whileai-sdk
CLAUDE.md21 starsChanged 51 days agoApache-2.0View on GitHub
Read CONSTITUTION.md first: what the library is, what we believe, and where each belief is enforced. Then docs/reference/style.md for the ergonomics standard. One command, run from any checkout, by anyone with write access: It cuts the next number (0.N then 0.N+1, never 1.0) from whatever sits under ## Unreleased in CHANGELOG.md, lands the bump on main, and starts publish.yml. Runs queue on a concurrency group, so two agents shipping in the same minute get two releases in order, or one…
- Commits and pushes
1 other project uses a version of this.
Adopted by
Most recently taken up by
- whilehq/whileai-sdkword for word, first seen Sep 22, 2026
Other versions1 project uses a version of this.
Word for word · 1
- whilehq/whileai-sdk★ 0100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
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 repository12 other files it gives its agents.
- audit-your-judgeSkill
- characterSkill
- dpo-pairsSkill
- grpo-verifierSkill
- harness-searchSkill
- manage-experimentsSkill
- pick-a-methodSkill
- sft-from-tracesSkill
- strengthen-your-evalsSkill
- tool-call-efficiencySkill
- watchSkill
- whileai-simulationsSkill
The file
We split this CLAUDE.md into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 3
Read CONSTITUTION.md first: what the library is, what we believe, and where each belief is enforced.
- Very specific: line 23
Never bump version by hand.
- Very specific: line 25
Every change ships with an entry under ## Unreleased.
- Distinctive: line 49
Add files by path. Never git add -A.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md 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.
Files
Working in whileai-sdk
Read CONSTITUTION.md first: what the library is, what we believe, and
where each belief is enforced. Then docs/reference/style.md for the
ergonomics standard.
Shipping a release
One command, run from any checkout, by anyone with write access:
gh workflow run release.ymlIt cuts the next number (0.N then 0.N+1, never 1.0) from whatever sits under ## Unreleased in
CHANGELOG.md, lands the bump on main, and starts publish.yml. Runs queue
on a concurrency group, so two agents shipping in the same minute get two
releases in order, or one release and one "nothing to ship". Watch it with
gh run watch and confirm with pip index versions whileai.
Rules the tooling enforces (CI fails otherwise):
- Never bump
versionby hand. A PR that bumps it may touch only CHANGELOG.md, both pyprojects and uv.lock. - Every change ships with an entry under
## Unreleased. Keep that header; the cut renames it and puts a fresh one above. - To see what a cut would ship without doing it:
uv run python .github/scripts/release.py --dry-run
Coding standard
.claude/agents/ holds the three agents that apply the constitution on
demand (style-guide, docs-designer, user-sim). docs/reference/style.md is the standard: PyTorch/DSPy ergonomics. One
import (import whileai as wai), one command (wai login, never the
package name followed by a verb; scripts/check_cli_name.py fails CI on
the long spelling), objects carry configuration and calls
carry data, at most eight parameters on a public call, reports print
themselves (no format_* twins), verbs a scientist says. New public
names go one dot down or become methods on the rows object.
tests/api/test_style_ratchet.py pins today's counts of the retired
shapes; lower a pin when you retire a name, never raise one.
Pull requests
gh pr createfails on this repo ("must be a collaborator"); open PRs through the API:gh api -X POST repos/whilehq/whileai-sdk/pulls -f title=.. -f head=<branch> -f base=main -F body=@body.md- A PR that touches
whileai/must touchdocs/too, or carry theno-docslabel. - Add files by path. Never
git add -A; other sessions share this checkout.
Checks
uv run pytest -q
uv run ruff check . && uv run ruff format --check .
uv run mypy # the gate
uv run ty check # same, under a second; mypy stays the gate until ty is 1.0
uv run python scripts/check_no_hardcoding.pyDoes 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.