Timmy_Foundation/timmy-home
16
0
Fork 0
Code Issues 239 Pull Requests Actions Packages Projects Releases Wiki Activity

[UNIWIZARD] Routing modules — quality scorer, task classifier, self-grader, heartbeat #106

Closed
Rockachopa wants to merge 1 commits from feat/uniwizard-routing-modules into main
pull from: feat/uniwizard-routing-modules
merge into: Timmy_Foundation:main
Conversation 0 Commits 1 Files Changed 15 +5863
Rockachopa commented 2026-03-30 20:00:03 +00:00
Owner
Copy Link

What

5,863 lines of production code + design docs for the Uniwizard routing intelligence layer.

Built By

Kimi K2.5 running through OpenClaw gateway, dispatched via Gitea assigned-kimi labels.
This PR is the proof that the Kimi delegation pipeline works end to end.

Modules

Module Lines Tests Issue
quality_scorer.py 642 28 pass #98
task_classifier.py 655 32 pass #88
self_grader.py 769 26 pass #89
kimi-heartbeat.sh 110 — —

86 tests total, all passing.

Design Docs

  • task_classifier_design.md — Backend strength mapping, classification logic
  • self_grader_design.md — Session grading methodology, pattern detection
  • context_compression_review.md — Audit of existing Hermes compressor with improvement proposals
  • context_compressor.patch — Concrete diff for compressor improvements
  • job_profiles.yaml — Toolset profiles per job type for context narrowing
  • job_profiles_design.md — How profiles map to Hermes toolset system

How the Pipeline Worked

  1. Issues #98, #88, #89 labeled assigned-kimi in Gitea
  2. Kimi heartbeat picked them up, posted kimi-in-progress comments
  3. Kimi worked each issue through OpenClaw (kimi/kimi-code backend)
  4. Code produced at ~/.timmy/uniwizard/
  5. Tests written and verified (86/86 passing)
  6. Packaged into this PR

What This Enables

  • quality_scorer: Track which backends perform best at what. recommend_backend('code') returns ranked list.
  • task_classifier: Classify any prompt into task type + backend preferences. Wire into routing.
  • self_grader: Grade sessions automatically. Identify failure patterns. Weekly improvement reports.
  • kimi-heartbeat: Label an issue → Kimi works it → results posted back. Gitea IS the dashboard.

Closes #98, closes #88, closes #89

## What 5,863 lines of production code + design docs for the Uniwizard routing intelligence layer. ## Built By Kimi K2.5 running through OpenClaw gateway, dispatched via Gitea `assigned-kimi` labels. This PR is the proof that the Kimi delegation pipeline works end to end. ## Modules | Module | Lines | Tests | Issue | |--------|-------|-------|-------| | quality_scorer.py | 642 | 28 pass | #98 | | task_classifier.py | 655 | 32 pass | #88 | | self_grader.py | 769 | 26 pass | #89 | | kimi-heartbeat.sh | 110 | — | — | **86 tests total, all passing.** ## Design Docs - task_classifier_design.md — Backend strength mapping, classification logic - self_grader_design.md — Session grading methodology, pattern detection - context_compression_review.md — Audit of existing Hermes compressor with improvement proposals - context_compressor.patch — Concrete diff for compressor improvements - job_profiles.yaml — Toolset profiles per job type for context narrowing - job_profiles_design.md — How profiles map to Hermes toolset system ## How the Pipeline Worked 1. Issues #98, #88, #89 labeled `assigned-kimi` in Gitea 2. Kimi heartbeat picked them up, posted `kimi-in-progress` comments 3. Kimi worked each issue through OpenClaw (kimi/kimi-code backend) 4. Code produced at ~/.timmy/uniwizard/ 5. Tests written and verified (86/86 passing) 6. Packaged into this PR ## What This Enables - **quality_scorer**: Track which backends perform best at what. `recommend_backend('code')` returns ranked list. - **task_classifier**: Classify any prompt into task type + backend preferences. Wire into routing. - **self_grader**: Grade sessions automatically. Identify failure patterns. Weekly improvement reports. - **kimi-heartbeat**: Label an issue → Kimi works it → results posted back. Gitea IS the dashboard. Closes #98, closes #88, closes #89
Rockachopa added 1 commit 2026-03-30 20:00:04 +00:00
feat: uniwizard routing modules — quality scorer, task classifier, self-grader, heartbeat 041a91ba42 
Built by Kimi K2.5 via OpenClaw gateway, dispatched through Gitea labels.

Modules:
- quality_scorer.py: per-backend performance tracking (issue #98)
- task_classifier.py: prompt classification and routing (issue #88)
- self_grader.py: session grading and improvement patterns (issue #89)
- kimi-heartbeat.sh: Gitea-driven dispatch to Kimi via OpenClaw

86 tests, all passing.

Includes design docs, context compressor review/patch, and job profiles.

Closes #98, closes #88, closes #89
Rockachopa closed this pull request 2026-03-30 20:04:02 +00:00

Pull request closed

This pull request cannot be reopened because the branch was deleted.
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
Clear labels
alembic
The Alchemical Vessel - Kimi-powered transformation
 architecture
Auto-created architecture label
 assigned-claw-code
Queued for Code Claw (qwen/openrouter)
 assigned-kimi
Dispatched to Kimi via OpenClaw
 assigned-sonnet
auth-needed
auth-needed
 bizzalo
bizzalo
 blocker
blocker
 claw
Claw Code runtime (Rust)
 claw-code-done
Code Claw completed this task
 claw-code-in-progress
Code Claw is actively working
 deadline-7am
deadline-7am
 epic
Parent tracking issue
 golden_bilbo
Bilbo at peak performance - max churn, fast responses
 harness-engineering
Auto-created harness-engineering label
 intel
Competitive intelligence
 kimi-done
Kimi completed this task
 kimi-in-progress
Kimi is actively working on this
 morning-report
morning-report
 runtime
Runtime abstraction layer
 saiyah
Auto-created saiyah label
 sonnet-ready
study
Learning from external source code
 substratum
The dissolution engine - runtime layer
 substratum:invoke
Trigger Substratum execution
 urgent
urgent
 velocity-engine
Auto-generated by velocity engine
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
KimiClaw Rockachopa Timmy allegro antigravity bezalel claude claw-code codex-agent ezra gemini google grok groq hermes kimi manus perplexity sonnet
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Timmy_Foundation/timmy-home#106
Delete Branch "feat/uniwizard-routing-modules"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?