Two small follow-ups to the guard:
1. Fix the should_execute_tools docstring so it matches the actual code.
The previous version said "Only execute when finish_reason explicitly
signals tool intent" but the code also accepts finish_reason == "stop".
Explain why (some compliant providers emit "stop" with legitimate tool
calls — openai_compat_provider.py already mirrors this at lines ~633 /
~678 where ("tool_calls", "stop") are both treated as the terminal
tool-call state). Without this, a strict "tool_calls"-only guard would
regress 15 existing runner tests that construct LLMResponse with
tool_calls but no explicit finish_reason (default = "stop").
2. Add tests/providers/test_llm_response.py. This locks the three cases:
- no tool calls -> never executes
- tool calls + "tool_calls"/stop -> executes
- tool calls + refusal / content_filter / error / length / ... -> blocked
These are exactly the boundary cases the #3220 fix is about; without a
test here a future refactor could silently revert the guard.
Body + tests only, no behavior change beyond the existing PR's intent.
Made-with: Cursor
- Extract synthetic user message string to module-level constant
- Tighten comments in _snip_history recovery branch
- Strengthen no-user edge case test to verify safety net interaction
When _snip_history truncates the message history and the only user message
ends up outside the kept window, providers like GLM reject the resulting
system→assistant sequence with error 1214 ("messages 参数非法").
Two-layer fix:
1. _snip_history now walks backwards through non_system messages to recover
the nearest user message when none exists in the kept window.
2. _enforce_role_alternation inserts a synthetic user message
"(conversation continued)" when the first non-system message is a bare
assistant (no tool_calls), serving as a safety net for any edge cases
that slip through.
Co-authored-by: darlingbud <darlingbud@users.noreply.github.com>
When a subagent result is injected with current_role="assistant",
_enforce_role_alternation drops the trailing assistant message, leaving
only the system prompt. Providers like Zhipu/GLM reject such requests
with error 1214 ("messages parameter invalid"). Now the last popped
assistant message is recovered as a user message when no user/tool
messages remain.
When a non-transient LLM error occurs with image content, the retry
mechanism strips images from a copy but never updates the original
conversation history. Subsequent iterations rebuild context from the
unmodified history, causing the same error-retry cycle to repeat
every iteration until max_iterations is reached.
Add _strip_image_content_inplace() that mutates the original message
content lists in-place after a successful no-image retry, so callers
sharing those references (e.g. the runner's conversation history)
also see the stripped version.
Keep tool-call assistant messages valid across provider sanitization and avoid trailing user-only history after model errors. This prevents follow-up requests from sending broken tool chains back to the gateway.
Resolved conflict in azure_openai_provider.py by keeping main's
Responses API implementation (role alternation not needed for the
Responses API input format).
Made-with: Cursor
Merge the three retry-after header parsers (base, OpenAI, Anthropic)
into a single _extract_retry_after_from_headers on LLMProvider that
handles retry-after-ms, case-insensitive lookup, and HTTP date.
Remove the per-provider _parse_retry_after_headers duplicates and
their now-unused email.utils / time imports. Add test for retry-after-ms.
Made-with: Cursor
Register xiaomi_mimo as an OpenAI-compatible provider with its API base URL,
add xiaomi_mimo to the provider config schema, and document it in README.
Signed-off-by: Lingao Meng <menglingao@xiaomi.com>
Some LLM providers (OpenAI-compat, Azure, vLLM, Ollama) reject requests
with consecutive same-role messages or trailing assistant messages. Add
_enforce_role_alternation() to merge consecutive same-role user/assistant
messages and strip trailing assistant messages before sending to the API.
Replace the flatten/unflatten approach (merging extra_content.google.*
into provider_specific_fields then reconstructing) with direct pass-through:
parse extra_content as-is, store on ToolCallRequest.extra_content, serialize
back untouched. This is lossless, requires no hardcoded field names, and
covers all three parsing branches (str, dict, SDK object) plus streaming.
Preserve the provider and agent-loop streaming primitives plus the CLI experiment scaffolding so this work can be resumed later without blocking urgent bug fixes on main.
Made-with: Cursor
Replace the static provider-level supports_vision check with a
reactive fallback: when a model returns an image-unsupported error,
strip image_url blocks from messages and retry once. This avoids
maintaining an inaccurate vision capability table and correctly
handles gateway/unknown model scenarios.
Also extract _safe_chat() to deduplicate try/except boilerplate
in chat_with_retry().
- Remove trailing whitespace and normalize blank lines
- Unify string quotes and line breaks for long lines
- Sort imports alphabetically across modules
