feat(codex): support reasoning effort targets config#1294
Merged
christso merged 1 commit intoJun 3, 2026
Conversation
Deploying agentv with
|
| Latest commit: |
be81c07
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://893e38a9.agentv.pages.dev |
| Branch Preview URL: | https://feature-codex-reasoning-effo.agentv.pages.dev |
ce93619 to
be81c07
Compare
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Codex targets can now configure SDK reasoning effort from
targets.yaml, so eval runs can use env-backed settings likeCODEX_MODEL=gpt-5.5withCODEX_REASONING_EFFORT=low. The wire field remains snake_case-only asmodel_reasoning_effort, resolves to internalmodelReasoningEffort, and is forwarded intostartThreadwith the SDK-supported enum values.The Codex SDK dependency was stale in the lockfile at
0.104.0, even though the shellcodexbinary was0.136.0. This PR bumps@openai/codex-sdk/bundled@openai/codexto0.136.0and makes the SDK provider honor the existing Codexexecutabletarget setting viacodexPathOverride, so targets can use profile shims such ascodex-eng.The root Codex target now includes
executable,model, andmodel_reasoning_effortenv bindings, and the Codex target docs list all three fields.Verification
bun test packages/core/test/evaluation/providers/targets.test.tsbun test packages/core/test/evaluation/providers/codex-sdk.test.tsbun test packages/core/test/evaluation/validation/targets-validator.test.tsbunx biome check .agentv/targets.yaml apps/web/src/content/docs/docs/targets/coding-agents.mdx apps/cli/package.json packages/core/package.json packages/core/src/evaluation/providers/codex.ts packages/core/src/evaluation/providers/targets.ts packages/core/src/evaluation/providers/types.ts packages/core/src/evaluation/validation/targets-validator.ts packages/core/test/evaluation/providers/codex-sdk.test.ts packages/core/test/evaluation/providers/targets.test.ts packages/core/test/evaluation/validation/targets-validator.test.tsbun run buildSmoke passed with:
AGENT_TARGET=codex LLM_TARGET=azure GRADER_TARGET=azure CODEX_EXECUTABLE=codex-eng CODEX_MODEL=gpt-5.5 CODEX_REASONING_EFFORT=low CODEX_LOG_DIR=.agentv/logs/codex bun apps/cli/src/cli.ts eval evals/self/azure-smoke.eval.yaml --target codex --workers 3Result:
PASS(2/2, mean 100%). Artifacts were written to:.agentv/results/runs/default/2026-06-03T11-24-27-757Z.agentv/results/runs/default/2026-06-03T11-24-27-757Z/index.jsonl.agentv/results/otel-2026-06-03T11-24-27-757Z.json.agentv/logs/codex/2026-06-03T11-24-28-160Z_codex_simple-math_attempt-1_985a8833.log.agentv/logs/codex/2026-06-03T11-24-28-182Z_codex_capital-of-france_attempt-1_2faff0e9.log