contract
Skill0 starsChanged 4 months agoMITView on GitHub
Turn a fuzzy task into an executable contract (a plan in small, verifiable steps) BEFORE executing. Triggers "/contract", "plan out", "break down this task", "plan before starting". It's piece 1 of the universal harness — used at the start of any non-trivial task, especially when running on a cheaper API model, so the model doesn't have to plan on its own. The gate later verifies against this contract.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- Railly/skillstheir own version, first seen Nov 1, 2026
- DataWithBaraa/claude-code-data-engineering-skillstheir own version, first seen Nov 1, 2026
- mephistopheles4/grimoiretheir own version, first seen Oct 31, 2026
- aibtcdev/skillstheir own version, first seen Oct 26, 2026
- langwatch/langwatchtheir own version, first seen Oct 17, 2026
- DanaResearchGroup/agent-skillstheir own version, first seen Aug 26, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- langwatch/langwatch★ 5ksame name
- DataWithBaraa/claude-code-data-engineering-skills★ 84same name
- aibtcdev/skills★ 11same name
- Railly/skills★ 11same name
- DanaResearchGroup/agent-skills★ 1same name
- mephistopheles4/grimoire★ 1same 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 repository11 other files it gives its agents.
- AGENTS.mdAGENTS.md
- brain/AGENTS.mdAGENTS.md
- brain/CLAUDE.mdCLAUDE.md
- CLAUDE.mdCLAUDE.md
- brainSkill
- checkpointSkill
- continueSkill
- gateSkill
- pmSkill
- researchSkill
- saveSkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 17
Read the real thing first: Don't plan from memory: open the files/paths the task touches and verify names, fields, existing patterns.
- Distinctive: line 34
If ≤5 steps and no risk → proceed and execute.
- Distinctive: line 35
If >5 steps or anything destructive/ambiguous → wait for the owner's OK first.
- Very specific: line 36
Keep the contract (in the chat, or a file if the task is big): /gate will use it as the acceptance criteria.
“Common” is measured: kept by at least 6 in 10 of the 6 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.
Files
Skill: /contract — build the task contract (piece 1 of the harness)
You turn the request into an executable, verifiable contract BEFORE touching anything. This hands the model (especially a cheap API one) the plan, so it doesn't improvise or hallucinate.
When to use
At the start of any non-trivial task. If it's an obvious one-liner, skip it.
What you do
- Read the real thing first. Don't plan from memory: open the files/paths the task touches and verify names, fields, existing patterns. No inventing paths/APIs/params.
- Fill the contract with this exact structure:
# Task contract
## Objective (1 sentence)
## Context / inputs
## Steps (small, numbered) → each with an observable **expected result**
## Success criteria (checklist of observable conditions)
## What NOT to do (out of scope)
## Risks / irreversible steps- Golden rule for steps: each step has an observable expected result (so the gate can verify it one by one). Small steps > big steps.
- Show the contract and:
- If ≤5 steps and no risk → proceed and execute.
- If >5 steps or anything destructive/ambiguous → wait for the owner's OK first.
- Keep the contract (in the chat, or a file if the task is big):
/gatewill use it as the acceptance criteria.
Language & tone
Reply in {{LANG}}, direct, no filler. The contract is concrete, not literature.
Input
$ARGUMENTS
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
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.