cap-evolve
CLAUDE.md69 starsChanged 5 months agoApache-2.0View on GitHub
Benchmark runs reach models through three IBM providers. Use them in this order: 1. RITS (ibm-rits/…): no budget. Use it whenever it serves the model. 2. ETE-INT (ibm-ete-int/…): the team's main gateway. 3. ETE (ibm-ete/…): a second gateway with its own budget. Use it only when ETE-INT cannot serve the model or is over budget. How to follow this: - When you dispatch benchmarks.yml, pick a plain model name (claude-opus-5, gemma-4-31B-it, …). CI then applies the order above by itself,…
In this repository22 other files it gives its agents.
- llms.txtllms.txt
- agent-optimizeSkill
- evographSkill
- gepaSkill
- hill-climbSkill
- skilloptSkill
- mcp-toolSkill
- skill-packageSkill
- system-promptSkill
- toolsSkill
- wikiSkill
- run-optimizerSkill
- orchestrateSkill
- using-cap-evolveSkill
- baselineSkill
- diagnoseSkill
- evaluateSkill
- finalizeSkill
- gateSkill
- implement-and-checkSkill
- intakeSkill
- reportSkill
The file
We split this CLAUDE.md into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
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 14
When you dispatch benchmarks.yml, pick a plain model name (claude-opus-5, gemma-4-31B-it, …).
- Very specific: line 20
After a run, the job summary and the run record (agent_provider, optimizer_provider, *_provider_skipped) show which provider was used.
≈ “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
cap-evolve — notes for Claude Code sessions
Benchmark CI: which model provider to use
Benchmark runs reach models through three IBM providers. Use them in this order:
- RITS (
ibm-rits/…): no budget. Use it whenever it serves the model. - ETE-INT (
ibm-ete-int/…): the team's main gateway. - ETE (
ibm-ete/…): a second gateway with its own budget. Use it only when ETE-INT cannot serve the model or is over budget.
How to follow this:
- When you dispatch
benchmarks.yml, pick a plain model name (claude-opus-5,gemma-4-31B-it, …). CI then applies the order above by itself, before the run starts. - Pick a prefixed id (
ibm-ete/aws/claude-opus-5) only when an experiment must stay on one gateway, and say why in the issue or PR. - The order and the model list live in
ci/benchmarks/model_catalog.txt. Read its header before you add a model: list two ids under one name only when they are the same model. - After a run, the job summary and the run record (
agent_provider,optimizer_provider,*_provider_skipped) show which provider was used. Report spend per provider from these.
Details: ci/benchmarks/README.md, section "Model providers and their order".
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.