Skip to content

Roadmap · subject to change

Roadmap

Where Tavronus Forge is and where it is going. One honest caveat up front: this is an early prototype, and both the order and the scope of everything below may change as we learn. Nothing here is a commitment to a date.

Available in prototype

  • Interactive workflow demoAvailable in prototype

    The full task lifecycle in simulation, including plan approval and failure & replan.

  • Review & diff experienceAvailable in prototype

    Verification report, self-review, and annotated diff with approve / request changes / reject.

  • Inspectable memory ledgerAvailable in prototype

    Categorized project memory with explicit vs. inferred entries and confidence.

  • Demo analytics dashboardAvailable in prototype

    Illustrative task and throughput analytics at /dashboard.

  • Task detail viewsAvailable in prototype

    Per-task state, trace, and artifacts.

In active development

  • Early-access programIn active development

    A real intake channel for teams who want to shape the product. Today the form is a preview.

  • Richer demo scenariosIn active development

    More task types, failure modes, and replan paths in the simulated workspace.

Planned

  • Real repository connectionPlanned

    Scoped read access and isolated task workspaces against actual repositories.

  • GitHub integration & PR creationPlanned

    Tempered tasks land as reviewable pull requests.

  • Local runnerPlanned

    Execution on your own hardware — see the local runner page.

  • VS Code extensionPlanned

    Cast and review without leaving the editor.

  • Desktop appPlanned

    The control center as a native application.

  • Team memoryPlanned

    Shared project memory across a team, with ownership and review of entries.

  • Enterprise permission policiesPlanned

    Org-level policy for what tasks may do without approval.

  • Audit exportsPlanned

    Task ledgers exportable for compliance and incident review.

Researching

  • Multi-agent orchestrationResearching

    Several VULCAN instances cooperating on decomposed work.

  • Self-hosting at scaleResearching

    Runner fleets on customer infrastructure.

  • Model auto-routing telemetryResearching

    Learning which execution strategy fits which task shape.

Following along

Shipped changes are recorded in the changelog. If you want to influence what gets built next, the early-access program is the channel.