iflytek/astron-agent · warning · CotFormatIncorrectExc

无效的插件名称

Error message

无效的插件名称'{action}'

What it means

After extracting the Action name between the Action and Action Input markers, parse_cot_step validates it via is_valid_plugin(action). If the name does not match any registered/available plugin, it raises CotFormatIncorrectExc '无效的插件名称' (invalid plugin name). This prevents dispatching tool calls to nonexistent plugins.

Solutions

1. Keep the tool list in the system prompt generated dynamically from the actual plugin registry so names always match.
2. Validate the prompt's tool descriptions/names against is_valid_plugin for each; remove stale entries.
3. Add fuzzy/retry handling: on this exception, re-prompt with the list of valid plugin names.
4. Check agent/plugin configuration to ensure the intended plugin is installed and enabled for the tenant.

Example fix

// before (prompt hardcodes stale name)
Action: web_search_v1
// after (names from registry)
Available plugins: baidu_search, amap_weather ...
Action: baidu_search

valid_names = {p.name for p in agent.plugins}
# assert every tool name mentioned in the system prompt is in valid_names

try:
    step = await runner.parse_cot_step(content)
except CotFormatIncorrectExc as e:
    if "无效的插件名称" in str(e):
        content = await reprompt_with_valid_plugin_list(e, messages)
    raise

When it happens

Trigger: The model hallucinates a tool name, uses a display name/alias not in the registry, misspells a plugin name, or calls a plugin not enabled for the current agent/space.

Common situations: Prompt lists tools in a different naming scheme than the registry; plugin disabled or removed but still described in the prompt; model inventing tools like 'google_search' when only built-ins exist; case/underscore mismatches.

Understand the failure class

Background: 'Could not be found', 'does not exist', 'not found in database': the resource-not-found family when an ID, slug, key, or URI lookup comes back empty — this error's family across 20 libraries.

Related errors

• 无效的插件参数JSON格式
• 无效的推理格式，Action字段不完整
• 无效的推理格式，缺少必要的标识字段
• PERSONALITY_AI_GENERATE_ERROR
• MODEL_CHECK_FAILED

AI-assisted analysis of iflytek/astron-agent@5e758547a8 (2026-09-12). Data as JSON: /api/errors/d4924121921993ea. Report an issue: GitHub.

Appendix: source

Thrown at core/agent/engine/nodes/cot/cot_runner.py:191

        action_marker = self._find_marker(step_content, "action")
        action_input_marker = self._find_marker(step_content, "action_input")
        if (
            action_marker is None
            or action_input_marker is None
            or action_marker.end() > action_input_marker.start()
        ):
            raise cot_exc.CotFormatIncorrectExc("无效的推理格式，Action字段不完整")

        thought = ""
        thought_marker = self._find_marker(step_content, "thought")
        if thought_marker is not None and thought_marker.end() <= action_marker.start():
            thought = step_content[thought_marker.end() : action_marker.start()].strip()

        action = step_content[action_marker.end() : action_input_marker.start()].strip()
        action = action.strip("`*_")

        if not await self.is_valid_plugin(action):
            raise cot_exc.CotFormatIncorrectExc(f"无效的插件名称'{action}'")

        action_input_end = len(step_content)
        observation_marker = self._find_marker(step_content, "observation")
        if (
            observation_marker is not None
            and observation_marker.start() >= action_input_marker.end()
        ):
            action_input_end = observation_marker.start()
        action_input_raw = step_content[
            action_input_marker.end() : action_input_end
        ].strip()
        action_input = await self._parse_action_input(action_input_raw)
        return thought, action, action_input

    async def parse_cot_step(
        self, step_content: str, allow_plain_final_answer: bool = False
    ) -> CotStep:
        final_answer_marker = self._find_marker(step_content, "final_answer")

View on GitHub (pinned to 5e758547a8)