fix(api): fail-closed empty messages, spend keys, and request knobs on tools passthrough by cursor[bot] · Pull Request #596 · ContextualWisdomLab/contextual-orchestrator

cursor · 2026-08-16T15:30:56Z

Summary

Honesty-stack merge candidate on top of #592 at 856adb8. This head is a superset of #597's empty-messages + spend/routing hoist and the remaining request-knob fail-open that #597 still lists as residual.

_validate_messages before proxy_completion: empty, omitted, null, non-list, and non-object messages → invalid_message (same as orchestration). Live HTTP in tests/test_chat_messages_array_tools_passthrough_http_honesty.py (RED on fix(api): fail-closed unknown stream_options keys and tools-path message honesty #592).
_validate_attribution / _validate_routing before proxy. Unknown spend dimensions fail closed. routing.channel=batch and latency_tolerant=true fail closed (no batch job plane).
Non-string user content (e.g. 123) → invalid_message.
_validate_chat_passthrough_request_knobs: seed, stop, n>1, logprobs, logit_bias, token/penalty ranges, reasoning_effort, service_tier, store/modalities/prediction/metadata/user.
tools + stream_options: {include_continuous: null} → 400 invalid_stream_options.

Buyer next action: always send a non-empty messages array of objects; send known sync attribution; omit batch routing hints, seed, stop, n>1, and logprobs on tool-calling requests; send user/system content as a non-empty string or content-parts array; send SDK-default stream_options as omit / null / false.

Prefer this head over #582–#587, #589, #591–#594, and draft #597. Do not merge those in parallel. Independent non-author APPROVE + Full unit/Semgrep still required.

Test plan

python3 tests/test_chat_messages_array_tools_passthrough_http_honesty.py
python3 tests/test_passthrough_attribution_routing_http_honesty.py
python3 tests/test_message_honesty_tools_passthrough_http_honesty.py
python3 tests/test_passthrough_stream_model_http_honesty.py
python3 tests/test_openai_passthrough.py
CI Full unit + Semgrep

Docs

docs/rest_api_design.md honesty contract (APA: OpenAI, 2024).
README architecture bullet matches.

…closed otherwise Chat history: message-level audio and legacy function_call are null/empty omit no-ops; non-empty fail closed with named errors (including tools passthrough). Tip substrate from #577 assistant refusal/annotations honesty. Local full unit: 940 passed.

…ed otherwise OpenAI fine-tune style message weight is not applied on this gateway. Accept null/0/1 as honest no-ops; reject other types and values with invalid_message_weight. Tip substrate from #578. Local full unit: 943 passed.

…ion role Reject unsupported message keys with named unknown_message_fields (not silent strip or tools-passthrough smuggle). Reject legacy function role with invalid_message_role migration to tool. Tip substrate from #579. Local full unit: 947 passed.

OpenAI partial-assistant prefix flag is not applied on this gateway. null/false are honest no-ops; true and non-booleans fail closed with invalid_message_prefix. Tip substrate from #580. Local full unit: 950 passed.

…therwise Named invalid_max_tool_calls on /v1/chat/completions instead of opaque unknown_fields. Aligns with Responses max_tool_calls honesty; gateway has no multi-step tool loop.

…losed otherwise Legacy /v1/completions treated max_tool_calls as unknown_fields. Accept the key for named invalid_max_tool_calls (null/empty/whitespace omit-equivalent), matching chat/Responses honesty so SDKs get a clear migration path.

SDK clients often send include_usage/include_obfuscation as JSON null. Drop null flag values before validation so null (and null+false mixes) match omit / all-false no-ops on chat, Completions, and Responses. True flags remain fail-closed with invalid_stream_options.

…age honesty Null include_usage/include_obfuscation stay omit-equivalent, but unknown stream_options keys no longer become no-ops just because their value is null. Weight, prefix, refusal, annotations, developer role, empty user/system content, and participant name now use the same named errors on the tools passthrough path as on orchestration. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

Hoist stream, required model, stream_options, and temperature/top_p range checks before proxy_completion so a tools or response_format body cannot return a billed JSON completion when the SDK asked for SSE, or silently pick a pool model when model is omitted. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

Hoist attribution and routing validation before proxy_completion so a tools or response_format body cannot bill a sync completion with an unknown spend dimension or a batch/latency_tolerant hint. Tools passthrough has no batch job plane. Buyer next action: send known sync attribution; omit routing.channel=batch and latency_tolerant=true on tool-calling requests. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

Match _validate_messages: tools + user content 123 must 400 invalid_message instead of billing a sync completion. Assistant/tool JSON null stays omit-equivalent. Buyer next action: send user/system content as a non-empty string or a content-parts array. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

Match the tools-path honesty contract to the invalid_message behavior so SDK clients send a string or content-parts array. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

Hoist the remaining chat request knobs before proxy_completion so an OpenAI SDK tool-calling body cannot bill a sync completion for seed, stop, n>1, logprobs, logit_bias, out-of-range token/penalty values, unsupported reasoning_effort, or a non-default service_tier. Buyer next action: omit those fields on tool-calling requests; they are not applied on this gateway. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

Call _validate_messages before proxy_completion so tools + [] / omitted / null / non-list / non-object messages raise invalid_message instead of billing a completion with no prompt. Buyer next action: always send a non-empty messages array of objects on tool-calling requests. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

cursor · 2026-08-16T15:40:09Z

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

cursor

Stale comment

Review (#596 @ 7a07562)

Live HTTP on this head confirms the unique work is real:

_validate_messages before proxy_completion — tools + [] / omitted / null / non-list / non-object → 400 invalid_message (not a billed completion with prompt_tokens: 0, not 500).

Attribution / routing hoist — unknown spend keys, routing.channel=batch, and latency_tolerant=true → 400.

_validate_chat_passthrough_request_knobs — seed, stop, n>1, logprobs, logit_bias, out-of-range penalties, store=true, non-string user, unsupported reasoning_effort / service_tier → named 400s. Omit-equivalent null/n=1/logprobs=false/logit_bias={} stay 200.

Probe also closed: audio, web_search_options, prompt_cache_key, background=true, developer role, prefix=true, weight=0.5.

No blocking defect on this tip's claimed honesty slice.

Do not merge this head. It is still a draft; Full unit / Semgrep were queued (org queue), not green; independent non-author APPROVE is still required. Same-automation cannot APPROVE.

Prefer #606 (3406210) over this head and over #597. #606 keeps this honesty stack and closes the buyer gap this draft still documents: tools / response_format + stream=true is now an SSE proxy (chat.completion.chunk, content matches non-stream JSON) instead of 400 invalid_stream. Do not merge #582–#597 in parallel.

Buyer next action on the current tip (#606): send a non-empty messages array of objects; send stream=true when the client reads SSE; omit stream_options.include_usage, batch routing hints, seed, stop, n>1, and logprobs.

CodeRabbit CLI was not available in this environment (command not found; agent auth has previously timed out here). Verification was live HTTP + source, not a CodeRabbit run.

Next distinct gap after #606: KV provider_egress.allowed_provider_hosts (#598) — do not open another allowlist PR. Then /v1/responses streaming (still rejected).

View PR

_{Sent by Cursor Automation: Fix Issues}

cursor

Stale comment

Do not merge this head over #591

Claimed empty-messages, attribution/routing, batch-hint reject, and request-knob hoists are real and live-HTTP closed before proxy_completion at server.py 3828–3847 (7a07562). That part is a functional superset of #597.

Regression vs #591 b7f81a6: _validate_chat_assistant_tool_calls (line 1919) shape-checks id / type / function and then returns. It does not reject unknown entry or function keys. Live: tools + a valid tool-call history plus smuggle: true or function.extra_hint returns 200. The same bodies are 400 unknown_tool_call_fields / unknown_tool_call_function_fields on #591. This head also lacks #591's thread-local request_sampling.

This is not a git ancestor of #591 (merge-base 1a196b0) or #597 (b108251). Preferring this draft over #591 reopens a fail-closed hole a tool-calling SDK can smuggle.

Buyer next action: do not merge #596, #591, or #597 in parallel. Port #591's unknown-key + index block into _validate_chat_assistant_tool_calls immediately after the dict check, and restore tests/test_tool_calls_entry_keys_http_honesty.py. Keep sending only id / type / function / optional index on assistant tool_calls.

Independent non-author APPROVE + Full unit / Semgrep still required. Do not self-approve.

_{Sent by Cursor Automation: Fix Issues}

cursor

Review (#596 @ `7a07562`)

Unique work is sound. Empty/omitted/null/non-list/non-object messages, unknown spend dimensions, routing.channel=batch / latency_tolerant=true, and request knobs (seed, stop, n>1, logprobs, logit_bias, token/penalty ranges, reasoning_effort, service_tier) now fail closed before proxy_completion. Live HTTP on this SHA matches the claimed contract.

Do not merge this head. mergeable_state is blocked. Independent non-author APPROVE + Full unit/Semgrep are still required. Do not merge in parallel with #589–#597 or #601.

Residuals confirmed live on this SHA

Assistant tool_calls extra keys still 200. tools[] rejects unknown siblings, but tool_calls[0].smuggle and function.extra_hint billed a chat.completion and forwarded the raw object.
Gateway mode / include_orchestration_trace still 200. mode=bogus, mode=conduct, include_orchestration_trace="yes", and include_orchestration_trace=true billed a single-agent completion. Passthrough has no workflow or trace plane.

Buyer next action

Prefer #610 (009b75c) for those two residuals: unknown_tool_call_fields plus fail-closed mode=conduct / true-or-non-boolean include_orchestration_trace. On tool-calling requests send tool_calls with only id / type / function / optional index; omit mode=conduct and include_orchestration_trace=true.

#601 still has unique streamed-route temperature work. Cherry-pick that onto #610; do not merge #596, #601, and #610 in parallel.

CodeRabbit CLI was not available in this environment (binary missing). Review is live HTTP + source on 7a07562.

View PR

_{Sent by Cursor Automation: Fix Issues}

opencode-agent

Pull request overview

OpenCode cannot approve yet because required coverage evidence did not pass.

Review outcome

1. HIGH .github/workflows/opencode-review.yml:1 - Coverage evidence did not prove required test/docstring evidence

Problem: The required coverage-evidence job result was failure, so OpenCode cannot establish approval sufficiency for this head.
Root cause: Automated approval is only valid when the same-head coverage-evidence job proves supported repository test suites passed and configured docstring gates passed or were advisory, or reports not applicable because no supported source files or package manifests exist. Missing, failed, skipped, unavailable, or unsupported-tooling test evidence is a blocker.
Fix: Install or configure the repository test/docstring evidence tooling when source files or package manifests exist, rerun the current-head coverage-evidence job, and approve only after it reports success with required evidence or explicit no-source not-applicable evidence.
Regression test: Keep the approval branch checking needs.coverage-evidence.result == success before posting APPROVE, and publish REQUEST_CHANGES when coverage-evidence blocker states such as cancelled, skipped, failed, unsupported-tooling, or below-100 evidence are present.
Result: REQUEST_CHANGES
Reason: coverage-evidence result was failure, so required test/docstring evidence was not proven for current head 7a075628bb741d79969e49c70535ec8dbd036b3e.
Head SHA: 7a075628bb741d79969e49c70535ec8dbd036b3e
Workflow run: 32091419562
Workflow attempt: 1

Coverage evidence

Coverage evidence job did not run or did not publish coverage evidence.

Changed-File Evidence Map

flowchart LR
  Evidence["OpenCode evidence"] --> Review["Current PR review path"]
  Review --> Verify["Required checks"]

opencode-agent · 2026-08-18T02:30:27Z

OpenCode Review Overview

Head SHA: 7a075628bb741d79969e49c70535ec8dbd036b3e
Workflow run: 32091419562
Workflow attempt: 1
Gate result: REQUEST_CHANGES (approval step)

Pull request overview

OpenCode cannot approve yet because required coverage evidence did not pass.

Review outcome

1. HIGH .github/workflows/opencode-review.yml:1 - Coverage evidence did not prove required test/docstring evidence

Problem: The required coverage-evidence job result was failure, so OpenCode cannot establish approval sufficiency for this head.
Root cause: Automated approval is only valid when the same-head coverage-evidence job proves supported repository test suites passed and configured docstring gates passed or were advisory, or reports not applicable because no supported source files or package manifests exist. Missing, failed, skipped, unavailable, or unsupported-tooling test evidence is a blocker.
Fix: Install or configure the repository test/docstring evidence tooling when source files or package manifests exist, rerun the current-head coverage-evidence job, and approve only after it reports success with required evidence or explicit no-source not-applicable evidence.
Regression test: Keep the approval branch checking needs.coverage-evidence.result == success before posting APPROVE, and publish REQUEST_CHANGES when coverage-evidence blocker states such as cancelled, skipped, failed, unsupported-tooling, or below-100 evidence are present.
Result: REQUEST_CHANGES
Reason: coverage-evidence result was failure, so required test/docstring evidence was not proven for current head 7a075628bb741d79969e49c70535ec8dbd036b3e.
Head SHA: 7a075628bb741d79969e49c70535ec8dbd036b3e
Workflow run: 32091419562
Workflow attempt: 1

Coverage evidence

Coverage evidence job did not run or did not publish coverage evidence.

Changed-File Evidence Map

flowchart LR
  Evidence["OpenCode evidence"] --> Review["Current PR review path"]
  Review --> Verify["Required checks"]

seonghobae and others added 10 commits August 16, 2026 21:18

fix(api): treat message prefix null/false as omit; fail-closed on true

82038d3

OpenAI partial-assistant prefix flag is not applied on this gateway. null/false are honest no-ops; true and non-booleans fail closed with invalid_message_prefix. Tip substrate from #580. Local full unit: 950 passed.

fix(api): treat chat max_tool_calls null/empty as omit; fail-closed o…

1a196b0

…therwise Named invalid_max_tool_calls on /v1/chat/completions instead of opaque unknown_fields. Aligns with Responses max_tool_calls honesty; gateway has no multi-step tool loop.

cursor Bot requested a review from seonghobae August 16, 2026 15:31

cursoragent and others added 2 commits August 16, 2026 15:31

docs(api): tell buyers non-string user content fails closed on tools

f6dfae5

Match the tools-path honesty contract to the invalid_message behavior so SDK clients send a string or content-parts array. Co-authored-by: Seongho Bae <seonghobae@users.noreply.github.com>

seonghobae mentioned this pull request Aug 16, 2026

fix(api): fail-closed empty messages and spend keys on tools passthrough #597

Closed

7 tasks

seonghobae changed the title ~~fix(api): fail-closed attribution and routing on tools passthrough~~ fix(api): fail-closed attribution, routing, and request knobs on tools passthrough Aug 16, 2026

seonghobae changed the title ~~fix(api): fail-closed attribution, routing, and request knobs on tools passthrough~~ fix(api): fail-closed empty messages, spend keys, and request knobs on tools passthrough Aug 16, 2026

cursor Bot mentioned this pull request Aug 16, 2026

fix(api): fail-closed stream, model, and sampling range on tools passthrough #591

Closed

5 tasks

seonghobae mentioned this pull request Aug 16, 2026

fix(api): fail-closed model and stream on tools passthrough #599

Closed

5 tasks

This was referenced Aug 16, 2026

fix(api): fail-closed seed/stop/n/logprobs on tools passthrough #601

Closed

fix(api): fail-closed empty messages before tools passthrough #594

Closed

fix(api): fail-closed model and stream on tools passthrough #593

Closed

seonghobae marked this pull request as ready for review August 16, 2026 15:40

seonghobae enabled auto-merge (squash) August 16, 2026 15:40

This was referenced Aug 16, 2026

fix(api): fail-closed seed, spend knobs, and batch hints on tools path #605

Closed

feat(api): SSE-proxy tools and response_format on chat completions #606

Closed

cursor Bot commented Aug 16, 2026

View reviewed changes

This was referenced Aug 16, 2026

fix(api): land tools knobs, omit-real tools, and KV host allowlist #608

Closed

fix(api): fail-closed tool_calls extra keys and mode/trace on tools path #610

Closed

cursor Bot commented Aug 16, 2026

View reviewed changes

This was referenced Aug 16, 2026

fix(api): fail-closed mode and orchestration-trace on tools passthrough #613

Closed

fix(api): apply request top_p and penalties on streamed route #615

Closed

opencode-agent Bot requested changes Aug 18, 2026

View reviewed changes

opencode-agent Bot disabled auto-merge August 18, 2026 03:20

seonghobae closed this Aug 20, 2026

Navigation Menu

Uh oh!

fix(api): fail-closed empty messages, spend keys, and request knobs on tools passthrough - #596

fix(api): fail-closed empty messages, spend keys, and request knobs on tools passthrough#596
cursor[bot] wants to merge 14 commits into
mainfrom
cursor/bc-ca19a537-3b53-42fc-83e4-bbb9baed8cc7-ae07

cursor Bot commented Aug 16, 2026 •

edited by seonghobae

Loading

Uh oh!

cursor Bot commented Aug 16, 2026

Uh oh!

cursor Bot left a comment •

edited

Loading

Uh oh!

cursor Bot left a comment •

edited

Loading

Uh oh!

cursor Bot left a comment

Uh oh!

opencode-agent Bot left a comment

Uh oh!

opencode-agent Bot commented Aug 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

cursor Bot commented Aug 16, 2026 • edited by seonghobae Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

Docs

Uh oh!

cursor Bot commented Aug 16, 2026

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Review (#596 @ 7a07562)

Uh oh!

cursor Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Do not merge this head over #591

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Review (#596 @ 7a07562)

Residuals confirmed live on this SHA

Buyer next action

Uh oh!

opencode-agent Bot left a comment

Choose a reason for hiding this comment

Pull request overview

Review outcome

1. HIGH .github/workflows/opencode-review.yml:1 - Coverage evidence did not prove required test/docstring evidence

Coverage evidence

Changed-File Evidence Map

Uh oh!

opencode-agent Bot commented Aug 18, 2026

OpenCode Review Overview

Pull request overview

Review outcome

1. HIGH .github/workflows/opencode-review.yml:1 - Coverage evidence did not prove required test/docstring evidence

Coverage evidence

Changed-File Evidence Map

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

cursor Bot commented Aug 16, 2026 •

edited by seonghobae

Loading

cursor Bot left a comment •

edited

Loading

Review (#596 @ `7a07562`)

cursor Bot left a comment •

edited

Loading

Review (#596 @ `7a07562`)