SpecJudge scores your specs locally and ranks the best AI model for the job
jokiruiz · reddit · 2026-07-22
SpecJudge is a local-first CLI that reads Spec-Driven Development artifacts and recommends the right-sized AI model for implementing them.
It runs the judge locally on Ollama, so project specs never leave the machine, and it produces a ranked podium based on fit first and price second. The tool also exports a self-contained HTML report, uses a YAML model catalog that can be extended without changing code, and is MIT licensed. The author invites feedback on the judging logic and the model/price catalog.
Install path:
- ollama pull llama3.1:8b
- pip install specjudge
- specjudge /path/to/your/project
Related event: SpecJudge Locally Evaluates Project Specs to Recommend AI Models(2 posts)→
More from coding & agent
- Google’s CodeMender is now standalone, but its best version still needs an invite — shashib · 2026-07-22
- Meta VR CLI adds an AI performance analysis workflow for Unity devs — Vjeux · 2026-07-22
- Agent alert feed shows a loop of self-generated code proposals — nptacek · 2026-07-22
- Whatbroke diffs agent traces to catch silent regressions after a model swap — Impossible-Alarm-738 · 2026-07-22
- People are using LLMs to handle long software installations — domdod9 · 2026-07-22
- Cloudflare-style infra is making agent-first apps feel radically easier to build — threepointone · 2026-07-22