fix(eve): run messages as turns while tool approvals stay open

[ruiconti]

Fixes #1224: a message sent while a tool approval was pending disappeared into deferredStepInput with no model call — the session looked dead until someone answered the approval. Now the message runs as an ordinary turn, the approval stays open and answerable, and a later structured answer still restores the parked batch and runs the tool. The first commit adds the acceptance-gate integration test proving the AI SDK accepts an approval batch restored after intervening turns (it passes against the unchanged harness); the second commit builds the behavior on top of it.

What

• Messages never wait behind approvals. resolvePendingInput runs a follow-up message as a normal turn when the pending work is approval-bearing. Session-limit prompts keep blocking (the budget gate must not run the model), and a message behind a single question-only batch keeps its dismiss-and-continue behavior. Text that matches an option (approve) still resolves directly — that removal is a separate, contract-stage change.
• Pending batches become an ordered collection (pending-input-batches.ts, split out of input-requests.ts). Required because the first change makes a previously-unreachable state reachable: an intervening turn can park with its own requests, and the old singleton key was overwritten unconditionally — silently destroying the open approval and its withheld model output. Each batch now resolves independently, restoring only its own output.
• Wedged sessions self-heal. The legacy singleton key reads as a one-element list, and the old deferral's held message replays through the existing consumeDeferredStepInput path on the next delivery — no migration step needed; upgrading un-wedges stuck sessions.

Data flow

• before → message during approval: resolvePendingInput → queueDeferredStepInput, turn parks with no model call; second park unreachable, singleton safe by loop order only.
• after → message during approval: resolvePendingInput returns continue, model runs, batch untouched; a park in that turn appends a second batch; a late response resolves exactly its own batch and splices its withheld output behind the grown history.

Coverage

• Unit: batch coexistence, legacy-key shim, multi-batch message (nothing dismissed), partial responses deferring without blocking the message, wedge-release, plus the flipped former wedge tests.
• Integration: tool-loop-generate-approval-resume (first commit) proves the late splice against the real AI SDK, including the intervening-turn case.
• E2e: unrelated-message-queued is replaced by unrelated-message-runs (message runs immediately, approval still completes once); the Anthropic provider: pending approval-parked tool calls break every subsequent model call (tool_use without adjacent tool_result → 400 → session.failed) #533 repro keeps its no-replay assertions but drops the no-model-step pin.

Scope notes

• With several batches open, a bare option word is ambiguous and stays a message (text matching is single-batch only).
• With any approval open in a multi-batch state, messages dismiss nothing — questions stay answerable. Actor-scoped supersession and lifecycle events remain contract-stage work (research/hitl-request-lifecycle.md).
• docs/tools/human-in-the-loop.md updated in the same PR.

Checks run: pnpm test:unit, pnpm test:integration, pnpm typecheck, pnpm lint, pnpm guard:invariants, pnpm docs:check — all green locally.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
eve-docs	Building	Preview	Aug 10, 2026 8:53pm
eve-docs-4759	Ready	Preview, v0	Aug 10, 2026 8:53pm

[github-actions]

Bundle + Package Summary: apps/fixtures/weather-agent

Key takeaways

• Package delta: install footprint 72.23 MB -> 72.83 MB (+596.7 kB ⚠️).
• Runtime delta: public routes 13 -> 14 (+1); 1 changed payload.

Delta vs main (f835ad5)

Area	Metric	Baseline	Current	Delta
Package	Packed tarball	7.69 MB	7.87 MB	+179.1 kB ⚠️
Package	Unpacked publish size	28.79 MB	29.39 MB	+596.7 kB ⚠️
Package	Installed footprint	72.23 MB	72.83 MB	+596.7 kB ⚠️
Package	Published files	3063	3092	+29
Package	Installed files	6773	6802	+29
Runtime	Unique function payloads	2	2	0
Runtime	Total function bytes	17.06 MB	17.14 MB	+74.6 kB ⚠️
Runtime	Public routes	13	14	+1

Changed function payloads vs main (f835ad5) (2)

Function	Status	Baseline	Current	Delta	Route changes
functions/__server.func	changed	8.53 MB	8.57 MB	+37.3 kB ⚠️	public +/eve/v1/connections/[name]/callback/[attemptId]/[token]
functions/.well-known/workflow/v1/flow.func	changed	8.53 MB	8.57 MB	+37.3 kB ⚠️	none

eve init install

Metric	Baseline	Current	Delta
Installed footprint	110.64 MB	111.24 MB	+596.7 kB ⚠️
Installed packages	121	121	0
dependencies	4	4	0
devDependencies	2	2	0
Dependency package bytes	43.28 MB	43.88 MB	+596.7 kB ⚠️
devDependency package bytes	5.04 MB	5.04 MB	0 B ➖

Build Metadata

• Preset: vercel
• Nitro: nitro@3.0.260610-beta
• Output directory: apps/fixtures/weather-agent/.vercel/output
• Build metadata timestamp: 2026-08-10T20:54:56.193Z
• Route aliases: 14 public, 1 internal (15 total aliases)
• Vercel routes in config: 17
• Severity legend: 🔴 dominant/large, 🟠 notable, 🟡 watch, ⚪ small

Package Drill-Down

Package Details

• Package: eve@0.31.3
• Package directory: packages/eve
• Tarball: 7.87 MB (eve-0.31.3.tgz)
• Unpacked payload: 29.39 MB across 3092 published files
• Installed footprint: 72.83 MB across 6802 installed files
• Installed root package: 28.04 MB
• Installed dependencies: 44.79 MB
• Runtime dependencies: 2
• Peer dependencies: 5 (4 optional)

Installed footprint is measured from an isolated temporary npm install of the packed tarball.

Heavy installed dependencies

• eve: 28.04 MB (38.5%)
• @rolldown/binding-linux-x64-gnu: 19.35 MB (26.6%)
• ai: 6.67 MB (9.2%)
• zod: 5.07 MB (7.0%)
• undici: 3.50 MB (4.8%)

Publish payload breakdown

Published file size
🔴 dist/src/compiled/shadcn-registry/index.js       [################........] 9.68 MB 32.9%
🟠 dist/src/compiled/@photon-ai/chat-adapter-ime... [#####...................] 2.92 MB 10.0%
🟠 dist/src/compiled/experimental-ai-sdk-code-mo... [###.....................] 1.51 MB 5.1%
🟡 dist/src/compiled/@vercel/sandbox/index.js       [#.......................] 638.9 kB 2.2%
🟡 dist/src/compiled/_chunks/workflow/undici-CUI... [#.......................] 508.8 kB 1.7%
🔴 Other published files                            [########################] 14.13 MB 48.1%

Installed footprint breakdown

Installed package size
🔴 eve                             [########################] 28.04 MB 38.5%
🔴 @rolldown/binding-linux-x64-gnu [#################.......] 19.35 MB 26.6%
🔴 ai                              [######..................] 6.67 MB 9.2%
🔴 zod                             [####....................] 5.07 MB 7.0%
🟠 undici                          [###.....................] 3.50 MB 4.8%
🟠 nitro                           [##......................] 2.41 MB 3.3%
🔴 Other installed packages        [#######.................] 7.79 MB 10.7%

Runtime dependencies (2)

Package	Range	Notes
nitro	3.0.260610-beta
undici	8.9.0

Peer dependencies (5)

Package	Range	Notes
@opentelemetry/api	^1.0.0	optional peer
ai	catalog:
braintrust	^3.0.0	optional peer
just-bash	^3.0.0	optional peer
microsandbox	^0.5.0	optional peer

eve init install drill-down

eve init install details

• Command: eve init my-agent
• Package manager: npm
• Installed footprint: 111.24 MB across 8678 installed files
• Installed packages: 121 total (115 transitive-only)
• dependencies: 4 direct packages totaling 43.88 MB
• devDependencies: 2 direct packages totaling 5.04 MB
• Other transitive package files: 62.32 MB

Installed footprint is measured from an isolated temporary eve init my-agent using the current packed eve tarball.

Heavy installed dependencies

• eve: 28.04 MB (25.2%)
• @typescript/typescript-linux-x64: 27.95 MB (25.1%)
• @rolldown/binding-linux-x64-gnu: 19.35 MB (17.4%)
• zod: 9.02 MB (8.1%)
• ai: 6.67 MB (6.0%)

Installed footprint breakdown

Installed package size
🔴 eve                              [########################] 28.04 MB 25.2%
🔴 @typescript/typescript-linux-x64 [########################] 27.95 MB 25.1%
🔴 @rolldown/binding-linux-x64-gnu  [#################.......] 19.35 MB 17.4%
🔴 zod                              [########................] 9.02 MB 8.1%
🔴 ai                               [######..................] 6.67 MB 6.0%
🟠 undici                           [###.....................] 3.50 MB 3.2%
🔴 Other installed packages         [##############..........] 16.71 MB 15.0%

dependencies (4)

Package	Range	Installed size	Share
@vercel/connect	0.4.3	141.0 kB	0.1%
ai	^7.0.38	6.67 MB	6.0%
eve	file:eve-0.31.3.tgz	28.04 MB	25.2%
zod	4.4.3	9.02 MB	8.1%

devDependencies (2)

Package	Range	Installed size	Share
@types/node	24.x	2.54 MB	2.3%
typescript	7.0.2	2.50 MB	2.2%

Function Drill-Down

Payload Size Graph

Unique function payload size and share of total
🔴 functions/.well-known/workflow/v1/flow.func     [########################] 8.57 MB 50.0%
🔴 functions/__server.func                         [########################] 8.57 MB 50.0%

Top Function Payloads

🟠 functions/.well-known/workflow/v1/flow.func • 1 public route • 8.57 MB

Metric	Value
Public routes	/.well-known/workflow/v1/flow
Runtime	nodejs24.x
Handler	index.mjs
Payload	8.57 MB
Function files	8.57 MB across 44 files
Traced dependencies	0 B
Signal	🟠 Bundled file index.mjs is 2.36 MB (27.5%)

🟠 🔎 Dependency Analysis

📦 Bundled files:

Bundled file size
🟠 index.mjs                       [########################] 2.36 MB 27.5%
🟠 _chunks/runtime-artifacts.mjs   [################........] 1.63 MB 19.1%
🟡 _libs/undici.mjs                [##########..............] 980.5 kB 11.4%
🟡 _chunks/sandbox.mjs             [########................] 777.1 kB 9.1%
🟡 _libs/@ai-sdk/gateway+[...].mjs [####....................] 432.8 kB 5.1%
🟠 Other bundled files             [########################] 2.38 MB 27.8%

🧾 Vercel Config

{
  "handler": "index.mjs",
  "launcherType": "Nodejs",
  "shouldAddHelpers": false,
  "supportsResponseStreaming": true,
  "runtime": "nodejs24.x",
  "maxDuration": "max",
  "experimentalTriggers": [
    {
      "type": "queue/v2beta",
      "topic": "__eve776561746865722d6167656e74_wkf_workflow_*",
      "consumer": "default",
      "retryAfterSeconds": 5,
      "initialDelaySeconds": 0
    }
  ],
  "environment": {
    "WORKFLOW_PRECONDITION_GUARD": "1"
  }
}

---

🟠 functions/__server.func • 13 public routes, 1 internal alias • 8.57 MB

Metric	Value
Public routes	/ /eve/v1/callback/[token] /eve/v1/connections/[name]/callback/[attemptId]/[token] /eve/v1/connections/[name]/callback/[token] /eve/v1/health /eve/v1/info /eve/v1/session /eve/v1/session/[sessionId] /eve/v1/session/[sessionId]/cancel /eve/v1/session/[sessionId]/clear /eve/v1/session/[sessionId]/compact /eve/v1/session/[sessionId]/reset /eve/v1/session/[sessionId]/stream
Internal aliases	/__server
Runtime	nodejs24.x
Handler	index.mjs
Payload	8.57 MB
Function files	8.57 MB across 44 files
Traced dependencies	0 B
Signal	🟠 Bundled file index.mjs is 2.36 MB (27.5%)

🟠 🔎 Dependency Analysis

📦 Bundled files:

Bundled file size
🟠 index.mjs                       [########################] 2.36 MB 27.5%
🟠 _chunks/runtime-artifacts.mjs   [################........] 1.63 MB 19.1%
🟡 _libs/undici.mjs                [##########..............] 980.5 kB 11.4%
🟡 _chunks/sandbox.mjs             [########................] 777.1 kB 9.1%
🟡 _libs/@ai-sdk/gateway+[...].mjs [####....................] 432.8 kB 5.1%
🟠 Other bundled files             [########################] 2.38 MB 27.8%

🧾 Vercel Config

{
  "handler": "index.mjs",
  "launcherType": "Nodejs",
  "shouldAddHelpers": false,
  "supportsResponseStreaming": true,
  "runtime": "nodejs24.x"
}

Build Timing: e2e/fixtures/agent-tools-sandbox

This is an informational timing measurement inside eve build, from preflight through publication. Output-size measurement and profile writing are excluded.

Build mode: deployable Vercel build with sandbox template prewarm included.

• Build pipeline: 2.02 s -> 2.08 s (+61.8 ms) vs main (f835ad5).
• Timing is informational: shared GitHub runners are too variable for a hard timing budget.

Detailed phase timings vs `main (f835ad5)`

Phase	Baseline	Current	Delta
extension.check	9.4 ms	1.1 ms	-8.3 ms
project.resolve	3.5 ms	0.5 ms	-3.0 ms
workspace.create	1.4 ms	0.6 ms	-0.8 ms
host.prepare	234.1 ms	193.9 ms	-40.2 ms
vercel.service-prefix.resolve	1.9 ms	2.0 ms	+0.1 ms
nitro.create	194.2 ms	231.5 ms	+37.3 ms
sandbox.prewarm	249.0 ms	296.9 ms	+47.9 ms
nitro.cache.prepare	0.3 ms	0.2 ms	-0.1 ms
nitro.prepare	0.8 ms	0.7 ms	-0.1 ms
nitro.public-assets	0.7 ms	0.7 ms	0.0 ms
nitro.prerender	0.4 ms	0.5 ms	+0.1 ms
nitro.bundle	1.30 s	1.33 s	+29.1 ms
nitro.cache.write	0.3 ms	0.3 ms	0.0 ms
vercel.workflow-function.materialize	20.1 ms	20.0 ms	-0.1 ms
agent-summary.emit	0.5 ms	0.6 ms	+0.1 ms
nitro.close	0.2 ms	0.1 ms	-0.1 ms
output.publish	3.2 ms	3.3 ms	+0.1 ms
workspace.remove	1.9 ms	1.7 ms	-0.2 ms