How idikuterke/onoz-brain works
onoz-brain is a model-independent Python CLI that shares skills across linked projects and records their outcomes for evidence-based promotion. The main loop is project linking → agent briefing → skill execution/verification → run logging → stats and human-approved promotion. It also provides project health/activity dashboards, evals, lessons, and anonymized exports. Most orchestration lives in brain.py; the README documents several capabilities whose implementation details were not sampled, so those links cite the README rather than implying source inspection.
Main entry points: brain.py Brain CLI · brain_export.py Anonymous exporter · kos.py Skill runner · runs Run ledger
Components 20
Project workflow
- Brain CLI
brain.py - Project registry
brain.py - Project repositories
- Agent briefing
brain.py - Skill runner
skills/kos/kos.py
Skills and evidence
- Skill library
skills - Lessons
memory/lessons
Measurement and oversight
- Run ledger
memory/runs - Promotion rules
brain.py - Run statistics
brain.py - Eval task sets
evals - Eval runner
brain.py - Eval records
memory/evals - Project health
brain.py - Commit activity
brain.py - Read-only dashboard
brain.py
Data export
- Anonymous exporter
skills/anonim-export/scripts/brain_export.py - Export package
Other components
- Engineer
- External agent
Connections 34
- Engineer → Brain CLI: runs commands
README.md - Brain CLI → Project registry: links projects
README.md - Project registry → Project repositories: marks project root
README.md - Brain CLI → Agent briefing: builds context
README.md - Agent briefing → External agent: briefs
README.md - External agent → Project repositories: works in
README.md - Engineer → Skill runner: runs skill
skills/kos/kos.py - Skill runner → Skill library: loads verification
skills/kos/kos.py - Skill runner → Project repositories: runs verification
skills/kos/kos.py - Skill runner → Brain CLI: invokes log
skills/kos/kos.py - Brain CLI → Run ledger: records runs
README.md - Engineer → Brain CLI: adds lessons
README.md - Brain CLI → Lessons: stores rules
README.md - Brain CLI → Run statistics: summarizes runs
README.md - Brain CLI → Promotion rules: calculates candidates
README.md - Run ledger → Promotion rules: supplies evidence
README.md - Engineer → Promotion rules: approves apply
README.md - Engineer → Eval runner: runs eval
README.md - Eval runner → Eval task sets: selects task
README.md - Eval runner → Project repositories: tests project
brain.py - Eval runner → Eval records: records results
brain.py - Brain CLI → Project health: collects status
README.md - Project health → Project repositories: reads git state
README.md - Brain CLI → Commit activity: classifies commits
README.md - Commit activity → Project repositories: reads commit history
README.md - Brain CLI → Read-only dashboard: generates view
README.md - Project health → Read-only dashboard: feeds health
README.md - Commit activity → Read-only dashboard: feeds activity
README.md - Engineer → Anonymous exporter: runs export
skills/anonim-export/scripts/brain_export.py - Anonymous exporter → Run ledger: reads runs
skills/anonim-export/scripts/brain_export.py - Anonymous exporter → Eval records: reads evals
skills/anonim-export/scripts/brain_export.py - Anonymous exporter → Lessons: reads lessons
skills/anonim-export/scripts/brain_export.py - Anonymous exporter → Project registry: reads project metadata
skills/anonim-export/scripts/brain_export.py - Anonymous exporter → Export package: writes package
skills/anonim-export/scripts/brain_export.py