windmill-labs/windmill · error

This ${type} draft "${path}" was started from an older deplo

Error message

This ${type} draft "${path}" was started from an older deployed version (forked from ${base}, latest is ${head}). Deploying now would overwrite the version deployed since. Call rebase_draft to discard the stale draft and get your changes back as a diff, then re-apply them (the new draft re-bases onto the latest version) and deploy. To deploy as-is and replace the newer version, call deploy_workspace_item again with force: true.

What it means

The deploy_workspace_item tool detects that the chat's draft was forked from deployed version `base` while the item's latest deployed version is now `head` (base !== head). Deploying would silently clobber the newer deployment, so it throws with recovery guidance: rebase the draft, or pass force: true to overwrite deliberately.

Source

Thrown at frontend/src/lib/components/copilot/chat/global/core.ts:6194

			triggerKind
		},
		null,
		2
	)
}

// A draft started from an older deploy would silently overwrite whatever was
// deployed since. Block the deploy and point the model at rebase_draft, unless it
// explicitly forces the overwrite. `base`/`head` undefined ⇒ can't tell ⇒ allow.
function assertDraftBasedOnLatest(
	type: WorkspaceItemType,
	path: string,
	base: string | number | undefined,
	head: string | number | undefined,
	force: boolean | undefined
): void {
	if (force || base == null || head == null || base === head) return
	throw new Error(
		`This ${type} draft "${path}" was started from an older deployed version (forked from ${base}, ` +
			`latest is ${head}). Deploying now would overwrite the version deployed since. Call rebase_draft to ` +
			`discard the stale draft and get your changes back as a diff, then re-apply them (the new draft ` +
			`re-bases onto the latest version) and deploy. To deploy as-is and replace the newer version, call ` +
			`deploy_workspace_item again with force: true.`
	)
}

// Discard a stale draft and return its own changes (vs the fork base) as a diff so
// the model can re-apply them on the latest deploy. Discarding rather than resetting
// keeps the base pointer honest (the next write re-bases on the current head) and
// fails safe: a premature deploy hits "no draft" instead of silently shipping the
// latest unchanged.
async function rebaseDraft(
	args: { type: WorkspaceItemType; path: string },
	ctx: WriteDraftCtx
): Promise<string> {
	switch (args.type) {

View on GitHub (pinned to e474e8803c)

Solutions

1. Call rebase_draft to discard the stale draft and receive the changes as a diff, re-apply them onto the latest version, then deploy
2. If overwriting the newer deployment is intentional, re-call deploy_workspace_item with force: true
3. Coordinate with whoever deployed the newer version before choosing force

Example fix

// before
deploy_workspace_item({ type: 'script', path: 'u/admin/cleanup' })
// after — option 1 (safe)
rebase_draft({ type: 'script', path: 'u/admin/cleanup' }) // then re-apply changes and deploy
// after — option 2 (overwrite intentionally)
deploy_workspace_item({ type: 'script', path: 'u/admin/cleanup', force: true })

const deployed = await getDeployedVersion(type, path)
if (draft.baseVersion !== deployed.version) {
  // draft is stale: prefer rebase_draft over deploying
}

try {
  await deployWorkspaceItem(args)
} catch (e) {
  if (e instanceof Error && e.message.includes('older deployed version')) {
    await rebaseDraft(args) // then re-apply changes and deploy
    // or intentionally: await deployWorkspaceItem({ ...args, force: true })
  }
}

When it happens

Trigger: Someone (or another chat/CLI deploy) deployed a newer version of the same script/flow/app/trigger after this chat's draft was created, and the chat then calls deploy_workspace_item without force.

Common situations: Parallel editing via UI and AI chat; a teammate deployed while the agent worked; re-running an old chat session whose draft predates recent deployments.

Related errors

• trigger_kind is required when discarding a trigger draft.
• No draft found for ${type} "${path}".
• This forked workspace '${workspaceId}' (${workspaceName}) al
• Invalid AI agent provider configuration: ${errors.join('\n')
• AI agent modules ${providerless.map((id) => `"${id}"`).join(

AI-assisted analysis of windmill-labs/windmill@e474e8803c (2026-09-03). Data as JSON: /api/errors/d17f176d1820ad26. Report an issue: GitHub.