From agent-kit
Use when starting a session after /handoff was run in a previous session - load .agents/handoff.md, verify git state matches what the previous session expected, print Target + Next Step verbatim, and wait for the user's "go" before doing anything.
How this skill is triggered — by the user, by Claude, or both
Slash command
/agent-kit:handoff-resumeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Invoked manually via `/handoff-resume` when the user wants to continue work captured by `/handoff` in a prior session. Load the state, verify the working tree matches, and **stop**. Do not execute the next step until the user explicitly says "go" (or equivalent).
Invoked manually via /handoff-resume when the user wants to continue work captured by /handoff in a prior session. Load the state, verify the working tree matches, and stop. Do not execute the next step until the user explicitly says "go" (or equivalent).
Read .agents/handoff.md (or the path the user provided). If missing, fall back to ./handoff.md for backwards compat; if still missing, say so and stop.
Verify git state. Single bash call, in parallel:
git rev-parse --abbrev-ref HEAD && git log -1 --oneline && git status --short
Compare branch, last commit, dirty status against the ## Current State section.
Report in this format:
=== RESUMING SESSION ===
Target: <one-line from ## Target>
Branch: <current> (handoff: <handoff value>) [MATCH | DRIFT]
Last commit: <current> (handoff: <handoff value>) [MATCH | DRIFT]
Dirty: <current> (handoff: <handoff value>) [MATCH | DRIFT]
Next step:
<verbatim content of ## Next Step>
⚠️ DRIFT DETECTED: <describe> ← only if any DRIFT
Stop. No tool calls, no edits. Wait for user "go", "yes", "continue", or similar.
.agents/handoff.md during resume.npx claudepluginhub egor-xyz/agent-kit --plugin agent-kitGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.