fix(eve): manage long-running sandbox commands - #2593
Open
cmpadden wants to merge 5 commits into
Open
Conversation
Signed-off-by: Colton Padden <colton.padden@vercel.com>
Contributor
![vercel[bot]](https://avatars.githubusercontent.com/in/8329?s=60&v=4){kind=small}
Contributor
Bundle + Package Summary:
|
| Area | Metric | Baseline | Current | Delta |
|---|---|---|---|---|
| Package | Packed tarball | 8.24 MB | 8.25 MB | +3.8 kB |
| Package | Unpacked publish size | 31.09 MB | 31.10 MB | +15.0 kB |
| Package | Installed footprint | 74.72 MB | 74.74 MB | +15.0 kB |
| Package | Published files | 3534 | 3538 | +4 |
| Package | Installed files | 7249 | 7253 | +4 |
| Package | Installed package instances | 35 | 35 | 0 |
| Package | Distinct installed package names | 34 | 34 | 0 |
| Package | Installed dependency edges | 53 | 53 | 0 |
| Package | Installed optional peer edges | 47 | 47 | 0 |
| Runtime | Unique function payloads | 2 | 2 | 0 |
| Runtime | Total function bytes | 18.45 MB | 18.47 MB | +25.8 kB |
| Runtime | Public routes | 16 | 16 | 0 |
Changed function payloads vs main (a3a5ddd) (2)
| Function | Status | Baseline | Current | Delta | Route changes |
|---|---|---|---|---|---|
functions/__server.func |
changed | 9.22 MB | 9.24 MB | +12.9 kB |
none |
functions/.well-known/workflow/v1/flow.func |
changed | 9.22 MB | 9.24 MB | +12.9 kB |
none |
eve init install
| Metric | Baseline | Current | Delta |
|---|---|---|---|
| Installed footprint | 113.16 MB | 113.18 MB | +15.0 kB |
| Installed packages | 120 | 120 | 0 |
| dependencies | 4 | 4 | 0 |
| devDependencies | 2 | 2 | 0 |
| Dependency package bytes | 45.69 MB | 45.71 MB | +15.0 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: 2026-08-26T18:27:33.382Z
- Route aliases: 16 public, 1 internal (17 total aliases)
- Vercel routes in config: 19
- Severity legend: 🔴 dominant/large, 🟠 notable, 🟡 watch, ⚪ small
Package Drill-Down
Package Details
- Package:
eve@0.45.0 - Package directory:
packages/eve - Tarball: 8.25 MB (
eve-0.45.0.tgz) - Unpacked payload: 31.10 MB across 3538 published files
- Installed footprint: 74.74 MB across 7253 installed files
- Installed root package: 29.75 MB
- Installed dependencies: 44.99 MB
- Installed package instances: 35
- Distinct installed package names: 34
- Installed dependency edges: 53
- Installed optional peer edges: 47
- Runtime dependencies: 2
- Peer dependencies: 5 (4 optional)
Installed footprint is measured from an isolated temporary npm install of the packed tarball.
Graph metrics read only package.json files in package directories directly beneath a node_modules boundary, including nested boundaries. Each directory is one package instance; distinct names come from those manifests. Dependency edges count each unique name in dependencies or optionalDependencies per instance; optional peer edges count peerDependencies marked optional.
Heavy installed dependencies
eve: 29.75 MB (39.8%)@rolldown/binding-linux-x64-gnu: 19.33 MB (25.9%)ai: 6.77 MB (9.1%)zod: 5.07 MB (6.8%)undici: 3.50 MB (4.7%)
Publish payload breakdown
Published file size
🔴 dist/src/compiled/shadcn-registry/index.js [##############..........] 9.76 MB 31.4%
🟠 dist/src/compiled/@photon-ai/chat-adapter-ime... [####....................] 2.44 MB 7.9%
🟠 dist/src/compiled/experimental-ai-sdk-code-mo... [##......................] 1.51 MB 4.8%
🟡 dist/src/compiled/_chunks/workflow/undici-Dzn... [#.......................] 512.7 kB 1.6%
🟡 dist/src/compiled/_chunks/node/undici-CUIybhQ... [#.......................] 508.8 kB 1.6%
🔴 Other published files [########################] 16.37 MB 52.6%
Installed footprint breakdown
Installed package size
🔴 eve [########################] 29.75 MB 39.8%
🔴 @rolldown/binding-linux-x64-gnu [################........] 19.33 MB 25.9%
🔴 ai [#####...................] 6.77 MB 9.1%
🔴 zod [####....................] 5.07 MB 6.8%
🟠 undici [###.....................] 3.50 MB 4.7%
🟠 nitro [##......................] 2.41 MB 3.2%
🔴 Other installed packages [######..................] 7.92 MB 10.6%
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 |
^3.1.0 |
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: 113.18 MB across 9149 installed files
- Installed packages: 120 total (114 transitive-only)
- dependencies: 4 direct packages totaling 45.71 MB
- devDependencies: 2 direct packages totaling 5.04 MB
- Other transitive package files: 62.43 MB
Installed footprint is measured from an isolated temporary eve init my-agent using the current packed eve tarball.
Heavy installed dependencies
eve: 29.75 MB (26.3%)@typescript/typescript-linux-x64: 27.95 MB (24.7%)@rolldown/binding-linux-x64-gnu: 19.33 MB (17.1%)zod: 9.02 MB (8.0%)ai: 6.77 MB (6.0%)
Installed footprint breakdown
Installed package size
🔴 eve [########################] 29.75 MB 26.3%
🔴 @typescript/typescript-linux-x64 [#######################.] 27.95 MB 24.7%
🔴 @rolldown/binding-linux-x64-gnu [################........] 19.33 MB 17.1%
🔴 zod [#######.................] 9.02 MB 8.0%
🔴 ai [#####...................] 6.77 MB 6.0%
🟠 undici [###.....................] 3.50 MB 3.1%
🔴 Other installed packages [##############..........] 16.86 MB 14.9%
dependencies (4)
| Package | Range | Installed size | Share |
|---|---|---|---|
@vercel/connect |
1.0.0 |
167.9 kB | 0.1% |
ai |
^7.0.58 |
6.77 MB | 6.0% |
eve |
file:eve-0.45.0.tgz |
29.75 MB | 26.3% |
zod |
4.4.3 |
9.02 MB | 8.0% |
devDependencies (2)
| Package | Range | Installed size | Share |
|---|---|---|---|
@types/node |
24.x |
2.54 MB | 2.2% |
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 [########################] 9.24 MB 50.0%
🔴 functions/__server.func [########################] 9.24 MB 50.0%
Top Function Payloads
🟠 functions/.well-known/workflow/v1/flow.func • 1 public route • 9.24 MB
| Metric | Value |
|---|---|
| Public routes | /.well-known/workflow/v1/flow |
| Runtime | nodejs24.x |
| Handler | index.mjs |
| Payload | 9.24 MB |
| Function files | 9.24 MB across 108 files |
| Traced dependencies | 0 B |
| Signal | 🟠 Bundled file index.mjs is 1.82 MB (19.7%) |
🟠 🔎 Dependency Analysis
📦 Bundled files:
Bundled file size
🟠 index.mjs [##########..............] 1.82 MB 19.7%
🟡 _libs/undici.mjs [#####...................] 980.5 kB 10.6%
🟡 _chunks/sandbox.mjs [####....................] 781.9 kB 8.5%
🟡 _chunks/frames-Dl2Hcma4.mjs [####....................] 678.7 kB 7.3%
🟡 _chunks/chatgpt-model.mjs [####....................] 665.1 kB 7.2%
🔴 Other bundled files [########################] 4.31 MB 46.7%
🧾 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 • 15 public routes, 1 internal alias • 9.24 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/[parentSessionId]/subagents/[callId]/[childSessionId]/stream/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/eve/v1/task-input/[token] |
| Internal aliases | /__server |
| Runtime | nodejs24.x |
| Handler | index.mjs |
| Payload | 9.24 MB |
| Function files | 9.24 MB across 108 files |
| Traced dependencies | 0 B |
| Signal | 🟠 Bundled file index.mjs is 1.82 MB (19.7%) |
🟠 🔎 Dependency Analysis
📦 Bundled files:
Bundled file size
🟠 index.mjs [##########..............] 1.82 MB 19.7%
🟡 _libs/undici.mjs [#####...................] 980.5 kB 10.6%
🟡 _chunks/sandbox.mjs [####....................] 781.9 kB 8.5%
🟡 _chunks/frames-Dl2Hcma4.mjs [####....................] 678.7 kB 7.3%
🟡 _chunks/chatgpt-model.mjs [####....................] 665.1 kB 7.2%
🔴 Other bundled files [########################] 4.31 MB 46.7%
🧾 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: 2.48 s -> 2.33 s (-150.3 ms) vs
main (a3a5ddd). - Timing is informational: shared GitHub runners are too variable for a hard timing budget.
Detailed phase timings vs `main (a3a5ddd)`
| Phase | Baseline | Current | Delta |
|---|---|---|---|
extension.check |
11.4 ms | 12.1 ms | +0.7 ms |
project.resolve |
0.4 ms | 0.3 ms | -0.1 ms |
workspace.create |
0.7 ms | 0.7 ms | 0.0 ms |
host.prepare |
410.9 ms | 376.7 ms | -34.2 ms |
vercel.service-prefix.resolve |
1.6 ms | 1.6 ms | 0.0 ms |
nitro.create |
281.7 ms | 241.3 ms | -40.4 ms |
sandbox.prewarm |
366.9 ms | 295.1 ms | -71.8 ms |
nitro.cache.prepare |
0.3 ms | 0.2 ms | -0.1 ms |
nitro.prepare |
0.6 ms | 1.0 ms | +0.4 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 |
1.37 s | 1.37 s | -3.5 ms |
nitro.cache.write |
0.3 ms | 0.3 ms | 0.0 ms |
vercel.workflow-function.materialize |
33.8 ms | 32.4 ms | -1.4 ms |
agent-summary.emit |
0.7 ms | 0.7 ms | 0.0 ms |
nitro.close |
0.1 ms | 0.1 ms | 0.0 ms |
output.publish |
3.0 ms | 2.8 ms | -0.2 ms |
workspace.remove |
1.5 ms | 1.6 ms | +0.1 ms |
Signed-off-by: Colton Padden <colton.padden@vercel.com>
Signed-off-by: Colton Padden <colton.padden@vercel.com>
Signed-off-by: Colton Padden <colton.padden@vercel.com>
Signed-off-by: Colton Padden <colton.padden@vercel.com>
cmpadden
marked this pull request as ready for review
August 26, 2026 18:25
5 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Closes #440. A long-running command or stalled sandbox exit stream can hold the built-in
bashtool—and its workflow step—open indefinitely, leaving the model and user without a way to inspect or stop it.This change moves command lifecycle management behind an internal eve-owned sandbox boundary:
runwaits foryieldTimeMs(30 seconds by default), then returns a process id without terminating a command that is still running.pollimmediately,waitfor another bounded interval (five minutes by default), orkillthe command.SandboxProcesshandles and bounded output in the app runtime; Vercel uses detached command ids andgetCommand()to reconnect after app-runtime relocation.Yielded commands are pull-based and do not proactively start a new turn when they complete. This does not expand the public
SandboxSessionAPI, and recovery when an initial submission fails before its process id reaches durable history remains separate in #2575.Validation
Focused coverage exercises completion, yielding, polling, waiting, termination, cancellation, idempotent starts, capacity limits, bounded output, provider-compatible schemas, Vercel command reconnection, and sandbox cleanup. The full unit tier passed with 7,152 tests and one skip, and the hosted-runtime scenario passed all nine tests.
A manual smoke test used a real
openai/gpt-5.6-solagent against the Docker backend. A command yielded after 100 ms, reused one process id across three polls, accumulated output through completion, and returned exit code 0; its start marker remained exactly one. The local workflow transport also retried an inline step during this run, and recovery completed without duplicating the command. A separate termination smoke test returnedkilledand confirmed that a heartbeat file stopped growing after termination.Checklist
CONTRIBUTING.mdevepackagegit commit --signoff)Diff size
Docs — 2 files ·
+7 / -1Documents the user-facing tool behavior and includes the required release note.
Implementation — 6 files ·
+629 / -111Adds the managed-command registry, bounded stream capture, Bash lifecycle actions, Vercel reconnection adapter, and sandbox cleanup integration. Most of the new code is the eve-owned process and output lifecycle needed to avoid shell protocols while keeping provider handles out of the public
SandboxSessionAPI.Tests — 6 files ·
+488 / -43Covers the managed command lifecycle and schemas across unit, provider, scenario, and isolated prompt-cache fixtures, including retry identity and output truncation behavior.