metis
Small Python framework and GitHub Action for Jev-powered workflows, with issue triage and PR screening built as examples.
Overview
metis-workflow is a small Python framework, with no runtime dependencies, for workflows powered by Jev. You define a workflow class with named steps; a step can use the Jev integration to check an event against yes/no questions, and later steps read earlier outputs by name. The README's example routes a support ticket to billing when the billing probability is at least 0.8, to support when at most 0.2, and to manual review otherwise.
The Workflow base class owns execution, skips, error handling and a run report; configuration has defaults and validation; integrations are shared clients; and a dry-run flag suppresses external changes in write steps. Conditional steps are supported. It ships a CLI and a reusable GitHub Action, and both automatic GitHub issue triage and pull-request screening are built entirely with the framework as examples. An earlier standalone triage package is archived. The version is 0.1.0, installable from the checkout, with migration and custom-workflow guides.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Does this ticket concern a payment or invoice? | — |
| Choice | Which label or route should this GitHub issue get?(inferred) | — |
How it works
- Define a Workflow subclass with steps
- Assess step asks Jev typed questions about the event
- Route step applies thresholds to outputs
- Workflow produces a run report; dry-run suppresses writes
- Run via CLI, application code or GitHub Action
Strengths
- Clear separation of judgment and routing
- No runtime dependencies
- Reusable GitHub Action
- Dry-run support
Caveats
- Early version (0.1.0)
- Not yet on PyPI
- Requires a TypeSafe key
Getting started
python -m pip install . then run SupportRouting().run(event={...}) or use the CLI.At a glance
- Type
- Project
- Status
- beta
- Language
- Python
- Runtime
- library, CLI and GitHub Action
- Has code
- yes
Use cases and integrations
How curated lists describe it
Metis: automatic GitHub issue triage powered by TypeSafe AI Jev. A reusable GitHub Action.
hellogumbo/awesome-jev
Frequently asked questions
What does Jev decide?
Answers to workflow-defined questions, such as whether a ticket concerns billing.
Can it triage GitHub issues?
Yes, issue triage and PR screening are included as examples via a GitHub Action.
Does it change anything in dry-run?
No, write steps suppress external changes.
Related projects
Glue that turns one Linux machine into a team of coding agents, with Jev making typed calls like who builds a task and whether it may merge.
ProjectbetaTypeScriptSelf-hosted Next.js API and GitHub Actions workflow that has Jev pick one label for each new issue or pull request.
ServicebetaTypeScriptLabels GitHub pull requests with Jev via OpenRouter: type, area, platform, attention and a size based on conceptual scope.
ProjectbetaPythonProbot GitHub app that classifies, scores, checks completeness and dedupes issues and PRs with Jev, routing uncertain ones to humans.
ProjectbetaTypeScriptSelf-hosted Jira alternative where AI agents join Scrum teams; Jev auto-assigns tasks, fills fields and routes workflow conditions.
ServicebetaTypeScriptGitHub Action that applies a repository's existing labels to new issues and PRs, asking Jev one question per label group in one request.
ProjectbetaTypeScript