agentleFS
autsachiautsachi/codex-skills/skills/repo-next-task/SKILL.md

repo-next-task

Skill0 starsChanged 3 months agoMITView on GitHub

Inspect a software repository and recommend the most valuable next implementation task from current state, recent commits, roadmaps, TODOs, and issue context. Use when the user asks what to do next, asks to inspect the next task, requests backlog prioritization, or says phrases such as "ดูงานถัดไป", "ควรทำอะไรต่อ", or "next task". This is a read-only planning workflow unless the user separately asks to implement.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    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:

    Read applicable repository instructions such as AGENTS.md.

    line 12
  2. Distinctive:

    Read recent commit subjects and enough diff or commit detail to understand the work just completed.

    line 14
  3. Distinctive:

    Use a connected issue tracker only when it is available and materially improves the answer.

    line 16
  4. Distinctive:

    Treat dirty files as user-owned.

    line 18
  5. Distinctive:

    Prefer finishing an in-progress milestone before starting the next one unless evidence supports a deliberate switch.

    line 25
  6. Distinctive:

    Prefer a small task that can be completed and verified over a broad theme with unclear boundaries.

    line 39
  7. Distinctive:

    Do not edit files, create issues, change branches, or begin implementation.

    line 54

≈ “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

Repository Next Task

Recommend one concrete next slice using evidence from the repository rather than a generic backlog guess.

Gather current state

  1. Read applicable repository instructions such as AGENTS.md.
  2. Inspect the current branch, upstream state, working tree, and staged or unstaged changes.
  3. Read recent commit subjects and enough diff or commit detail to understand the work just completed.
  4. Locate the repository's planning sources, such as a roadmap, milestones, TODO files, README status, changelog, or issue references.
  5. Use a connected issue tracker only when it is available and materially improves the answer. Prefer local sources for a local repository question.

Treat dirty files as user-owned. Identify whether they are active work, documentation drift, or unrelated changes; never edit, stage, or discard them in this workflow.

Reconcile plans with reality

  • Detect roadmap items that are marked pending even though recent commits implemented part of them.
  • Detect implementation that is blocked by an unfinished prerequisite, migration, configuration decision, or release gate.
  • Distinguish the current milestone from later attractive work.
  • Prefer finishing an in-progress milestone before starting the next one unless evidence supports a deliberate switch.
  • Separate user-facing work from maintenance, release metadata, and optional someday items.

Rank candidates

Evaluate candidates by:

  1. ability to complete or unblock the current milestone;
  2. user value and frequency of the affected workflow;
  3. dependency and regression risk;
  4. size of the smallest independently verifiable slice;
  5. strength of repository evidence;
  6. availability of a proportional verification path.

Prefer a small task that can be completed and verified over a broad theme with unclear boundaries.

Recommend

Lead with one primary task. Include:

  • the intended outcome;
  • why it is next, citing repository evidence;
  • what recent work already covers;
  • the exact first slice and the main surfaces it touches;
  • risks or user changes that must be preserved;
  • the smallest useful verification plan.

Optionally list at most two follow-up tasks in order. State when roadmap status itself needs reconciliation.

Do not edit files, create issues, change branches, or begin implementation. If the user later says to proceed, treat that as a separate implementation request and re-check the working tree first.

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.