33616664384
66%

Ran 02 Sep 2026 09:54AM UTC

Jobs 1

Files 79

Run time 1min

Badge

Embed ▾

Committed 01 Sep 2026 10:47AM UTC coverage: 66.056% (+0.7%) from 65.374%

Build # 33616664384

Build Type

push

github

Committed by

cod3smith

Commit Message

Document local-LLM ops: native Ollama on macOS, context and parallel pins

Docker-on-Mac Ollama is CPU-only in a VM (measured 30-60x slower than
Ollama.app with Metal); the compose local-llm profile is Linux-only
now. Ops doc covers the two silent quality traps: the 4096 default
context truncates qualification prompts, and NUM_PARALLEL splits the
window per slot.

Coverage Stats

3460 of 5238 relevant lines covered (66.06%)

0.66 hits per line

Coverage Regressions

Lines	Coverage	∆	File
112	74.91	1.31%	karani/ingestion/storage.py
64	0.0	0.0%	karani/cli.py
13	33.33	1.33%	karani/qualification/anthropic.py
5	90.54	0.0%	karani/qualification/agent.py
2	96.08	15.31%	karani/qualification/factory.py
2	88.24	2.52%	karani/qualification/__init__.py
2	92.86	21.43%	karani/qualification/openrouter.py
1	96.15	0.0%	karani/qualification/client.py

Jobs

ID	Job ID	Ran	Files	Coverage
1	33616664384.1	02 Sep 2026 09:54AM UTC	79	66.06	GitHub Action Run

cod3smith / karani / 33616664384
66%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Coverage Regressions

Jobs

Source Files on build 33616664384

cod3smith / karani / 33616664384 66%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Coverage Regressions

Jobs

Source Files on build 33616664384

cod3smith / karani / 33616664384
66%

README BADGES
x