Val-sss 76fa84da08 feat: wire model_capability into pipeline — tier-adaptive prompts + ctx auto-detect ...

Model tier detection now active (was orphaned code):
- orchestrator.__init__: detect tier from generator.llm.ollama_model
- ctx.model_tier written at run() start, flows to Generator
- Generator._build_system(): SMALL gets strict format constraints
  (1 function, ≤10 lines, preserve indent, no explanation, ≤15 words between tools)
  LARGE gets minimal constraint; MEDIUM uses base rules as-is
- _get_max_retries() respects model_strategy.max_retries

Context window auto-detection (get_effective_ctx):
- 4-layer probe: llama.cpp /props → vLLM /v1/models → Ollama modelinfo.llama.context_length → tier default
- Passed to LLMBackend n_ctx in build_pipeline
- SMALL=16K, MEDIUM=32K, LARGE=64K defaults; native ctx capped at 64K×0.8

501 tests green, 0 regression.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-05-06 20:33:59 +08:00

..

commands

feat: wire model_capability into pipeline — tier-adaptive prompts + ctx auto-detect

2026-05-06 20:33:59 +08:00

__init__.py

Kaiwu v3 MVP: deterministic expert pipeline for local-model coding agent

2026-04-26 12:47:34 +08:00

formatters.py

fix: unify version to 1.5.1 from single source (pyproject.toml)

2026-05-06 19:47:24 +08:00

main.py

feat: audit log + model cmd + indent alignment fix

2026-05-06 20:13:26 +08:00

onboarding.py

feat: three flywheels + anonymous telemetry + prompt quantification

2026-05-06 19:39:17 +08:00

repl.py

refactor: split CLI main.py (1861→173 lines) + expert EventBus emit

2026-05-06 17:45:09 +08:00

status_bar.py

v0.7.0: P1+P2+搜索重构+UI全面优化

2026-04-28 17:52:59 +08:00