agentleFS
uwuclxdyuwuclxdy/agenticat/skills/todo/SKILL.md

todo

Skill7 starsChanged 18 days agoMITView on GitHub

Authors and normalizes tasks into `docs/todo.md`.

37 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use standup, update-docs 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.
  • AnthropicSkill

    standup

    Generate a standup update from recent activity. Use when preparing for daily standup, summarizing yesterday's commits and PRs and ticket moves, formatting work into yesterday/today/blockers, or structuring a few rough notes into a shareable update.

    Also in 3 projects that use this

  • Hugging FaceSkill

    update-docs

    Update documentation across the repo after API changes. Finds stale references in docs, examples, docstrings, and fixes them.

    Also in 3 projects that use this

  • sunnyclinches44Skill

    explain

    Explains code using real-world analogies and ASCII diagrams

    Also in 3 projects that use this

  • Hugging FaceSkill

    implement

    Make tests pass. Invoke after /write-tests produces failing tests.

    Also in 5 projects that use this

  • Google CloudSkill

    triage

    Gather context on a change or bug and plan the implementation.

    Also in 4 projects that use this

  • photon-groveSkill

    unslop

    Detects and rewrites generic, overly polished, or AI-sounding prose while preserving meaning. Use when the user asks to "unslop", humanize, de-AI, tighten, or review whether writing sounds generated.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 34 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When the user dumps loose notes ("inconsistent cursor..", "remove j/k"):

    line 60
  2. Distinctive:

    Read the relevant code first.

    line 62
  3. Distinctive:

    Ask clarifying questions for anything ambiguous (batch them, don't trickle).

    line 63
  4. Distinctive:

    Never delete an unresolved or in-progress task without the user saying so.

    line 70
  5. Distinctive:

    When asked to "resolve the todo", route execution through your multi-task runner or workflows (one task each).

    line 71

“Common” is measured: kept by at least 6 in 10 of the 34 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

Todo

docs/todo.md is the per-repo task inbox.

Core Rule

Every task must be executable by a fresh agent with no session context. Self-contained: current behavior, expected behavior, and a verify step you watched fail once on an input where it must fail. If a task needs a decision the user hasn't made, AskUserQuestion before writing it down (AskUserQuestion is Claude Code's question tool; other harnesses ship their own native tool: opencode's question, gemini-cli's ask_user, Codex CLI's request_user_input. Use it if present, else fall back to a plain numbered message). Never park open questions inside a task.

Tasks say what needs to change and, sometimes, how. Never where in the code: no file:line into source, no path that only locates code. A path stays where the file is the task's subject (an instruction file, a hook, a config the task edits), since there the path is the identity.

State observed behavior as fact; label attributed causes as hypotheses. A task's "current behavior" is what you saw or reproduced. Any "why" you did not verify (an engine's timing, an API's state model, a component's intent) must read as a hypothesis (hypothesis:, suspected:), never as fact. Baking an unverified root cause into a task sends the executing agent to fix the wrong thing (a retry built on an inferred "the API hasn't synced" missed that the API never stored the data at all). If the cause is load-bearing for the fix, the verify step must confirm it first against the vendor source (swagger/docs/code) or a repro before the fix is built on it.

Size tasks by scope. Combine small related nits into one task; split anything too big for a single reviewable change into independent tasks. The unit is "one agent, one clean commit".

Two Formats, Picked by Source

Audit Findings / Bugs -> Punch-List

For audit results, contract violations, bug reports:

# <audit name> punch-list

<one-line provenance: what was audited against what, date>

## <component / feature area>

- [ ] <current behavior> -> <required behavior>. <optional how. ordering note if other items cascade from it>
  • Group by component or feature area, not by severity.
  • Note cascade order explicitly ("fix first: X findings depend on it").

Feature Ideas / Specs -> Checkbox Spec

For new features, enhancements, raw idea notes:

## <N>. <feature title>

<one-line intent if the title doesn't carry it>

### <surface: Backend / Frontend / TUI / CLI / Schema>
- [ ] <concrete, verifiable step>
  • Number tasks; subsection per surface touched; one reviewable change per checkbox.

Normalizing Raw Notes

When the user dumps loose notes ("inconsistent cursor..", "remove j/k"):

  1. Read the relevant code first. Resolve each note to specific files and current behavior.
  2. Ask clarifying questions for anything ambiguous (batch them, don't trickle).
  3. Rewrite as tasks in the matching format above. The original note text can be dropped once the task captures it fully.

Resolving

  • docs/todo.md holds OPEN work only. A landed item gets propagated to where it can't rot (a test that fails when the behavior changes, else the owning docs/ page or CLAUDE.md), then deleted from the list instead of left as a checked-off record. A ship narrative parked in the backlog drifts against the code it describes; nothing re-reads it to notice.
  • A blocked task's gate names the OBSERVABLE it waits on, never the event. One written as "has the export arrived" cleared on delivery across two sessions and the delivered file was useless both times, because what was actually needed was a populated download link inside it. Phrase every gate as a check someone can run and get a yes/no from.
  • Never delete an unresolved or in-progress task without the user saying so. Deletion is reserved for a task you have confirmed landed and propagated; it is never a way to prune work you chose to skip.
  • When asked to "resolve the todo", route execution through your multi-task runner or workflows (one task each); this skill only owns the file.

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.