agentleFS
blauwtjeblauwtje/exo/skills/save/SKILL.md

save

Skill3 starsChanged 57 days agoMITView on GitHub

Writes a handoff file for unfinished work and reads it back in a fresh session. Use when the user asks to save, pause or hand off this session, or to continue from a saved handoff ('continue where I left off'). Not for finished work.

38 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use load, ingest 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.
  • TheLatentSpaceLabSkill

    load

    Load a document into memory for RAG

    Also in 4 projects that use this

  • JaimeGBASkill

    ingest

    Digest an external source (article, paper, transcript, URL, book chapter) and integrate its content into the knowledge base

    Also in 8 projects that use this

  • ArtemXTechSkill

    query

    Query data from this vault. USE WHEN user asks about projects, clients, tasks, daily notes. Use grep to extract frontmatter - do NOT read full files.

    Also in 4 projects that use this

  • szweibelSkill

    idea

    Generate a wide field of options when the user is brainstorming, exploring, expanding, or pressure-testing — not recommending or implementing

    Also in 3 projects that use this

  • GoogleSkill

    dashboard

    Configures and launches the GKE Capacity & Triage Dashboard scoped to a GCP project ID, GKE cluster name, or target ML workload ID. Uses the configure-and-launch-dashboard action and launch-workload-dashboard workflow to verify local dashboard health, discover cluster capacity, trigger live scans, and generate deep-linked URLs. Use when the user types "/dashboard", says "launch dashboard", "open dashboard", "show capacity", or "view triage dashboard" with a project, cluster, or workload. Don't use for general workload log querying or automated failure attribution (use the troubleshoot skill).

    Also in 4 projects that use this

  • Microsoft AzureSkill

    lint

    Run all linting and spell checking for TypeScript and Rust code. Run this before committing to catch style issues, clippy warnings, and misspellings.

    Also in 5 projects that use this

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 32 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run node "${CLAUDE_SKILL_DIR}/scripts/handoff.mjs" path for the location, create its parent directories, and write the file whole, …

    line 21
  2. Distinctive:

    Write plain words in the user's language, with these sections in order:

    line 22
  3. Very specific:

    When the user asks to continue saved work, run node "${CLAUDE_SKILL_DIR}/scripts/handoff.mjs" path, read the file it prints and follow its …

    line 33

“Common” is measured: kept by at least 6 in 10 of the 32 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Save

Freeze what this session knows and the next one cannot rebuild. Point to files and commands, never paste the transcript, a diff or a log, because the fresh session would pay again for that context.

Before writing

  • Push nothing and open no pull request.
  • Finish or undo the step in progress and start no new work.
  • On a plan's run branch, commit nothing: save the partial work of the task in progress with git diff > .exo/carry/task<n>-partial.patch and list it under Progress.
  • On any other non-default branch, commit every uncommitted edit as one chore: wip <topic> commit, and say in its body if the tree is broken.
  • On the default branch, leave edits uncommitted and list them under Progress.

Write the file

  1. Run node "${CLAUDE_SKILL_DIR}/scripts/handoff.mjs" path for the location, create its parent directories, and write the file whole, replacing an older handoff.
  2. Write plain words in the user's language, with these sections in order:
    • One line on where the work stands and the first thing to do next, then Written: <date>, last commit <short hash>, branch <name>.
    • ## Goal: what the work is for, and what it will not do.
    • ## Progress: from git status --short, not memory. Open with the plan path and task number when a plan runs. List what is committed, then every uncommitted path in full, because a next session that cannot tell them apart discards working-tree edits. Add each decision as "<decision>, decided by <the user or this session>".
    • ## Evidence: Proven: claim, command, result; Unproven: what is open and who can answer. Add each approach already ruled out.
    • ## Next step: the first action, then each later step with its full command. For a run, say how to wait, what to do on failure and where to stop.
    • ## Resume: copy this paragraph exactly: "Diff the work this handoff records as done against the work it still lists as next or open, name the resume point, and repeat no step it records as proven. Verify an inherited claim against the real artifact before building on it: a prior session's report is not the proof. Compare the Written date and last commit with git log -1 and git status --short, and name what changed since before relying on a claim."
  3. End the turn by naming the written path and telling the user to run /clear, then say "continue from the handoff". Continue no work.

Continue from a handoff

When the user asks to continue saved work, run node "${CLAUDE_SKILL_DIR}/scripts/handoff.mjs" path, read the file it prints and follow its ## Resume paragraph. With no file there, say which branch has none and write nothing.

Judgment

  • Confirm path, branch and commit with a command before writing them down.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.