Cluster related controls, keep operational context visible, and move through settings with the same confidence as Cursor or VS Code.
Choose which product surfaces appear in the workspace and which backend feature flags are active for the selected team.
Control which parts of Salsa are visible to this workspace so the product can feel minimal, operational, or fully loaded depending on how the team works.
Choose a preset to quickly configure which plug-ins are visible. You can customise individual plug-ins below.
Plug-ins extend Salsa with dedicated productivity tools — like having several apps under one unified AI and interface. Each plug-in is Team-scoped, so visibility settings apply to the whole team. Chat is always available.
Choose a team from the sidebar or add `?team_id=` to the URL to edit team-scoped backend feature flags.
Tune the shell, color system, motion, and workspace data presentation so the product feels closer to an editor than a control form.
Treat the interface like an editor surface: dense when you need throughput, calm when you need focus, and always legible.
Select the tonal system that sets the workbench personality.
Pin the workbench to dark, light, or let the OS decide.
Dark keeps chrome understated, light improves ambient readability, and system follows the host machine.
Tune density for fast triage or long review sessions.
Balanced spacing for dense dashboards and board work.
Keep states legible when status colors carry meaning.
Use native accents and contrast defaults across the workbench.
Choose whether replies land instantly or type out like a live operator.
Balanced pacing for day-to-day chat and review.
Notifications
Workspace
Manage connected accounts and identity providers that expand how Salsa signs in, pairs, and routes work.
Account access is team-scoped. Jump straight into the team surfaces that own OAuth, shared inboxes, and delivery credentials.
Create a team first, then return here to manage the accounts, inboxes, and external systems attached to it.
Set provider credentials, routing defaults, and the guardrails that decide which model stack handles each job.
Manage the external model providers Salsa can route into. Keep endpoints, keys, and provider type visible instead of buried in a modal.
Keep the configuration terse and operator-friendly. You can test the connection after saving.
Add your first provider to unlock external model routing, provider testing, and team-scoped delivery workflows.
Assign the right model to the right assistant surface, then inspect which capabilities each model can advertise back to the workbench.
Each assistant surface below resolves to one model from the live catalog.
Expose what each catalog model can do. These switches shape how higher-level agent contracts classify tools and output formats.
Inspect the active model roster, routing decisions, and what happens when frontier capacity or credits are unavailable.
AI Models
Enable models for chat, set a default, and test connectivity.
Test Results
Control the autonomous agents available to Salsa and how they surface across planning, execution, and monitoring flows.
Treat agents like installed capabilities, not a hidden backend list. Review what is live, what is available, and how much operating surface each agent exposes.
This workspace does not have any attached agents yet. Jump into the Marketplace tab to install one.
Browse the marketplace and attach agents that extend planning, execution, or research workflows.
Manage the workspace identity layer used for secure collaboration, verification, and external connections.
Verify your primary email so secure settings, invitations, and team-facing actions can rely on a confirmed identity.
Curate the MCP registry, adjust visibility, and inspect the tools Salsa is exposing to agents in practice.
Manage the registered tools Salsa can expose to agents. This is the source that should echo into Team Edit and Marketplace surfaces.
Browse the capability catalog and install focused tools without leaving the workbench.
Marketplace
Turn Salsa into a spoken operator surface with narration rules, voice AI controls, and live session feedback.
Control voice narration for plan and workspace cards.
Connect to Onzoe for cloud-based speech-to-text and text-to-speech in chat.
Get your API key from Onzoe dashboard. Leave empty to use local voice processing.
Choose the premium Onzoe voice Salsa should use for spoken replies. Katie is the recommended default for an executive assistant.
Adjust how fast the AI voice speaks. Affects both Onzoe voices and browser speech.
Onzoe Voice AI real-time speech-to-text and text-to-speech.
Watch the live operating state of the local stack, deployed services, and build alignment.
Real-time infrastructure health, deployed versions, and operational status.
Compares deployed exec-daemon and salsa-web git hashes. Mismatches indicate incomplete deployments.
Inspect Salsa Brain training state, peer connectivity, and runtime learning signals in one place.
Salsa Brain neural telemetry and federated learning status.
Review the knowledge graph layer, indexing state, and memory surfaces that keep the product grounded.
Knowledge
Knowledge source connectors are now managed per-team via the Team Marketplace.
Sync Settings
AI Agents
Configure in APIs tab
Decide when autonomous execution is allowed and keep the runtime state visible while plans move.
Configure and monitor autonomous execution between scheduled hours.
Control data retention, export and import flows, and whether anonymized usage can improve the system.
Manage exports, imports, and privacy posture from one place so recovery and data governance feel operational instead of hidden.
Define which local paths the workspace can see and treat as safe working context.
Define the exact directories this workspace can read and operate on. Keep the scope narrow so automation stays predictable.
Manage team members, access invitations, and shared workspace responsibilities.
Manage who can access the team, what invites are still pending, and which devices are already trusted.
Keep track of linked machines and devices that participate in local-first workflows and cloud sync.
Review which phones and desktops are linked to this workspace, then create a fresh pairing code when you need to add another device.
Audit current spend, tighten refill rules, and keep model and cloud costs inside the boundaries you expect.
Watch current burn, tune guardrails, and review refill policy behavior from one operational surface.
Pair this workspace with cloud, inspect sync health, and decide which rules apply when state moves across environments.
Version details, platform metadata, and the runtime facts you need before debugging or shipping a build.