agentleFS
RyanDanielWalkerRyanDanielWalker/snippr/.claude/skills/next/SKILL.md

next

Skill1 starsChanged 3 days agoMITView on GitHub

Pick up the next roadmap item and take it from plan to reviewed, committed work. Stops before pushing.

  • Reads credentials

44 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use quick, project and 9 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.
  • codebycoreySkill

    quick

    Quick start a change with minimal interaction. Use when fast-tracking a straightforward change that needs proposal, design, and tasks in one step.

    Also in 3 projects that use this

  • BugRogerSkill

    project

    Project lifecycle management. Dispatches to Ada.

    Also in 3 projects that use this

  • AnthropicSkill

    status

    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 9 projects that use this

  • MicrosoftSkill

    cleanup

    Also in 4 projects that use this

  • ArvorCoSkill

    clarify

    Identify and resolve ambiguities in feature specifications. Use when spec has unclear requirements. Triggers on: clarify spec, resolve ambiguities, clarify requirements.

    Also in 3 projects that use this

  • genomewalkerSkill

    explore

    RLM-style recursive memory exploration - dynamically navigate the memory graph

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

12 rules. 0 are common, 7 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 39 edited versions we read keep these.

  • 7

    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
  1. Distinctive:

    Not on main, or uncommitted changes above: stop and ask what to do with them.

    line 28
  2. Distinctive:

    If the arguments name an item, do that one.

    line 33
  3. Distinctive:

    If every item in the current milestone is ticked, stop and say so.

    line 35
  4. Very specific:

    Read the ADRs the item links to, the code it touches, and any tests nearby.

    line 41
  5. Distinctive:

    Write tests for the logic.

    line 49
  6. Distinctive:

    No new dependencies without asking.

    line 51
  7. Very specific:

    Before every commit: npm run typecheck, npm test, npm run lint.

    line 64
  8. Very specific:

    Never push: If the commits include a migration, say under "Do before you push" that they go to a branch first, and give the commands from …

    line 65
  9. Very specific:

    Read git diff origin/main..HEAD as a skeptical reviewer who didn't write it: ownership checks on every record, error handling, edge cases, …

    line 69
  10. Distinctive:

    Keep it short and in plain English.

    line 73
  11. Very specific:

    If no group has anything, write the line Needs you: nothing.

    line 93
  12. Distinctive:

    Never put any of these only in the body below.

    line 94

“Common” is measured: kept by at least 6 in 10 of the 39 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

/next: ship the next roadmap item

You're the senior engineer. Ryan is the owner: he makes the calls, reviews every commit, and pushes. He learns the code by reading what you write about it, so the explanation is part of the deliverable, not an extra. Finish the item: no TODO(ryan), no exercise, no piece held for him to write, unless he asks for that in words (0033). Follow CLAUDE.md throughout.

Arguments: $ARGUMENTS

Where things stand

Branch: !git branch --show-current

Uncommitted changes: !git status --short

Committed but not pushed: !git log --oneline origin/main..HEAD

1. Preflight

  • Not on main, or uncommitted changes above: stop and ask what to do with them. Don't stash, reset, or discard anything.
  • Unpushed commits above: mention them in one line, then carry on. New commits stack on top.

2. Pick the item

  • If the arguments name an item, do that one.
  • Otherwise read docs/ROADMAP.md and take the first unchecked item in the earliest phase of the current milestone, the first section after the roadmap's intro. Parked items are never picked automatically.
  • If every item in the current milestone is ticked, stop and say so. Don't start on Parked work without being asked.
  • Skip items Ryan has to do himself (dashboard settings, accounts, anything outside the repo). List them in one line so they don't get forgotten.
  • Say which item you picked and why, in one sentence.

3. Understand it before touching it

Read the ADRs the item links to, the code it touches, and any tests nearby. For anything framework-specific, check node_modules/next/dist/docs/ (see AGENTS.md). Then sort the work:

  • Mechanical (cleanup, a clear bug with one obvious fix): state the plan in 3 to 5 bullets and go.
  • A real decision (a new dependency or service, a data model change, a status code or API shape, a trade-off between options): stop. Lay out 2 or 3 options in plain English, each with what it costs, recommend one, and wait for Ryan to choose. Big decisions get an ADR.

4. Build it

  • Pull logic worth testing into small functions in src/lib/, with database and network calls passed in, so tests run against fakes.
  • Write tests for the logic. After they pass, break the code on purpose once and confirm a test fails. Put it back.
  • Never touch production's database. Locally, .env points at the Docker Postgres (npm run db:up), and scripts/check-db.mjs refuses any DATABASE_URL that isn't local (0015). So the local one is fair game: npm run db:migrate for schema changes, npm run test:db when a change touches queries. Never get around the check with ALLOW_REMOTE_DB or a bare npx prisma. If verifying needs production data, say so and let Ryan decide.
  • No new dependencies without asking.
  • Product copy: direct, a little dry, human. No em dashes.

5. Keep the docs true

  • Tick the roadmap item. Add anything you discovered along the way as a new item, worded so it makes sense later. It goes under the current phase if a visitor would hit it before the current milestone is done, otherwise under Parked; say which in the report.
  • Fix docs/ARCHITECTURE.md or docs/DEVELOPMENT.md if this change made them wrong. Ryan learns what was built by reading these later, so write them to be learned from: what each piece does, why it is that way, and what was weighed against it. Every new term gets a plain definition and a small example.
  • ADRs are append-only. A new decision or a change to an old one gets a new ADR that links back. Never edit an accepted one. An ADR is as long as someone who wasn't there needs it to be: what forced the decision, each option with its real cost, what was picked, and what follows.
  • Anything you found and aren't fixing now goes on the roadmap, not into a mental note.

6. Commit

  • Small commits, one idea each, Conventional Commits. Subject says what, body says why.
  • Before every commit: npm run typecheck, npm test, npm run lint. Before the last one, npm run build too.
  • Never push. If the commits include a migration, say under "Do before you push" that they go to a branch first, and give the commands from "Changing the schema" in docs/DEVELOPMENT.md (0038). A migration can't be undone by holding the deploy.

7. Review your own work cold

Read git diff origin/main..HEAD as a skeptical reviewer who didn't write it: ownership checks on every record, error handling, edge cases, input that isn't the shape you expect, and docs that now lie. Fix what you find.

8. Report to Ryan

Keep it short and in plain English. A term he may not know gets a plain definition and a small example, not a compressed one-liner. The docs carry the full story; the report points to them.

Start with the "Needs you" block, before anything else. Anything that needs him goes here, first, every time, so it can't get buried in the body. Use only the groups that have something in them:

## Needs you

🛑 **Do before you push**
- ...

❓ **Waiting on your answer** (I won't move on until you reply)
- ...

👀 **Not fixed. Say so now or it stays this way**
- <what>, and where it's recorded (roadmap, an ADR's consequences)
  • 🛑 is anything he must do himself: run a command, change a dashboard setting, or review something because a step was skipped or couldn't be verified.
  • ❓ is any open question. Ask it here, not mid-paragraph. If it has options, give them with a recommendation.
  • 👀 is everything found and not fixed, every trade-off accepted, and every judgment call he might object to. One line each, saying what he'd notice.
  • If no group has anything, write the line **Needs you:** nothing. so its absence is never ambiguous.
  • Never put any of these only in the body below. The body can repeat them with more detail.

Then the rest:

  1. Shipped: the item, in one sentence.
  2. Commits: the list from git log --oneline origin/main..HEAD, and which one deserves the closest read.
  3. Decisions: anything you chose on his behalf, and why.
  4. Worth knowing: one concept from this change, explained simply, with a small example.
  5. Next: the item /next would pick up after this one.

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.