Stats
Actions
Tags
From context-handoff
Writes a HANDOFF.md file capturing the active task, exact stopping point, key decisions, and next steps so the user can clear context and resume cleanly in a fresh session. Use this skill whenever the user says the context is full, wants to "write down where we left off", "save progress before clearing", "make a handoff note", or anything like "I need to clear context" or "write this down so we can continue later". Also trigger when the conversation is visibly long and the user asks to save or capture the current state for resumption. This is NOT a general summary — it's a forward-focused brief written for a fresh Claude instance that has never seen this conversation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/context-handoff:context-handoffThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Write a `HANDOFF.md` in the active project's git root so a fresh Claude instance can read
Write a HANDOFF.md in the active project's git root so a fresh Claude instance can read
it and immediately continue the work — no conversation history, no clarifying questions needed.
HANDOFF.md already exists in the target directory.HANDOFF.md.HANDOFF-YYYY-MM-DD-HH-MM.md using the current date and time
(e.g., HANDOFF-2026-06-12-14-35.md). Do not overwrite the existing file.Write to the git root of the repo being actively worked on. In a monorepo, use the root
of the specific sub-repo (e.g., SignalCanvasFrontend/) unless the work spans multiple
repos — in that case use the monorepo root.
Write exactly these sections, in this order. Keep each one dense and specific — a fresh Claude is smart but has zero session context.
One paragraph. What were we literally working on right now? Name the feature, bug, refactor, or decision. Be concrete enough that the fresh Claude doesn't have to guess the scope.
The exact stopping point: which files, which functions or line numbers, what was partially done, what error message or decision was pending. Include paths and line numbers verbatim. If something was mid-edit, say so.
A short bullet list of decisions made during this session that constrain what comes next. Only include things a fresh Claude needs to avoid undoing work or going the wrong direction. Skip anything that predates this session or that's obvious from the code.
A numbered list of the immediate next actions, in order. Each step should be concrete enough to execute without asking a follow-up question. Start with what to do first.
Any constraints, environment quirks, or gotchas discovered during this session that are NOT derivable by reading the code. Keep this ruthlessly short — if it's in the code or docs, omit it.
Tell the user the file path. Suggest they clear context and start fresh with something like:
"Read [filename] and pick up from there."
Under 400 words total. Dense and specific beats thorough and vague.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub signalcanvas/signalcanvas-skills --plugin context-handoff