fix(request): pin strict: false on function tools missing Responses strictness by SteveLTN · Pull Request #294 · ndycode/oc-codex-multi-auth

SteveLTN · 2026-10-10T07:59:56Z

Summary

Function tools forwarded to the Codex Responses backend now carry an explicit strict: false when the host left strictness unset, applied at a single choke point in transformRequestForCodex so both the native (default) and legacy request-transform modes are covered.
Why: OpenCode v2's AI SDK adapter (@ai-sdk/openai) omits strict when serializing function tools, and the Responses backend treats an omitted flag as strict mode — every schema property becomes effectively required. Models then fabricate values for optional parameters, most visibly sessionID on OpenCode's subagent tool, and fresh subagent creation fails with Subagent session not found / Session <id> is not a child of the current session. This is the same mechanism behind the GitLab Duo failures in anomalyco/opencode#51464, which GitLab fixed provider-side by defaulting unspecified Responses tool strictness to false (gitlab-ai-provider MR 48).
The behavior matches Codex CLI, which hardcodes strict: false on every function tool (codex-rs/tools/src/responses_api.rs), and OpenCode's own parity fix for its native providers (anomalyco/opencode#33392). Explicitly set strict values are preserved; hosted (non-function) tools are untouched.
Fixes [BUG] OpenCode V2 subagent calls fabricate sessionIDs when Responses tool strictness is omitted #293.

Testing

npm run lint
npm run build
npm run typecheck
npm test (5126 passed; 2 failures in request-hardening / storage-io-migration reproduce on unmodified main in this environment and are unrelated)
npm test -- test/doc-parity.test.ts
Not applicable

New coverage:

test/tool-utils.test.ts — unit tests for the new ensureFunctionToolStrictFalse helper: Responses flat shape, Chat Completions nested shape, explicit-value preservation, hosted-tool passthrough, non-array input.
test/fetch-helpers.test.ts — wire-level tests for transformRequestForCodex: native mode (classic shape), native mode on the responses-lite shape (tools inside the additional_tools input item), and legacy mode; all assert strict: false reaches the serialized body with required arrays intact.

End-to-end verification on a live install (OpenCode v2.0.26, plugin built from this branch): before the change, gpt-6-luna fabricated session IDs (ses_not_needed, …) and every fresh subagent call failed; after the change, fresh subagent delegation completes successfully.

Docs and Governance Checklist

README updated (if user-visible behavior changed) — not needed: internal request fix, no surface change
docs/getting-started.md updated (if onboarding flow changed) — not needed
docs/tools-and-cli.md updated (if the codex-* tool or CLI surface changed) — not needed
docs/configuration.md / docs/development/CONFIG_FIELDS.md updated (if config keys or env vars changed) — not needed: no new config keys or env vars
docs/upgrade.md updated (if migration or rename behavior changed) — not needed
SECURITY.md and CONTRIBUTING.md reviewed for alignment

Compliance Confirmation

This change stays within the repository scope and OpenAI Terms of Service expectations.
This change uses official authentication flows only and does not add bypass, scraping, or credential-sharing behavior.
I updated tests and documentation when the change affected users, maintainers, or repository behavior.

Notes

Linked issue: [BUG] OpenCode V2 subagent calls fabricate sessionIDs when Responses tool strictness is omitted #293
Follow-up work or rollout notes: the root cause sits in OpenCode v2's AI SDK adapter, which still omits strict on the current release line; this plugin-side fix is version-independent and matches Codex CLI behavior, so it stays correct even if OpenCode later fixes its adapter (an explicit strict: false from the host would simply be preserved as-is).

Summary by CodeRabbit

Bug Fixes
- Function tools without an explicit strict setting are now sent with strict mode disabled across native, legacy, and lite request formats. Existing strict settings and non-function tools remain unchanged.

note: greptile review for oc-chatgpt-multi-auth. cite files like `lib/foo.ts:123`. confirm regression tests + windows concurrency/token redaction coverage.

Confidence Score: 5/5

this change appears safe to merge.

Summary

defaults missing function-tool strict values to false before both request-transform modes.

Function tools without a strict setting keep optional arguments optional.

_{Reviews (1) · Last reviewed commit: "fix(request): pin strict: false on funct..." · Reviewed by Greptile}

…trictness OpenCode's AI SDK adapter omits `strict` when serializing function tools for the Responses API, and the backend treats an omitted flag as strict mode: every schema property becomes effectively required. Models then fabricate values for optional parameters — notably `sessionID` on OpenCode's subagent tool — and fresh subagent creation fails with "Subagent session not found" / "not a child of the current session". Fill in `strict: false` on function tools that don't set it, at the single choke point in transformRequestForCodex so both the native and legacy transform modes are covered. Explicitly set values are preserved; hosted (non-function) tools are untouched. This matches Codex CLI, which hardcodes strict: false on every function tool (codex-rs/tools/src/responses_api.rs), and OpenCode's own fix for its native providers (anomalyco/opencode#33392). Fixes ndycode#293

chatgpt-codex-connector · 2026-10-10T08:00:08Z

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-10T08:03:43.069330Z	`6498691`	PR opened

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.

coderabbitai · 2026-10-10T08:00:15Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository: ndycode/oc-codex-multi-auth/.coderabbit.yaml
Review profile: CHILL
Plan: Advanced
Run ID: 42d20c65-435e-4645-ae33-4bcb34985446

📥 Commits

Reviewing files that changed from the base of the PR and between 9dd0012 and 6498691.

📒 Files selected for processing (4)

lib/request/fetch-helpers.ts
lib/request/helpers/tool-utils.ts
test/fetch-helpers.test.ts
test/tool-utils.test.ts

Included review availability: This review used your included allowance. Your plan provides up to 2 included reviews per hour; 1 remain after this review.

📝 Walkthrough

Walkthrough

Function tools without an explicit strict value now receive strict: false before native or legacy request transformation. Existing strictness values and other tool fields remain unchanged.

Changes

Function Tool Strictness

Layer / File(s)	Summary
Strictness helper and utility tests `lib/request/helpers/tool-utils.ts`, `test/tool-utils.test.ts`	Added `ensureFunctionToolStrictFalse` for Responses- and Chat Completions-shaped function tools. Tests cover non-array inputs, non-function tools, and preservation of existing values.
Request transformation integration `lib/request/fetch-helpers.ts`, `test/fetch-helpers.test.ts`	`transformRequestForCodex` applies the helper before native or legacy transformation. Tests check canonical and serialized bodies, including the lite representation.

Priority: ➖ Normal

Estimated code review effort: 2 (Simple) | ~12 minutes

Change: Bug fix · Severity of issue fixed: Medium

Merge Risk: ⚪ Minimal · up to 64986

The change appears ready to merge after normal checks; no actionable risk remains from this review.

Pre-merge checks | 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check		The title clearly and concisely describes the main change: setting strict: false on function tools when Responses strictness is missing.
Description check		The description is complete. It explains what changed, why it was needed, testing results, new coverage, documentation impact, governance checks, compliance, and the linked issue.
Linked Issues check		Issue `#293` requires fresh subagent calls to omit `sessionID` and complete successfully. The PR applies `strict: false` only when function-tool strictness is unset, before both native and legacy reques…
Out of Scope Changes check		The reported changes are limited to the strictness helper, its integration in `transformRequestForCodex`, and focused unit and wire-level tests. These changes support issue `#293` directly. No unrelated…
Docstring Coverage		Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 3 functions across 4 files.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create a new PR

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

ndycode · 2026-10-11T13:26:22Z

Shipping verdict: PASS — verified independently against head 6498691 with CI green and no unresolved threads.

SteveLTN requested a review from ndycode as a code owner October 10, 2026 07:59

ndycode merged commit 078a26b into ndycode:main Oct 11, 2026
2 checks passed

Navigation Menu

Repository navigation

fix(request): pin strict: false on function tools missing Responses strictness - #294

fix(request): pin strict: false on function tools missing Responses strictness#294
ndycode merged 1 commit into
ndycode:mainfrom
SteveLTN:fix/responses-tool-strict-false

SteveLTN commented Oct 10, 2026 •

edited by greptile-apps Bot

Loading

Uh oh!

chatgpt-codex-connector Bot commented Oct 10, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Oct 10, 2026 •

edited

Loading

Walkthrough

Changes

Uh oh!

ndycode commented Oct 11, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

SteveLTN commented Oct 10, 2026 • edited by greptile-apps Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Testing

Docs and Governance Checklist

Compliance Confirmation

Notes

Summary by CodeRabbit

note: greptile review for oc-chatgpt-multi-auth. cite files like lib/foo.ts:123. confirm regression tests + windows concurrency/token redaction coverage.

Confidence Score: 5/5

Summary

Uh oh!

chatgpt-codex-connector Bot commented Oct 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codex Review Summary

Uh oh!

coderabbitai Bot commented Oct 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Uh oh!

ndycode commented Oct 11, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

SteveLTN commented Oct 10, 2026 •

edited by greptile-apps Bot

Loading

note: greptile review for oc-chatgpt-multi-auth. cite files like `lib/foo.ts:123`. confirm regression tests + windows concurrency/token redaction coverage.

chatgpt-codex-connector Bot commented Oct 10, 2026 •

edited

Loading

coderabbitai Bot commented Oct 10, 2026 •

edited

Loading