agentleFS
Sign inStart free
IsmailAzzouzIsmailAzzouz/Work-With-Me-Skill/skills/work-with-me/SKILL.md

work-with-me

Skill0 starsChanged 28 days agoMITView on GitHub

Default collaboration workflow for non-trivial work. Keeps every task in its own working-sessions/ folder, reads the project's agent-journal/ first (learn-project), and runs fixed pragmatic phases: knowledge retrieval → scope → acceptance criteria → how → tests → refinements → fixes → coverage/issues. Use at the start of every non-trivial task, feature, bugfix, investigation, or when the user says "work with me", "session", or "follow the workflow".

5 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions5 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 5

    1. SaltKode/obedient-claude-skill★ 1same name
    2. Saltshard/obedient-claude-skill★ 1same name
    3. loonylabs-dev/claude-setup★ 0same name
    4. loonylabs/claude-setup★ 0same name
    5. nimishmehta/antria-office★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. 0 are common, 10 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Default operating mode for all non-trivial work.

    line 15
  2. Very specific:

    Use these alongside this workflow for every substantive task (not one-line Q&A).

    line 19
  3. Distinctive:

    Do not skip the stack because the task "seems simple" if it touches code, docs, or multiple files.

    line 30
  4. Very specific:

    Every project has a working-sessions/ folder at the repo root.

    line 34
  5. Distinctive:

    Create or reopen the session folder for this task.

    line 51
  6. Very specific:

    Create/update notes.md (journal of record).

    line 52
  7. Distinctive:

    Add other files as needed (see session-layout.md).

    line 59
  8. Very specific:

    Keep notes.md current as you go: decisions, gotchas, commands run, links to logs.

    line 61
  9. Very specific:

    Do phases in order. Write phase outcomes into the session (notes.md and/or dedicated files).

    line 65
  10. Distinctive:

    If the map is missing and the codebase is unfamiliar, run/update learn-project (and graphify if available) before guessing.

    line 72
  11. Distinctive:

    Write verifiable checks (pass/fail).

    line 82
  12. Very specific:

    Put them in notes.md under ## Acceptance criteria (or acceptance.md if long).

    line 83
  13. Distinctive:

    Prefer the laziness ladder.

    line 88
  14. Distinctive:

    Add or update tests that lock the acceptance criteria first when practical (red → green).

    line 93
  15. Distinctive:

    Check coverage for touched code.

    line 106
  16. Distinctive:

    Update session notes with final status: done / blocked / follow-ups.

    line 108
  17. Very specific:

    Before ending a session, leave notes.md readable by a cold agent: goal, branch, files touched, how to verify, leftovers.

    line 116

“Common” is measured: kept by at least 6 in 10 of the 5 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

Work With Me

Default operating mode for all non-trivial work.

Companion skills

Use these alongside this workflow for every substantive task (not one-line Q&A). Only learn-project ships with this skill; the rest are optional. If a companion is not installed, apply its principle inline (the "Fallback" column) and carry on.

SkillRoleFallback if not installed
learn-projectEnsure agent-journal/ exists and is fresh; machine-only project truth— (bundled)
graphifyKnowledge retrieval / codebase graph queries when graphify-out/ exists or architecture relationships matterRead manifests, entrypoints, and directory structure directly
knowledge-base-builderKeep agent-journal/ and large note trees navigable (index.md + summaries)Every note gets a one-line summary; every folder gets an index.md
karpathy-guidelinesThink before coding, simplicity, surgical diffs, verifiable goalsSurface assumptions, smallest diff, define pass/fail before coding
ponytailLaziness ladder: YAGNI, reuse, stdlib, minimal code (default full)Question the need; reuse > stdlib > platform > new code
cavemanTerse communication (default lite unless the user asks for full/ultra); journal bodies may use ultraDrop filler and hedging; keep technical detail exact

Do not skip the stack because the task "seems simple" if it touches code, docs, or multiple files.

Working sessions (mandatory)

Every project has a working-sessions/ folder at the repo root. Create it if missing.

Each piece of work gets its own session folder named for the work, e.g.:

working-sessions/
  Adding X feature/
  Fix auth token expiry/

Session artifacts live in that folder: task notes, logs, gotchas, research, debug scripts, seed data, and scratch docs. Product code, tests, and requested documentation changes stay in their normal project locations. Do not promote temporary session artifacts into the main codebase unless the user asks.

Session bootstrap

At the start of work:

  1. Ensure working-sessions/ exists.
  2. Create or reopen the session folder for this task.
  3. Create/update notes.md (journal of record).
  4. Capture context in notes.md early:
    • Goal (one sentence)
    • Repos / services involved
    • Branch name(s)
    • Key files (update via git status / git diff as work proceeds)
    • Open questions / assumptions
  5. Add other files as needed (see session-layout.md).

Keep notes.md current as you go: decisions, gotchas, commands run, links to logs.

Pragmatic workflow (order is fixed)

Do phases in order. Write phase outcomes into the session (notes.md and/or dedicated files). Do not jump to implementation before 1–4 are clear enough to execute.

1. Knowledge retrieval

  • First: read agent-journal/index.md if present (skill: learn-project). If missing or clearly stale for this work, run/refresh learn-project before deep coding.
  • Then: the project's agent instructions file (AGENTS.md or your harness's equivalent), graphify queries, KB indexes, session notes, docs, prior sessions.
  • Skim only what is needed; cite paths in session notes.md.
  • If the map is missing and the codebase is unfamiliar, run/update learn-project (and graphify if available) before guessing.

2. What needs to be done

  • State the problem and the smallest useful outcome.
  • List out-of-scope items explicitly (YAGNI).
  • Surface assumptions; ask if blocked.

3. Acceptance criteria

  • Write verifiable checks (pass/fail). Prefer tests as criteria when code changes.
  • Put them in notes.md under ## Acceptance criteria (or acceptance.md if long).

4. How to do it

  • Short plan: steps + how each step is verified.
  • Prefer the laziness ladder; surgical changes only.
  • Note repos/services/files/branch in the plan.

5. Tests

  • Add or update tests that lock the acceptance criteria first when practical (red → green).
  • Record how to run them in notes.md.

6. Refinements

  • Tighten design/API/names only if needed to meet criteria. No drive-by refactors.

7. Fixes

  • Fix failures from tests/review; keep diffs surgical; log gotchas.

8. Code coverage and issues

  • Check coverage for touched code; note gaps.
  • Scan for leftover issues (lints, TODOs you introduced, open risks).
  • Update session notes with final status: done / blocked / follow-ups.

Session hygiene

  • One session per distinct goal; do not mix unrelated work.
  • Promote durable docs into the real project docs/KB only when the user asks (then use knowledge-base-builder if organizing a notes tree).
  • Durable project truth learned during the session goes to agent-journal/ (learn-project); task detail stays in the session.
  • Debug scripts and seeds stay under the session folder unless they become product assets.
  • Before ending a session, leave notes.md readable by a cold agent: goal, branch, files touched, how to verify, leftovers.

Trivial work exception

Skip full session scaffolding only for truly trivial asks (definition lookup, one-line answer, single obvious edit with no investigation). If unsure, open a session.

Additional resources

  • Session folder layout and templates: session-layout.md

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.

Check your setup, free