Lazy-load sandbox function invocation GCS blobs. by Fraggle · Pull Request #32767 · dust-tt/dust

Fraggle · 2026-09-18T19:19:32Z

Description

SandboxFunctionInvocationResource now fetches invocation rows without downloading their GCS payloads. ensureData coalesces on-demand loads, async getters expose blob-backed fields, and existsForFunction handles execution-side existence checks without a storage read. Updated Google Calendar, Pod Manager, Poke, and LLM serialization callers to await data explicitly.

Tests

Updated sandbox_function_invocation_resource.test.ts and run_sandbox_function_invocation.test.ts for async getters, lazy reload, persistence, migrations, execution, and error paths. No manual testing recorded.

Risk

Low-to-medium risk: callers now cross an async boundary when they access invocation data, and a lazy GCS read can fail at call time. Updated call sites and coverage mitigate missed awaits. Rollback is safe by reverting the front changes.

Deploy Plan

Deploy front. No SQL migrations or infrastructure changes are required.

Fetch the DB row without downloading the blob; load on demand via ensureData/getters, and use existsForFunction for existence checks. Co-authored-by: Cursor <cursoragent@cursor.com>

vercel · 2026-09-18T19:19:35Z

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

2 Skipped Deployments

Project	Deployment	Actions	Updated
playground	Ignored	Preview	Sep 21, 2026 8:40am UTC
storybook	Ignored	Preview	Sep 21, 2026 8:40am UTC

Fraggle · 2026-09-18T19:20:59Z

This change is part of the following stack:

Lazy-load sandbox function invocation GCS blobs. #32767 ◀
- Add: Timings informations for tool calls. #32768
  - Tighten dsbx action poll: 500ms initial wait, then 200ms until 2s, then 500ms. #32769
    - Stage MCP tool outputs in Redis and write GCS behind markAsSucceeded. #32770
      - Stage sandbox function invocation blobs in Redis; defer GCS on create and terminal. #32771
        
        Add create/run phase server timings for sandbox MCP tool calls. #32772

_{Change managed by git-spice.}

Nothing outside tests called it; getters already expose the same blob fields. Co-authored-by: Cursor <cursoragent@cursor.com>

getInput, getResult, getError, and getBundleSha256 had no callers outside tests. getContext stays; tests read the blob through toPokeJSON. Co-authored-by: Cursor <cursoragent@cursor.com>

spolu · 2026-09-19T08:14:11Z

Small comments as I re-read this file. @davidebbo a fair amount of noisy code attributes to you. listRows, SandboxFunctionInvocationRow, SandboxFunctionInvocationForLLM, toJSONForLLM (unused what is that?)

Please avoid throwing new types around especially if they are not critical. Poke stuff should be clearly distinguished from mission critical stuff. listRows is only used in poke but there is no clear way to understand that without investigating all call sites.

spolu · 2026-09-19T08:19:49Z

r? cc

github-actions

cc-verify: violations found!

no-catching-own-errors: Blob failures thrown

Findings and notifications outside the PR diff:

Source: front/lib/resources/sandbox_function_invocation_resource.ts:964

Pre-existing violation of no-catching-own-errors in CONTRACTS: loadDataFromGcs() throws a failed withRetry result instead of returning Err. The new lazy getContext() path exposes this exception inside Calendar's create_event after calendar.events.insert() succeeds, reporting a successful creation as a tool failure and risking duplicate events on retry; propagate a Result and handle the context load before the mutation.

Fraggle · 2026-09-19T09:38:02Z

@davidebbo a fair amount of noisy code attributes to you. listRows, SandboxFunctionInvocationRow, SandboxFunctionInvocationForLLM, toJSONForLLM (unused what is that?)

I removed a few that weren't used.

davidebbo · 2026-09-19T12:17:47Z

Small comments as I re-read this file. @davidebbo a fair amount of noisy code attributes to you. listRows, SandboxFunctionInvocationRow, SandboxFunctionInvocationForLLM, toJSONForLLM (unused what is that?)

@spolu SandboxFunctionInvocationForLLM and toJSONForLLM date back to the early days (see @flvndvd's #29088). At the time, we had a tool that let the model inspect invocations. We don't have it anymore, but that code ended up staying around and should be cleaned up.

Massive changes happened to go from old model to new, and we moved fast. We cleaned out a ton of obsolete code, but some obviously was missed.

davidebbo · 2026-09-19T12:19:41Z

Please avoid throwing new types around especially if they are not critical. Poke stuff should be clearly distinguished from mission critical stuff. listRows is only used in poke but there is no clear way to understand that without investigating all call sites.

@spolu Not sure that we have this properly encoded today (to name Poke-only types with Poke in their name). Agreed it's probably a good thing, though likely widely violated today. Could be a global clean up exercise.

davidebbo · 2026-09-19T12:29:03Z

    ? await UserResource.fetchByModelIds([invocation.userId])
    : [];

+  await invocation.ensureData();


Feels like something that should be in toPokeJSON(), so we don't put the onus on all the callers?

Yep, that would make toPokeJSON async but it's okay (i think we have prior art)

davidebbo

LGTM with minor comment.

spolu · 2026-09-21T06:30:17Z

    }: {
      sandboxFunction: SandboxFunctionResource;
      data?: SandboxFunctionInvocationData;
+      /** Set when `data` is the real blob (makeNew / after GCS load), not a placeholder. */


Code contract please

spolu · 2026-09-21T06:30:43Z

+      version: SANDBOX_FUNCTION_INVOCATION_DATA_VERSION,
+      input: undefined,
+    };
+    this.dataLoaded = dataLoaded;


It's surprising that dataLoaded can be set directly here?

I'll double check but it makes sense at creation, you are supposed to have the data.

spolu · 2026-09-21T06:32:13Z

      // record of a failure the stream classified precisely.
      error: callError,
    };
+    this.dataLoaded = true;


I don't understand this one. Since we fall ensureData above dataLoaded must be true already?

Correct, good catch.

spolu · 2026-09-21T06:33:53Z

+   * Load the GCS blob once and cache it on this instance. No-op when already loaded
+   * (including instances constructed from `makeNew` with in-memory data).
+   */
+  async ensureData(): Promise<void> {


Shouldn't we return data here fully loaded. This will make it more explicit in methods of this resolve to use that return value vs this.data when using the data post loading?

Something like getData() + _data & _pendingDataLoad & _dataloaded then, no ?

spolu · 2026-09-21T06:35:27Z

        "Invalid sandbox function invocation data"
      );
      this.data = { version: SANDBOX_FUNCTION_INVOCATION_DATA_VERSION };
+      this.dataLoaded = true;


It's not really loaded is it?

Yeah, I agree.

spolu · 2026-09-21T06:37:10Z

  }

+  /**
+   * DB-only existence check for execution-side pairing. Does not download the


This comment is not bringing much value is it?

spolu · 2026-09-21T06:37:46Z

    user: UserResource | null,
    mcpActions: PokeSandboxFunctionMCPAction[]
  ): PokeSandboxFunctionInvocationDetails {
+    assert(


Is that the only place where we need it such an assert?

Will refactor (see above).

spolu · 2026-09-21T06:38:08Z

   * their (function, invocation) ids come from server-minted inputs — workflow args or verified
   * sandbox JWT claims — never from user input.
+   *
+   * Existence only: does not download the invocation GCS blob.


What does existence only mean here?

spolu · 2026-09-21T06:41:31Z

cc @dust-tt/data-model-owners

Return the resolved blob from getData, load inside toPokeJSON, replace dataLoaded with dataResolved (including degraded empty parses), and add code contracts for construction/resolution and write-behind invariants. Co-authored-by: Cursor <cursoragent@cursor.com>

github-actions · 2026-09-21T08:42:02Z

  async succeed(result: unknown): Promise<boolean> {
+    const current = await this.getData();
+
    const claimed = await this.casStatus({


React Doctor · react-doctor/server-sequential-independent-await (warning)

This await doesn't use the previous result, so your users wait twice as long for nothing.

Fix → These two awaits don't depend on each other. Wrap them in Promise.all([...]) so they run at the same time.

Docs

Lazy-load sandbox function invocation GCS blobs.

bec4276

Fetch the DB row without downloading the blob; load on demand via ensureData/getters, and use existsForFunction for existence checks. Co-authored-by: Cursor <cursoragent@cursor.com>

Fraggle requested review from davidebbo and spolu September 18, 2026 20:49

Fraggle and others added 2 commits September 18, 2026 21:24

Remove unused SandboxFunctionInvocationResource.toJSONForLLM.

4fa061f

Nothing outside tests called it; getters already expose the same blob fields. Co-authored-by: Cursor <cursoragent@cursor.com>

Drop unused invocation blob getters.

e3300a4

getInput, getResult, getError, and getBundleSha256 had no callers outside tests. getContext stays; tests read the blob through toPokeJSON. Co-authored-by: Cursor <cursoragent@cursor.com>

github-actions Bot reviewed Sep 19, 2026

View reviewed changes

davidebbo reviewed Sep 19, 2026

View reviewed changes

davidebbo approved these changes Sep 19, 2026

View reviewed changes

spolu reviewed Sep 21, 2026

View reviewed changes

github-actions Bot reviewed Sep 21, 2026

View reviewed changes

Fraggle merged commit ff8cbcb into main Sep 21, 2026
53 checks passed

Fraggle deleted the sflory/lazy-load-invocation-blob branch September 21, 2026 12:53

Conversation

Fraggle commented Sep 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Tests

Risk

Deploy Plan

Uh oh!

vercel Bot commented Sep 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Fraggle commented Sep 18, 2026

Uh oh!

spolu commented Sep 19, 2026

Uh oh!

spolu commented Sep 19, 2026

Uh oh!

github-actions Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Fraggle commented Sep 19, 2026

Uh oh!

davidebbo commented Sep 19, 2026

Uh oh!

davidebbo commented Sep 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

davidebbo left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Fraggle Sep 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

spolu commented Sep 21, 2026

Uh oh!

github-actions Bot Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Fraggle commented Sep 18, 2026 •

edited

Loading

vercel Bot commented Sep 18, 2026 •

edited

Loading

davidebbo commented Sep 19, 2026 •

edited

Loading

Fraggle Sep 21, 2026 •

edited

Loading