resume
Skill4 starsChanged 2 months agoApache-2.0View on GitHub
Rehydrate context from the latest numbered handoff file written by /handoff and continue the work. Picks the highest handoff number in the chain. Use at the start of a fresh session, after clearing or compacting, or when the user says "resume", "pick up where we left off", "continue from the handoff", or points at a handoff file. Pairs with /handoff.
64 other projects use a version of this.
Adopted by
Most recently taken up by
- optiflowic/keyturntheir own version, first seen Nov 16, 2026
- theraihanrakibb/agent-skillstheir own version, first seen Nov 16, 2026
- DeepExtrema/Tesseracttheir own version, first seen Nov 16, 2026
- kele-dev/Claude-Orchestrator-Skilltheir own version, first seen Nov 16, 2026
- netarz/gisoo-skills-registrytheir own version, first seen Nov 15, 2026
- kuldeep-jadeja/claude-handoff-plugintheir own version, first seen Nov 15, 2026
Used with
Projects with it also use pause, status and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
kennethleungtySkillpause
Pause — stop the music (alias for /stop)
Also in 7 projects that use this
AnthropicSkillstatus
Case status summary by audience — client-facing (plain language), internal (for the professor), or court-ready (formal caption format per local rules). Same facts, different framing and depth. Use when a student needs to update the client, brief the professor, or prepare a court status report.
Also in 15 projects that use this
MicrosoftSkillreport
Turn an exploration (threads, findings, charts) into a single Markdown report — note, blog post, executive summary, KPI dashboard, slide brief, or multi-section analytical report, with embedded charts.
Also in 7 projects that use this
MicrosoftSkillconfig
Show or change TokenEconomizer's settings: turn micro-skills, tool pruning or the loop guard on or off, or tune them, for this repository or (with --global) every repository. Use when the user asks to configure, enable or disable TokenEconomizer or one of its mechanisms.
Also in 3 projects that use this
sahinSkillexecute
Also in 4 projects that use this
u9401066Skillbug-fix
Structured bug fixing workflow with root cause analysis and regression prevention. Triggers: BF, 修 bug, fix bug, debug, 除錯, 錯誤, error, 問題, issue, 修復, fix, 故障, fault, 異常, exception, 失敗, failure, 不能用, broken, 壞掉.
Also in 3 projects that use this
- +2 more from microsoft
Other versions64 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 64
- H-mmer/pentest-agents★ 815same name
- bitwize-music-studio/claude-ai-music-skills★ 668same name
- qbeka/jev-job-search★ 368same name
- thesun4sky/jobstack★ 216same name
- EmeaAppGbb/spec2cloud★ 100same name
- saksham10arora-dotcom/everything-claude-codex★ 96same name
- thiagofinch/mega-brain★ 95same name
- nimadorostkar/Claude-Skills-collection★ 37same name
- junjiezhou1122/ideascientist-claude-code★ 35same name
- wrg32786/aigent-os★ 26same name
54 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository19 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- autonomous-prepSkill
- autonomous-sessionSkill
- backlogSkill
- check-buildSkill
- copilot-test-setupSkill
- daily-reviewSkill
- e2e-loopSkill
- handoffSkill
- harness-auditSkill
- harness-refreshSkill
- night-briefingSkill
- openspec-apply-changeSkill
- openspec-archive-changeSkill
- openspec-exploreSkill
- openspec-proposeSkill
- openspec-sync-specsSkill
- review-iterationSkill
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 2 are distinctive, and 5 are very specific to this project.
0
Common
Most of the 56 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 19
If a path was given as an argument, use it.
- Very specific: line 29
If a LATEST.md pointer exists, read it as a cross-check, not as the answer.
- Distinctive: line 45
If no handoff exists anywhere, say so plainly and ask what the user wants to work on.
- Very specific: line 70
Check what the previous session left running: Read ## Environment left running and confirm each entry is still in the state the handoff …
- Very specific: line 85
Read the files under ## Read on resume, plus any file that next-step-1 will edit.
- Very specific: line 93
Before touching anything, internalize ## Constraints and user directives from the handoff, and read <handoff-dir>/DEAD-ENDS.md if it exists.
- Very specific: line 121
When this resumed session in turn fills up, run /handoff again.
“Common” is measured: kept by at least 6 in 10 of the 56 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Restore working context from a durable handoff file, verify it still describes reality, and continue.
Usage: /resume (loads the highest-numbered handoff) or /resume <path-to-handoff.md>.
Agent-agnostic: the handoff is plain markdown, so a session started by one CLI agent can be
resumed by another. The handoff's **Agent:** field records the model, the CLI and the
posture that wrote it — note it if the tooling assumptions differ from yours, and say so out
loud when the model changed, since that is the most common reason the previous session's style
or judgment does not match what you would have done.
Step 1 — Locate the latest handoff
If a path was given as an argument, use it. Otherwise list every known location — the previous session may have run under a different agent:
ls .handoffs/handoff-*.md .claude/handoffs/handoff-*.md .agents/handoffs/handoff-*.md 2>/dev/nullThe highest NNNN is the newest handoff. Numbering is the authority: file mtimes do not
survive a clone, a copy, or a checkout.
If a LATEST.md pointer exists, read it as a cross-check, not as the answer. If it
disagrees with the highest number, say so and prefer the number — a stale or unreachable
pointer is a known failure mode, especially where the pointer is tracked in version control
and the handoff it names is not.
Then read that file. If you need more history, follow its **Previous handoff:** link — only
as far as the current next-step requires, usually zero or one hop.
Two things to flag rather than silently absorb:
- Handoffs in more than one directory → the chain has been split, and the newest file may not be the newest work. Reconcile before trusting either.
- A gap in the numbering, or handoffs that stop partway → check whether the missing ones are untracked rather than absent. A partially tracked chain hands a clone the newest tracked handoff, which reads as entirely legitimate while being weeks stale.
If no handoff exists anywhere, say so plainly and ask what the user wants to work on. Do not invent context or guess from git history alone.
Step 2 — Verify the handoff against reality
A handoff is a snapshot; the tree may have moved. Check before trusting it:
git branch --show-current
git log --oneline -5
git status --shortCompare against the handoff's ## Git state:
- Same branch, same HEAD, same dirty files → trustworthy, proceed.
- HEAD moved forward → someone (possibly another session, possibly another agent)
committed since. Run
git log <handoff-sha>..HEAD --statand reconcile: the handoff's "next steps" may already be done. Flag anything that no longer applies. - Different branch → say so before doing anything. Confirm which branch the work belongs on.
- Files listed as touched are now clean/absent → the work was committed, stashed, or reverted. Determine which; do not assume.
Also verify the paths in ## Files touched and ## Read on resume still exist.
Check what the previous session left running. Read ## Environment left running and
confirm each entry is still in the state the handoff claims — a server it started may have
died, or may still be holding the port you are about to bind. Starting a second one is the
common failure. An unattended job left enabled is the expensive one.
Verify claims, do not inherit them. A handoff records what the previous session believed
it had finished. Where the next step builds on that, confirm the work still functions rather
than trusting the claim — especially anything the handoff's own ## Verification section
lists as not tested.
State any drift you found in one or two lines. Silent reconciliation is how a resumed session redoes finished work.
Step 3 — Reload only what the next step needs
Read the files under ## Read on resume, plus any file that next-step-1 will edit.
Do not bulk-read everything the handoff mentions. The point of a fresh window is that it is fresh — refilling it with context you will not use recreates the exact problem the handoff solved. Load lazily; you can always read more later.
Step 4 — Re-establish the guardrails
Before touching anything, internalize ## Constraints and user directives from the handoff,
and read <handoff-dir>/DEAD-ENDS.md if it exists.
These are what a fresh window loses most damagingly: without them a resumed session cheerfully violates a stated rule or re-pays for a known failure. Treat quoted user directives as still binding — they were not withdrawn, just forgotten.
DEAD-ENDS.md is the durable ledger; the handoff's own ## Dead ends section holds only what
that session hit. Read both, and verify before trusting a ledger entry that carries an old
date — a stale entry is worse than a missing one, because it is believed.
Step 5 — Confirm and continue
Report back, briefly:
- Where we are — one or two sentences from
## Current state. - Drift — anything that changed since the handoff, or "tree matches the handoff".
- Environment — what the handoff left running and whether it is still up.
- Constraints still in force — the quoted directives, condensed to a list.
- Next action — restate next-step-1 concretely, as you are about to do it.
Then do it. If the handoff has entries under ## Open questions for the user, or if drift
means the recorded next step no longer makes sense, ask before starting work rather than
proceeding on a guess. A question the handoff says has gone unanswered across several sessions
should be put to the user directly now, not carried forward again.
Chaining
When this resumed session in turn fills up, run /handoff again. It will read this handoff
and carry forward what is still true — so state is always re-derived from the live session
plus a durable file, never from a summary of a summary. That is what stops quality decaying
across a long chain of sessions, and it holds across agent switches too.
Durable tool and environment facts do not ride the chain at all: they go to DEAD-ENDS.md
once and stay there.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

