docs+test(custom-engine): finish http transport rollout#107
Merged
Conversation
The http transport (JSON request/response core, multipart http.files upload, and artifacts.files[].url download) has landed, but the docs still described it as "not yet implemented / planned / rejected by validation". Update the design doc status header and HTTP-transport section, the EN/ZH writing-evals guides (now with a worked http config example), and the stale SessionInput comment in custom.go to reflect that both local and http are supported. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The pipeline e2e only covered transport: local. Add
TestPipeline_CustomEngine_HTTPTransport, which stands up an in-process httptest
server, points ${CUSTOM_AGENT_ENDPOINT} at it via a new eval-http.yaml (reusing
the existing hello.yaml case), and asserts both that the framework POSTs a
SessionInput carrying the case prompt and that the returned SessionResult flows
into report.json with a PASS grading. Needs no POSIX shell fixture, so it also
runs on Windows.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Collaborator
Author
|
@codex review |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 7fde68b610
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".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Per review: uploads use the fixed form field name `files`, with the
workspace-relative path carried in each part's `filename` (CreateFormFile("files",
rel)). The previous wording ("keyed by its workspace-relative path") could lead
server authors to look for per-path form keys. Make the EN/ZH guides say the
server reads each `files` part's filename.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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
2 participants
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.
What
Finishing touches now that the Custom Engine
httptransport has fully landed (JSON core #99, multiparthttp.files#102,artifacts.files[].urldownload #103). The implementation works, but the docs and e2e suite hadn't caught up.Docs accuracy
docs/design/custom-engine.md: rewrite the stale "Phase 1 / not yet implemented / rejected by validation" status header and drop the "Not yet implemented in Phase 1" note above the HTTP transport section.docs/guide/writing-evals.md+docs/zh/guide/writing-evals.md: stop telling usershttpfails validation; add a workedcustom.httpconfig example (url/method/headers/files/request_body) plus the HTTP-specific rules (structured${session_input}injection, multipart payload, key-in-URL rejection, artifact URL download).internal/agent/custom.go: fix the stale "planned http transport" comment onSessionInput.Test coverage
e2e/custom_engine_test.go: addTestPipeline_CustomEngine_HTTPTransport. It stands up an in-processhttptestserver, points${CUSTOM_AGENT_ENDPOINT}at it via a neweval-http.yaml(reusing the existinghello.yamlcase), and asserts both that the framework POSTs aSessionInputcarrying the case prompt and that the returnedSessionResultreachesreport.jsonwith a PASS grading. No POSIX-shell fixture needed, so it also runs on Windows.No behavior change; the
[Unreleased]changelog already documents the feature.Verification
make verify→ 0 issuesgo test ./internal/agent/... ./internal/config/...→ okgo test -tags e2e -run TestPipeline_CustomEngine_... ./e2e→ all PASS (local, http, validate)Out of scope (separate follow-ups)
internal/customengineleaf package (config types still live ininternal/config/schema.go).agent_judgeruns on the case engine, breaking cross-protocol judge models.🤖 Generated with Claude Code