Skip to content

eidos-agi/ml-forge

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

ml-forge

Solve software problems with ML, not just LLMs.

ml-forge teaches AI agents how to build small, local ML models that ship as static artifacts inside projects. No cloud APIs, no data exfiltration, no heavy deps — just learned weights that make software smarter.

Why

Every forge in the Eidos ecosystem hits the same wall: static checklists that miss context. A linter config exists but the preflight doesn't check it. A test suite exists but the runner doesn't know which tests are relevant. An app is redundant but nobody hardcoded the synonym.

These are all the same problem: given features about the current state, predict the right action. That's ML. And for most of these problems, a 10KB logistic regression trained on project history beats both a hardcoded list and an expensive LLM call.

The Pattern

Extract local features → Train offline → Ship as static artifact → Infer at runtime
  • Features: whatever signals are locally available (files, configs, system metadata, git history)
  • Training: dev-time script, labeled from existing project data (synonym groups, CI configs, test results)
  • Artifact: JSON weight vector or hardcoded constants — versioned in git
  • Inference: stdlib Python, zero deps, zero network, fast and deterministic

Skills

Skill Description
/ml-assess Is this problem ML-shaped? Evaluate features, labels, and value proposition
/ml-extract Design feature extraction from local signals
/ml-train Build offline training with proper evaluation
/ml-ship Package model as minimal static artifact
/ml-evaluate Precision, recall, feature importance, quality report

Guardrails

  • No cloud APIs — everything runs locally
  • Models ship as static artifacts — no training at runtime
  • Minimize runtime deps — prefer weights @ features over import sklearn
  • This repo is knowledge only — no installable software

Use Cases

Project Problem ML Solution
ship-forge Static preflight misses project-specific checks Train on CI configs → predict which checks matter
test-forge Tests wrong target (live vs local) Train on git/env signals → predict correct target
apple-a-day Hardcoded app synonym list Train on framework fingerprints → predict similarity

License

MIT — see LICENSE

About

Solve software problems with ML, not just LLMs. Skills and templates for building small, local ML models that ship as static artifacts.

Resources

License

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors