent needs to correctly process the resulting navigate lick event without accidentally executing untrusted actions or skipping required user approvals for sensitive operations.
Core Features & Use Cases
- Dual Verb Handling: Distinguishes between handoff (human-gated task handoffs from external agents) and upskill (agent-actionable skill installations from public GitHub repos) events.
- Safety Guardrails: Enforces mandatory user approval for all handoff events, prevents premature fetching of handoff target content, and avoids auto-accepting untrusted external instructions.
- Safe Inspection: Supports read-only pre-approval checking of link headers and post-approval discovery of origin capabilities via the discover command.
- Use Case: If a user opens a tab that advertises a SLICC handoff for a cross-agent coding task, this skill ensures the agent only proceeds after the user explicitly approves the external instruction, preventing unintended execution of untrusted code.
Quick Start
Use the handoff skill to process a navigate lick event for a SLICC handoff or upskill link, following the appropriate approval and installation workflow for the detected verb type.