jev-harness (apixly-ai)
Python harness for open-ended goals: AI-authored tools and checks, Jev choosing the next tool and arguments from fresh state.
Overview
Jev Harness by Apixly runs AI-authored task scripts for goals whose route may be unknown. The primary AI defines tool capabilities, fresh parameter providers, observation and completion checks; Jev uses the goal, admitted conversation, new state and history to choose the next tool and its arguments; code executes, refreshes state and checks completion. Bounds control permissions and resources while Jev discovers the path from observations.
Example goals include finding three applicable self-hosted tools in an unfamiliar catalogue, diagnosing a repository issue from source and recorded failures, and diagnosing a task through a controlled API. For recurring work on new data with the same criteria, the README recommends judging first and capturing the procedure once. It is an independent community project at runtime 0.3.0, distributed as apixly-jev-harness from its Git URL, since the PyPI name jev-harness belongs to an unrelated project.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which tool should be used next, with which arguments? | — |
| Yes / no | Is there enough evidence to stop?(inferred) | — |
How it works
- Primary AI writes a task script with tools and checks
- Jev chooses the next tool and arguments from fresh state
- Code executes and refreshes observations
- Completion checks decide when the goal is met
Strengths
- Handles goals with unknown routes
- Code keeps execution and bounds
- Evidence reports and agent contract docs
Caveats
- Early runtime (0.3.0)
- Install from Git, not PyPI
Getting started
Install apixly-jev-harness from the repository's Git URL; see the documentation site and agent contract.At a glance
- Type
- Project
- Status
- beta
- Language
- Python
- Runtime
- Library and CLI
- Has code
- yes
Use cases and integrations
How curated lists describe it
Jev-native agent loops: typed tools and parameter choices, explicit context, verified outcomes.
hellogumbo/awesome-jev
Frequently asked questions
What does Jev decide?
The next tool and its arguments while exploring an open goal.
Who defines the tools?
The primary AI writes them in a task script; code executes them.
Can I pip install jev-harness?
No, that PyPI name is an unrelated project; use the Git URL.
Related projects
One skill and MCP server for Claude Code, Codex and Hermes where Jev gates large-model turns, routes models and guards Git steps.
MCP serverbetaPythonPython agent runtime where Jev Choice decisions pick tools and arguments, and low-confidence picks go to an LLM for review.
ServicebetaPythonDecision middleware for AI agents that uses local Laya to pick tools and judge results, falling back to an LLM when unsure.
ProjectbetaPythonSelf-improving multi-platform agent harness with an optional Jev companion for typed judgments and per-session model routing.
ServicebetaTypeScriptOmO/senpi agent plugin where Jev or Span-01 rates which skills and next tools fit the task and spots loops or completion.
Agent skillbetaTypeScriptClaude Code function hook that sends the agent's this-or-that questions to Jev and optionally ranks installed skills per prompt.
ProjectbetaTypeScript