Skip to main content

Agent mode

Describe an outcome; the agent plans the work, edits across files, and runs commands — with guardrails at every step.

The agent's tools​

Reading and searching files, proposing edits, creating and deleting files, running shell commands, fetching URLs, semantic codebase search, and image generation. Edits are always staged as diffs for your approval.

You control how much the agent can run on its own via editvolt.commands.runMode:

  • Auto-review (default) — every command shown to you first
  • Allowlist — commands matching your allowlist run automatically; everything else asks. The allowlist lives in .editvolt/permissions.json (workspace or global)
  • Run everything — no prompts (use with care)

Extra protections are on by default: file-deletion protection (editvolt.commands.protectFileDeletion) and browser-access protection (editvolt.commands.protectBrowser).

Checkpoints​

Before the agent changes anything, EditVolt records a checkpoint. Roll back to any point in the conversation if a direction doesn't work out — your workspace is restored exactly.

Multitask​

Multitask mode runs several agent workers in parallel on independent subtasks. Control concurrency with editvolt.multitask.maxConcurrency.

The Agents window​

Start a project from an empty window: the Agents window scaffolds a working project (static site or Next.js) deterministically, then hands off to the agent for your actual feature work. An opt-in in-editor browser (editvolt.browser.enabled) lets agents see what they build.