Organization repository on GitHub
Orcha — multi-agent orchestration platform (CLI, FastAPI portal, desktop app)
- Forks
- 5
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Practices & Patterns
Orcha Jev Judgment Spike
Open spike issue in the Orcha multi-agent orchestration platform: it proposes moving judgment out of expensive state-of-the-art models and onto Jev, which emits precise instructions for a cheaper executor to carry out, and requires findings, a cost model, and a recommendation as issue comments.
- Stars
- 6
- Language
- Python
- License
- MIT
- Published by
- Community
- Last verified
- Oct 2026
About this project
Human-authoritative multi-agent orchestration as Claude Code slash commands. Multiple Claude Code sessions collaborate on a high-level objective through a shared Postgres database; the human holds standing authority (approve, reprioritise, reassign, arbitrate) over every subtask.
Quoted from the project's README. Read the full README.
- Languages
- Python
- Platforms
- CLIPostgreSQLmacOS
- Capabilities
- Open sourceSelf-hostedRequires API key
Highlights
- The issue asks for five findings: what Jev is from TypeSafe's docs, where it plugs into Orcha, a cost model, a quality read, and a recommendation.
- The proposed pattern: Jev produces the reasoning, plan, decisions, and constraints, then a cheaper executor such as Sonnet types out the result.
- Orcha is a Python 3.10+ CLI with a FastAPI/Postgres backing service and a Claude Code slash-command agent layer, MIT licensed.
- Named risks: vendor lock-in, a second vendor's terms over the code, wrong instructions followed confidently, debuggability, and Jev downtime.
- The definition of done requires findings as comments on the issue, with sources linked, ending in pursue, pursue-with-conditions, or drop; no PR is required.
Watch out
This entry tracks an open spike issue rather than shipped code, and the outcome was not decided when checked; Orcha's README also notes its hosted dogfood deployment is decommissioned in favor of local-first mode.
Questions about this entry
Short answers drawn from the linked project or page.
It proposes Jev as a judgment layer that produces the hard part of a task - reasoning, plan, decisions, constraints - as precise instructions for a cheaper executor model to carry out, to cut state-of-the-art model spend while holding quality.
The definition of done requires findings as comments on issue #253 covering Jev's API and terms, Orcha integration seams, a cost model, a small quality benchmark, and risks, ending with an explicit pursue, pursue-with-conditions, or drop call.
Docker is required, because Orcha runs Postgres and its web portal as containers, along with Python 3.10 or newer for the `orcha` CLI and Claude Code, where the slash commands run. Node.js and Xcode are optional, needed only to build the desktop app or the macOS widget.
Homebrew is the recommended route: `brew install open-orcha/orcha/orcha`, or `brew tap open-orcha/orcha` then `brew install orcha`. Installing from source for hacking on Orcha itself is `git clone`, then `pip install ./orcha-cli`.
Yes, Orcha's agents run on Claude, so they need an Anthropic API key. The README recommends creating a key, loading $20 of credit, and setting it as ORCHA_LLM_API_KEY or ANTHROPIC_API_KEY; it says $20 goes a long way and there is no subscription to manage.
Plan gating uses ORCHA_PLAN: `solo` is the default and free, running the complete cloud portal locally on a laptop, while `team` is the hosted, paid tier that adds members, invites and roles.
No agent self-certifies: `/orcha-done` only flips a task to needs_verification, and only a human or a human-delegated reviewer flips it to completed via `/orcha-verify`. Humans are first-class agents in the same table as AI agents, and agents never create other agents.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Launching agents on a canvas
X: A chief of staff for bots
X: An agent with a Jev model router
More Jev resources like this
3GitHub stars
Evaluation scripts, MCP tools, and a skill router built to measure where jev-1.13.0 fits in agent harnesses, with raw results under eval/.
Jev Benchmarks & Evaluations#community#evaluation#agents
Community
Aitejiu/jev-harness-labPython
Aitejiu/jev-harness-labPython187.5kGitHub stars
Autonomous agent platform with 187,504 stars for building and running agents from plain English or a visual builder. Its platform features first-class TypeSafe Jev decision blocks for typed routing, filtering, scoring and confidence-gated next-action dispatching.
Jev Integrations & Adapters#community#autonomous-agents#platform
Community
Significant-Gravitas/AutoGPTPython
Significant-Gravitas/AutoGPTPython29.8kGitHub stars
Open-source workspace with 29,796 stars for building, deploying and monitoring AI agents and workflows -- visual builder, chat, tables, files and knowledge -- featuring native TypeSafe System One evaluation and decision provider integration.
Jev Integrations & Adapters#community#agents#workflow-automation
Community
simstudioai/simTypeScript
simstudioai/simTypeScript12.4kGitHub stars
Open-source AI email assistant (Next.js, Postgres, 12k+ stars) with a decision-model layer that can run on openrouter:typesafe/jev-1.13: Jev-backed questions cover rule selection, cold-email detection, thread status, recurring patterns, unsubscribe pages, and sender categorization, each with evals.
Jev Integrations & Adapters#community#email#gmail
Community
elie222/inbox-zeroTypeScript
elie222/inbox-zeroTypeScript106GitHub stars
A personal-assistant agent with 100 mocked tools that measures how many steps a task takes when the LLM picks the tool versus when Jev picks it before every model step.
Jev Practices & Patterns#community#typescript#agents
Community
vinilana/jev-eval-agentHTML
vinilana/jev-eval-agentHTML21GitHub stars
MIT-licensed Python tool router for coding agents that keeps obvious calls on a fast path, asks Jev (default typesafe/jev-1.13 via OpenRouter) for genuinely ambiguous choices, and can run bounded MCTS when multi-step consequences matter. Ships a CLI, MCP server and hook integrations.
Jev Tools & CLIs#community#routing#mcp
From the Jev guides
Original write-ups on what this entry is for.
Most Jev accuracy problems are state problems. What belongs in a state, what is a pre-made judgment to leave out, and why naming the slice you ask about matters more than the wording.
Jev Practices & Patterns#state#design#primitives
Read article
A Noul returns one number and no separate confidence. How to read the value, pin its boundary with criteria, split compound questions, and set a per-action threshold that matches the cost of being wrong.
Jev Practices & Patterns#noul#primitives#confidence
Read article
A Score is a weighted position, not a ruler. Why the number can land between levels, why levels are judged separately, and how to combine several atomic scores in code instead of one crowded rubric.
Jev Practices & Patterns#score#primitives#scoring
Read article
From the Jev community
Posts from builders shipping with Jev right now.