Agentic Workbench

The agentic workbench is Tangle’s shared workspace for teams building with AI agents. It is multiplayer by design, so engineering, product, and business teams can work in the same space with shared context. Runs capture inputs, outputs, and logs so results are reviewable.

What You Can Do

• Build and iterate on software with agent workflows in a shared project space.
• Run background tasks that keep working without babysitting.
• Compare variants across multiple approaches or branches.
• Evaluate at scale with simulations over large task sets.
• Control access with profiles, policies, and limits.
• Connect tools through integrations and shared workspace context.

Who This Is For

• Product teams shipping AI-assisted software. Start with workflows.
• Engineering orgs standardizing agent workflows. Define profiles.
• Builders who need repeatable, reviewable output. Run simulations.

How It Connects

The workbench sends execution to the sandbox runtime. Hosted runs use product/API-key access. Protocol-backed runs depend on a registered blueprint service instance, operator endpoint, and payment path.

Profiles Power The Runtime

Workbench profiles configure the selected agent harness, model routing, per-agent prompts, and tool permissions. OpenCode sits beside 12 other Sandbox SDK backend types: Claude Code, Kimi Code, Codex, AMP, Factory Droids, Pi, Hermes, Forge, OpenClaw, ACP, Cursor, and CLI base. Protocol-backed service instances must still read their own /api/capabilities response.

Start Here (By Role)

• Builders: Start with agent workflows.
• Team leads: Define guardrails in profiles and policies.
• Evaluators: Run simulations before shipping.
• Platform teams: Wire tools in integrations.

The workbench is available via partnership or early access.