VS Code-style IDE
A three-pane explorer, tabbed editor and docked Dev Workmate with Quick Open, in-file find and an integrated terminal.
DevSpace Studio is AgenTorQ's AI-native IDE — a VS Code-style, three-pane workspace with a Dev Workmate that browses, edits and ships Apex, LWC and Flows, and drives real git on GitHub. Ask, plan, edit, diff, approve, deploy, test, fix — every change is a diff you approve before anything deploys.
One IDE, two code homes — Salesforce metadata and GitHub repositories.
// Ask the Dev Workmate:
// "Query only open opps and enforce sharing."
--- proposed diff (approve to deploy) ---
- List<Opportunity> opps = [SELECT Id FROM Opportunity];
+ List<Opportunity> opps = [SELECT Id, Amount
+ FROM Opportunity
+ WHERE StageName != 'Closed Won'
+ WITH USER_MODE];
mode: Ask before edits # Plan never deploys
guard: exact-anchor + shrink-guard
next: Approve → Deploy → Run tests → Fix
Review the code, describe a change, and ship it as an approvable diff. An explorer of Apex, LWC and Flow, a diff-first editor, and a Developer Copilot that plans the change and proposes a governed deploy action — you approve before anything ships.
Here’s my plan to harden the callout:
DevSpace Studio brings developer-grade AI to the tools you already use — the same governed engine that powers every AI Workmate, focused on code. It runs in the standalone AgenTorQ Cloud workspace and inside the Salesforce-native managed package.
A three-pane explorer, tabbed editor and docked Dev Workmate with Quick Open, in-file find and an integrated terminal.
A role-aware AI pair that knows the open file and your selection, reasons over the codebase, and proposes surgical diffs.
Drive Apex, LWC and Flow metadata and real git on GitHub — two code homes in one workspace, no context switch.
Every change is a diff you approve. Enforced modes, deploy and test gates, and an immutable audit trail before anything ships.
The same layout you already think in — a file tree on the left, a tabbed editor in the middle, and an AI Workmate docked on the right that reads the open file and your selection to propose changes as diffs.
Browse Salesforce metadata and GitHub repositories in one tree. Quick Open jumps to any file; the Source Control view shows repos, changes and branches.
A tabbed, editable code editor with in-file find, syntax highlighting, a test-coverage gutter and related-test jump — plus an integrated terminal for the sf CLI and git.
Docked AI that knows the open file and selection. Describe a change or select code, get a surgical diff, and run senior code review before you approve.
DevSpace Studio speaks native Salesforce. Browse and edit Apex classes, Lightning Web Components and Flows, deploy to your org over the sf CLI, and run Apex tests with coverage — without leaving the platform boundary you control. In the Salesforce-native edition, work is enforced in USER_MODE.
# Dev Workmate deploys the approved change,
# then runs the related tests and reports coverage.
$ sf project deploy start -m ApexClass:OppRiskController
Status Succeeded Components 1/1
$ sf apex run test -t OppRiskController_Test -r human
Outcome Passing Pass 12 Fail 0
Coverage OppRiskController 86%
# A failing test? The Workmate proposes a fix diff
# — you approve before it redeploys. Nothing ships blind.
# GitHub is a first-class code provider —
# real git over the API, no local clone.
branch create feature/opp-risk ok
edit OppRiskController.cls staged
commit "Scope SOQL + enforce USER_MODE" ok
pr open #482 base: main ok
pr list 2 open ok
merge #482 squash merged
# Same layer drives GitLab, Bitbucket & Azure DevOps.
GitHub is a first-class code provider through a code-provider registry. Browse repositories down to files, edit and commit, and run real git operations over the API — create branches, open and list pull requests, and merge — with no local clone. A VS Code-style Source Control panel brings changes, commit, branch, PR and merge into the workspace.
Every change moves through the same governed loop. You stay in control at the gate that matters — a reviewable diff — and the Dev Workmate does the mechanical work around it.
Describe the change in plain language, or select the code you want the Dev Workmate to work on.
The Workmate reads the open file and codebase and lays out the change. In Plan mode it never deploys.
It applies a surgical edit against an exact anchor — not a blind whole-file rewrite.
Changes surface as a clear before/after diff, with senior code review on request.
You approve or reject — per diff or all at once. Nothing ships without your go-ahead.
The approved change deploys to your org over the sf CLI, or commits and ships to git.
Related Apex tests run automatically, with results and coverage reported back in the editor.
On a failure, the Workmate proposes a fix diff — you approve, and the loop closes.
Five enforced operating modes you switch per turn. The rules are enforced server-side — Plan never deploys, and destructive actions such as deletes always require confirmation.
Proposes a diff and waits for your approval before applying anything.
Applies safe edits, then runs the related tests automatically.
Designs the change end to end and never deploys — pure design mode.
Runs the multi-step loop with less prompting, inside your guardrails.
Maximum autonomy for trusted flows — deletes still confirm.
Developer autonomy without the risk. Approval gates, edit-integrity guards, an audit trail and per-workspace memory keep the Dev Workmate accountable for everything it touches.
Every change is a diff you approve before deploy. Plan never deploys, and deletes always require explicit confirmation.
Edits apply against an exact anchor with a shrink guard, so a whole-file rewrite can never silently truncate your code.
Every prompt, diff, deploy, test run and git operation is recorded to an immutable audit trail for review and compliance.
Each workspace keeps its own context — open work, decisions and results — and applies it on the next turn, private by default.
In the Salesforce-native edition, sharing, CRUD and field-level security are enforced — the Workmate only touches what you can.
The Dev Workmate is scoped to development work only, so a code session cannot wander into unrelated business actions.
AgenTorQ is built to be extended. Register custom actions, add custom MCP connectors, reach any system through the connector layer, and call governed, multi-model AI natively from Apex, LWC and Flows.
Register REST endpoints, Salesforce operations or internal tools as typed actions with role-based access and approval on writes.
Add any MCP server with OAuth 2.1 PKCE, Dynamic Client Registration and automatic tool discovery — run as many as you like.
Reach hundreds of apps and your own systems over REST, GraphQL, databases and webhooks through the connector layer.
The managed package exposes namespaced Apex, Lightning Web Components and Flow actions — call governed AI in-org.
// Call governed, multi-model AI natively from Apex.
public with sharing class OppRiskController {
@AuraEnabled
public static AgentResult score(Id oppId) {
return AgenTorQ.Agents.run(
new AgentRequest()
.agent('opportunity-risk')
.recordId(oppId) // record-aware
.mode('ask') // approval-gated
); // runs WITH USER_MODE
}
}
Illustrative. Namespaced Apex, LWC and Flow are real in the managed package; long-tail systems are reachable through the connector layer, REST or MCP — not bespoke certified integrations.
DevSpace Studio is AgenTorQ's AI-native, VS Code-style IDE. It has three panes — a file explorer, a tabbed editor, and a docked Dev Workmate that knows the open file and your selection. You browse, edit and deploy Apex, LWC and Flows, and run real git on GitHub, with Quick Open, in-file find, an integrated terminal (sf CLI and git), a test-coverage gutter, related-test jump, a Flow diagram visualizer, senior code review and Apply-with-AI.
Yes. GitHub is a first-class code provider through a code-provider registry. You browse repos to files, edit and commit, and run real git over the API — create branches, open and list pull requests, and merge — with no local clone. GitLab, Bitbucket and Azure DevOps are supported through the same layer, and a VS Code-style Source Control panel shows repos, changes, commit, branch, PR and merge.
Five enforced operating modes you switch per turn: Ask before edits (proposes a diff and waits for approval), Edit (applies safe edits then runs tests), Plan (designs the change and never deploys), Auto and Bypass. The rules are enforced server-side, and destructive actions such as deletes always require confirmation.
No. Every change surfaces as a surgical diff you review and approve before anything deploys. Plan mode never deploys, deletes always confirm, and each edit is applied against an exact anchor with a shrink guard so a whole-file rewrite can never silently truncate your code. Every action is recorded to an immutable audit trail.
Yes. Register custom actions with a typed schema and role-based access, add custom MCP connectors (OAuth 2.1 PKCE with Dynamic Client Registration and automatic tool discovery), reach systems over REST, GraphQL, databases and webhooks through the connector layer, and call governed, multi-model AI natively from Apex, LWC and Flows in the Salesforce managed package.
No. AgenTorQ is an enterprise AI platform with two deployment editions of one engine — a standalone AgenTorQ Cloud workspace and a Salesforce-native AppExchange managed package. DevSpace Studio drives both Salesforce metadata (Apex, LWC, Flows) and GitHub repositories, so it is an AI-native IDE for Salesforce and general code alike.
Open DevSpace Studio in the standalone workspace, or book a technical walkthrough of the Salesforce-native edition. Same governed engine — you choose where it runs.
reachout@agentorq.com