fix(kimi): send max_tokens, reasoning_effort, and thinking for Kimi/Moonshot - #13503
Merged
Conversation
…oonshot
Kimi/Moonshot endpoints require explicit parameters that Hermes was not
sending, causing 'Response truncated due to output length limit' errors
and inconsistent reasoning behavior.
Root cause analysis against Kimi CLI source (MoonshotAI/kimi-cli,
packages/kosong/src/kosong/chat_provider/kimi.py):
1. max_tokens: Kimi's API defaults to a very low value when omitted.
Reasoning tokens share the output budget — the model exhausts it on
thinking alone. Send 32000, matching Kimi CLI's generate() default.
2. reasoning_effort: Kimi CLI sends this as a top-level parameter (not
inside extra_body). Hermes was not sending it at all because
_supports_reasoning_extra_body() returns False for non-OpenRouter
endpoints.
3. extra_body.thinking: Kimi CLI uses with_thinking() which sets
extra_body.thinking={"type":"enabled"} alongside reasoning_effort.
This is a separate control from the OpenAI-style reasoning extra_body
that Hermes sends for OpenRouter/GitHub. Without it, the Kimi gateway
may not activate reasoning mode correctly.
Covers api.kimi.com (Kimi Code) and api.moonshot.ai/cn (Moonshot).
Tests: 6 new test cases for max_tokens, reasoning_effort, and
extra_body.thinking under various configs.
This was referenced Apr 21, 2026
teknium1
added a commit
that referenced
this pull request
Apr 22, 2026
Kimi's /coding endpoint speaks the Anthropic Messages protocol but has its own thinking semantics: when thinking.enabled is sent, Kimi validates the history and requires every prior assistant tool-call message to carry OpenAI-style reasoning_content. The Anthropic path never populates that field, and convert_messages_to_anthropic strips Anthropic thinking blocks on third-party endpoints — so after one tool-calling turn the next request fails with: HTTP 400: thinking is enabled but reasoning_content is missing in assistant tool call message at index N Kimi on chat_completions handles thinking via extra_body in ChatCompletionsTransport (#13503). On the Anthropic route, drop the parameter entirely and let Kimi drive reasoning server-side. build_anthropic_kwargs now gates the reasoning_config -> thinking block on not _is_kimi_coding_endpoint(base_url). Tests: 8 new parametric tests cover /coding, /coding/v1, /coding/anthropic, /coding/ (trailing slash), explicit disabled, other third-party endpoints still getting thinking (MiniMax), native Anthropic unaffected, and the non-/coding Kimi root route.
aj-nt
pushed a commit
to aj-nt/hermes-agent
that referenced
this pull request
May 1, 2026
…Research#13826) Kimi's /coding endpoint speaks the Anthropic Messages protocol but has its own thinking semantics: when thinking.enabled is sent, Kimi validates the history and requires every prior assistant tool-call message to carry OpenAI-style reasoning_content. The Anthropic path never populates that field, and convert_messages_to_anthropic strips Anthropic thinking blocks on third-party endpoints — so after one tool-calling turn the next request fails with: HTTP 400: thinking is enabled but reasoning_content is missing in assistant tool call message at index N Kimi on chat_completions handles thinking via extra_body in ChatCompletionsTransport (NousResearch#13503). On the Anthropic route, drop the parameter entirely and let Kimi drive reasoning server-side. build_anthropic_kwargs now gates the reasoning_config -> thinking block on not _is_kimi_coding_endpoint(base_url). Tests: 8 new parametric tests cover /coding, /coding/v1, /coding/anthropic, /coding/ (trailing slash), explicit disabled, other third-party endpoints still getting thinking (MiniMax), native Anthropic unaffected, and the non-/coding Kimi root route.
02356abc
pushed a commit
to 02356abc/hermes-agent
that referenced
this pull request
May 14, 2026
…Research#13826) Kimi's /coding endpoint speaks the Anthropic Messages protocol but has its own thinking semantics: when thinking.enabled is sent, Kimi validates the history and requires every prior assistant tool-call message to carry OpenAI-style reasoning_content. The Anthropic path never populates that field, and convert_messages_to_anthropic strips Anthropic thinking blocks on third-party endpoints — so after one tool-calling turn the next request fails with: HTTP 400: thinking is enabled but reasoning_content is missing in assistant tool call message at index N Kimi on chat_completions handles thinking via extra_body in ChatCompletionsTransport (NousResearch#13503). On the Anthropic route, drop the parameter entirely and let Kimi drive reasoning server-side. build_anthropic_kwargs now gates the reasoning_config -> thinking block on not _is_kimi_coding_endpoint(base_url). Tests: 8 new parametric tests cover /coding, /coding/v1, /coding/anthropic, /coding/ (trailing slash), explicit disabled, other third-party endpoints still getting thinking (MiniMax), native Anthropic unaffected, and the non-/coding Kimi root route.
gweeteve
pushed a commit
to gweeteve/hermes-agent
that referenced
this pull request
Jun 2, 2026
…Research#13826) Kimi's /coding endpoint speaks the Anthropic Messages protocol but has its own thinking semantics: when thinking.enabled is sent, Kimi validates the history and requires every prior assistant tool-call message to carry OpenAI-style reasoning_content. The Anthropic path never populates that field, and convert_messages_to_anthropic strips Anthropic thinking blocks on third-party endpoints — so after one tool-calling turn the next request fails with: HTTP 400: thinking is enabled but reasoning_content is missing in assistant tool call message at index N Kimi on chat_completions handles thinking via extra_body in ChatCompletionsTransport (NousResearch#13503). On the Anthropic route, drop the parameter entirely and let Kimi drive reasoning server-side. build_anthropic_kwargs now gates the reasoning_config -> thinking block on not _is_kimi_coding_endpoint(base_url). Tests: 8 new parametric tests cover /coding, /coding/v1, /coding/anthropic, /coding/ (trailing slash), explicit disabled, other third-party endpoints still getting thinking (MiniMax), native Anthropic unaffected, and the non-/coding Kimi root route.
waefrebeorn
pushed a commit
to waefrebeorn/slermes
that referenced
this pull request
Jul 2, 2026
…Research#13826) Kimi's /coding endpoint speaks the Anthropic Messages protocol but has its own thinking semantics: when thinking.enabled is sent, Kimi validates the history and requires every prior assistant tool-call message to carry OpenAI-style reasoning_content. The Anthropic path never populates that field, and convert_messages_to_anthropic strips Anthropic thinking blocks on third-party endpoints — so after one tool-calling turn the next request fails with: HTTP 400: thinking is enabled but reasoning_content is missing in assistant tool call message at index N Kimi on chat_completions handles thinking via extra_body in ChatCompletionsTransport (NousResearch#13503). On the Anthropic route, drop the parameter entirely and let Kimi drive reasoning server-side. build_anthropic_kwargs now gates the reasoning_config -> thinking block on not _is_kimi_coding_endpoint(base_url). Tests: 8 new parametric tests cover /coding, /coding/v1, /coding/anthropic, /coding/ (trailing slash), explicit disabled, other third-party endpoints still getting thinking (MiniMax), native Anthropic unaffected, and the non-/coding Kimi root route.
smasoftware
pushed a commit
to smasoftware/hermes-agent
that referenced
this pull request
Jul 18, 2026
…Research#13826) Kimi's /coding endpoint speaks the Anthropic Messages protocol but has its own thinking semantics: when thinking.enabled is sent, Kimi validates the history and requires every prior assistant tool-call message to carry OpenAI-style reasoning_content. The Anthropic path never populates that field, and convert_messages_to_anthropic strips Anthropic thinking blocks on third-party endpoints — so after one tool-calling turn the next request fails with: HTTP 400: thinking is enabled but reasoning_content is missing in assistant tool call message at index N Kimi on chat_completions handles thinking via extra_body in ChatCompletionsTransport (NousResearch#13503). On the Anthropic route, drop the parameter entirely and let Kimi drive reasoning server-side. build_anthropic_kwargs now gates the reasoning_config -> thinking block on not _is_kimi_coding_endpoint(base_url). Tests: 8 new parametric tests cover /coding, /coding/v1, /coding/anthropic, /coding/ (trailing slash), explicit disabled, other third-party endpoints still getting thinking (MiniMax), native Anthropic unaffected, and the non-/coding Kimi root route.
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
1 participant
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.
Salvages #13168 (@mengjian-github) onto current main with hostname-safe URL checks and upstream-faithful disabled-reasoning semantics.
Summary
Kimi/Moonshot endpoints stop returning
"Response truncated due to output length limit"— we now sendmax_tokens=32000, top-levelreasoning_effort, andextra_body.thinking={"type":"enabled"}, matching Kimi CLI'sgenerate()+with_thinking()defaults.Changes
run_agent.py_build_api_kwargs():max_tokens=32000+reasoning_effort.extra_body.thinkingfor Kimi/Moonshot hosts, independent of_supports_reasoning_extra_body().tests/run_agent/test_run_agent.py: 6 targeted tests covering kimi.com / moonshot.ai / moonshot.cn, custom effort, and disabled reasoning.Salvage adjustments on top of #13168
"api.kimi.com" in ...,"api.moonshot" in ...) withbase_url_host_matches(...)— matches the hostname-hardening sweep (dbb7e00, 5356797) already on main. Incidentally closeshttps://moonshot.ai.evil.com/v1andhttps://evil.com/api.moonshot.ai/v1false-positive paths (E2E verified).api.moonshot.cn(China endpoint) — samekimi-coding-cnroute already recognized inmodel_metadata.py.reasoning_config.enabled=False, omitreasoning_effortentirely (not "medium") — mirrors upstream Kimi CLI'swith_thinking("off")mappingreasoning_effort=None. Test updated to assertreasoning_effortis not present in that case.Note on max_tokens value
Upstream Kimi CLI default is
32000. Official Kimi docs list32768as the model default for K2.5/K2.6 and recommend128kfor reasoning /256kfor coding benchmarks. Starting conservative at32000to match CLI parity; we can bump this in a follow-up if coding tool-call workloads still truncate.Validation
TestBuildApiKwargstests pass (incl. 6 new).AIAgenton isolated HERMES_HOME: all three legit hostnames send the expected params; two attack-shape URLs (evil.com/api.moonshot.ai/v1,moonshot.ai.evil.com/v1) correctly receive nothing.Closes #13168.