[devin-ai-integration]

Summary

Applied the TDD skill (skills/tdd/SKILL.md) to packages/cli/tests and removed tests that don't catch any regression the remaining suite wouldn't. Tests only, plus deletion of dead helpers, so no changeset needed.

Dead code (tests + source removed)

• readStdinIfPiped / readStdinFrom in exec_helpers.ts have had no production caller since exec switched to streamStdinChunks. Deleted them and their 5 tests.
• chunkBytesBySize is now module-private. Its 5 direct tests duplicated the streamStdinChunks splitting and empty-chunk tests; maxBytes is a constant in production.

Duplicates

• sandbox/list.test.ts: dropped the JSON default-asc sort and "does not mutate input" tests. The table-path locale/tie-break/desc tests already go through sortSandboxes. Also dropped startedAt === startedAt.toLocaleString(), which just restated the implementation, and renamed that test to "formats state and metadata for display".
• create_lifecycle.test.ts: "lifecycle and timeout together" combined two mappings already tested on their own. "omits empty envs" was already asserted by the explicit-timeout test.
• fork.test.ts: dropped --count 101; 21 already covers values above the max.

Vacuous / never-run integration code

• backend_integration.test.ts "exec runs a command with piped stdin" accepted either 6 or 0, so it could not fail. Real piping coverage lives in exec_pipe.test.ts.
• exec_pipe.test.ts: removed the binary_random_sha256 case. It only ran behind E2B_PIPE_INTEGRATION_STRICT/BINARY/STRICT, which no workflow sets, and binary_nul_ff_hex already covers binary.
• Both integration files read userConfig.E2B_DOMAIN / userConfig.domain, fields that don't exist on UserConfig. Simplified to process.env.E2B_DOMAIN || 'e2b.app'.

Net: 265 → 249 tests. Offline, everything passes except template/create.test.ts, which fails the same way on main without E2B_API_KEY. backend_integration and exec_pipe were also run live with a key and passed.

Kept after review: the auth.test.ts stale-lock test, because it is the only test that runs the production CONFIG_LOCK. Also the table-path buildTableRows(..., 'desc') test.

Not changed (follow-up candidates): exec_passthrough, exec_stdin_flag, exec_close_stdin, create_lifecycle and project.test.ts mock the CLI's own modules (exec_helpers, src/terminal, src/utils/urls, src/api). They cover real behavior, but they are coupled to internals and would be better rewritten against a local HTTP server, as attribution.test.ts and template/delete.test.ts already do.

Link to Devin session: https://app.devin.ai/sessions/ba0ccf3e32064be4ad6191007b87c968
Open in Devin Desktop: https://app.devin.ai/desktop/session/ba0ccf3e32064be4ad6191007b87c968?variant=devin
Requested by: @mishushakov

---

🟢 Devin Review found nothing to fix · View findings

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring

[changeset-bot]

⚠️ No Changeset found

Latest commit: ad7ce5e

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[chatgpt-codex-connector]

Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-13T21:15:14.995805Z	cb76f8c	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[github-actions]

Package Artifacts

Built from c5d16df. Download artifacts from this workflow run.

JS SDK (e2b@2.64.1-devin-1794604309-prune-cli-tests.0, with @e2b/dockerfile-utils@0.1.3-devin-1794604309-prune-cli-tests.0):

npm install ./e2b-dockerfile-utils-0.1.3-devin-1794604309-prune-cli-tests.0.tgz ./e2b-2.64.1-devin-1794604309-prune-cli-tests.0.tgz

CLI (@e2b/cli@2.24.1-devin-1794604309-prune-cli-tests.0):

npm install ./e2b-cli-2.24.1-devin-1794604309-prune-cli-tests.0.tgz

Code Interpreter JS SDK (@e2b/code-interpreter@2.9.2-devin-1794604309-prune-cli-tests.0):

npm install ./e2b-code-interpreter-2.9.2-devin-1794604309-prune-cli-tests.0.tgz

Desktop JS SDK (@e2b/desktop@2.5.1-devin-1794604309-prune-cli-tests.0):

npm install ./e2b-desktop-2.5.1-devin-1794604309-prune-cli-tests.0.tgz

Python SDK (e2b==2.64.0+devin.1794604309.prune.cli.tests, with e2b-dockerfile-utils==0.1.2+devin.1794604309.prune.cli.tests):

pip install ./e2b_dockerfile_utils-0.1.2+devin.1794604309.prune.cli.tests-py3-none-any.whl ./e2b-2.64.0+devin.1794604309.prune.cli.tests-py3-none-any.whl

Code Interpreter Python SDK (e2b-code-interpreter==2.11.10+devin.1794604309.prune.cli.tests):

pip install ./e2b_code_interpreter-2.11.10+devin.1794604309.prune.cli.tests-py3-none-any.whl

Desktop Python SDK (e2b-desktop==2.7.2+devin.1794604309.prune.cli.tests):

pip install ./e2b_desktop-2.7.2+devin.1794604309.prune.cli.tests-py3-none-any.whl

[devin-ai-integration]

TASTE.md review: no violations (0). Checked T-49/T-50 (the new process.env.E2B_DOMAIN || 'e2b.app' treats an empty env var as unset), T-59/T-63 (chunkBytesBySize is now module-private, so its plain Error falls under the internal-helper exemption) and T-54 (the removed exec_helpers exports were never part of a package entry point). The rest of the diff only deletes tests.

Written by Devin

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.

[claude]

Nothing blocking. The comments below are optional suggestions. There is no need to push a fix for them before merging.

Apart from the inline findings, I checked two other things and neither is a problem.

• The removed exec_helpers exports have no importers left in packages/, and the CLI package publishes only a bin entry. Removing them doesn't change the public surface, so no changeset is needed.
• The integration tests no longer read the domain from ~/.e2b/config.json. This loses nothing: UserConfig (src/user.ts) has no E2B_DOMAIN or domain field, so that fallback could never supply a value.

Extended reasoning...

This change deletes tests in packages/cli and removes dead stdin helpers from exec_helpers.ts. It touches no security-sensitive production code paths. The inline findings already cover the coverage gaps, so this note only records the concerns that were checked and ruled out.

[mishushakov]

check findings ^

[devin-ai-integration]

Both findings were valid and are fixed in ad7ce5e:

• I restored the auth.test.ts stale-lock test, because it is the only test that runs the production CONFIG_LOCK.
• I restored the buildTableRows(..., 'desc') table test, because it is the only test that sends --order through the table path.

I didn't put back the toLocaleString() check: it just recomputes what the code does. That test is now named "formats state and metadata for display".