agentleFS
danielrosehilldanielrosehill/Claude-Hopper/skills/start-leftover/SKILL.md

start-leftover

Skill1 starsChanged 7 months agoMITView on GitHub

Pick one or more open items from `planning/leftover-tasks.md` and spawn a fresh Claude session focused on them. Counterpart to `capture-leftover`. Use when the user says "tackle a leftover", "pick up a parked task", "start on a leftover", "what's in the parking lot", or wants to clear deferred follow-ups without polluting the current session.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    After context-gate writes a handover, user wants to tackle one parked item rather than resume the main thread

    line 14
  2. Very specific:

    If target doesn't exist, tell the user there are no leftovers to start from and stop.

    line 25
  3. Distinctive:

    If zero open entries, print:

    line 31
  4. Very specific:

    If new-claude-here isn't available or the user prefers manual launch, print the command + payload and let the user run it.

    line 107
  5. Very specific:

    If the user picked multiple items, the new session gets all picked entries in one prompt.

    line 119
  6. Distinctive:

    Don't auto-pick: Always require user confirmation, even if there's only one open item.

    line 134

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Start Leftover

Picks open entries from planning/leftover-tasks.md, lets the user choose one or a batch to work on, and spawns a clean Claude session focused on just those items. The original session keeps its context untouched.

When to use

  • User wants to address a previously-deferred follow-up
  • "What's parked?" / "let's clear some leftovers" / "tackle that auth refactor we noted"
  • After context-gate writes a handover, user wants to tackle one parked item rather than resume the main thread

Steps

1. Locate and read

repo_root=$(git rev-parse --show-toplevel 2>/dev/null) || repo_root="$PWD"
target="$repo_root/planning/leftover-tasks.md"

If target doesn't exist, tell the user there are no leftovers to start from and stop. Don't auto-scaffold here — there's nothing to scaffold for.

2. Parse open entries

Each entry begins with ## YYYY-MM-DD HH:MM — <summary>. Read the file and extract entries where Status: open. Skip in-progress and done.

If zero open entries, print:

No open leftovers in planning/leftover-tasks.md. ✓

3. Present the list

Show open entries newest-first by capture timestamp:

Open leftovers (4):

1. 2026-04-27 18:53 — Refactor auth middleware to drop legacy session-token path
   Scope: medium  |  Captured 2h ago

2. 2026-04-26 14:10 — Investigate flaky CI run on linter step
   Scope: small  |  Captured 1d ago

3. 2026-04-25 09:30 — Add OG images to blog posts
   Scope: medium  |  Captured 2d ago

4. 2026-04-23 16:45 — Migrate utils.py to typed module structure
   Scope: large  |  Captured 4d ago

4. User picks

Ask:

Pick one or more (e.g. 1, 1,3, or all). Or cancel.

  • Single number → one task focus
  • Comma-list → batch mode
  • all → batch all opens
  • cancel → exit cleanly

5. Mark in-progress

For each picked entry, change **Status:** open → **Status:** in-progress in the file. Append a marker line:

- **Started:** 2026-04-27 21:05 (session: <new-session-id-or-cwd-marker>)

This way the original session knows what's been claimed.

6. Build a focused-session prompt

Construct an initial message for the new session that includes:

  • Which leftover(s) to tackle (full entry text from the file)
  • The original repo context (path, branch)
  • A one-line directive: "Tackle the items below. When complete, mark each as Status: done in planning/leftover-tasks.md and append a one-line outcome note."

Example payload:

Continue from these parked items in this repo:

[paste the full markdown entry blocks here]

Repo: /home/user/repos/myproject
Branch: main

When you finish each, mark `Status: done` in planning/leftover-tasks.md and append a one-line outcome below the entry. If any item turns out larger than the estimated scope, ask before continuing.

7. Spawn

Delegate to new-claude-here (same plugin) with the message pre-loaded. Default form:

# in a new Konsole window at $repo_root, opens claude with -m "<the payload>"

If new-claude-here isn't available or the user prefers manual launch, print the command + payload and let the user run it.

8. Return

Print one summary line in the original session:

✓ Started: "Refactor auth middleware..." in new session (1 picked, 3 still open)

Batch behaviour

If the user picked multiple items, the new session gets all picked entries in one prompt. Tell it to handle them sequentially. The original session's leftover-tasks.md reflects all picked items as in-progress.

Cleanup after completion

The new session is responsible for marking Status: done and appending an outcome line per task. If the user reports "I finished those" in the original session and the file still shows in-progress, offer to update it from this side — but the new session should normally do it.

Edge cases

  • Same item picked twice across sessions — second pick should detect Status: in-progress, warn the user, and let them either reclaim it (force) or pick something else.
  • Capture-while-in-flight — user might capture new leftovers while the picked-up ones are running. That's fine — file is append-only at the entry level, and Status updates are line-level.
  • No new-claude-here available — print the manual launch command instead. Don't fail the workflow.

Hard rules

  • Don't reorder entries. Newest-first is presentation only; the file stays in capture order.
  • Don't auto-pick. Always require user confirmation, even if there's only one open item.
  • Don't mark done from this skill. That's the new session's job — it ensures whoever did the work confirms it landed.
  • Don't spawn a session without the explicit prompt payload. A blank new-claude-here loses the focus.

Counterparts

  • capture-leftover — adds entries that this skill consumes
  • scaffold-planning — creates the file
  • new-claude-here — actual session-spawning mechanic this delegates to

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.