NousResearch/hermes-agent · error · ValueError

Plugin LLM structured output did not match schema: {exc.mess

Error message

Plugin LLM structured output did not match schema: {exc.message}

What it means

Error "Plugin LLM structured output did not match schema: {exc.message}" thrown in NousResearch/hermes-agent.

Source

Thrown at agent/plugin_llm.py:591

    if not (json_mode or json_schema is not None):
        return None, "text"
    if not text:
        return None, "text"

    try:
        parsed = json.loads(_strip_code_fences(text))
    except (json.JSONDecodeError, ValueError):
        return None, "text"

    if json_schema is not None:
        try:
            import jsonschema  # type: ignore[import-untyped]
            jsonschema.validate(parsed, json_schema)
        except ImportError:
            # jsonschema is optional; skip strict validation when absent.
            logger.debug("jsonschema unavailable; skipping schema validation")
        except jsonschema.ValidationError as exc:  # type: ignore[attr-defined]
            raise ValueError(
                f"Plugin LLM structured output did not match schema: {exc.message}"
            ) from exc

    return parsed, "json"


# ---------------------------------------------------------------------------
# Usage extraction
# ---------------------------------------------------------------------------


def _extract_usage(response: Any) -> PluginLlmUsage:
    """Pull token usage out of an OpenAI-shaped response object.

    Tolerant of provider differences — Anthropic via the auxiliary
    adapter exposes ``usage.prompt_tokens`` / ``usage.completion_tokens``;
    direct OpenAI also exposes ``cache_read_input_tokens``."""
    usage = PluginLlmUsage()

View on GitHub (pinned to c896c09c42)

Solutions

1. Adjust the plugin's schema or prompt so the model output validates.
2. Review the schema validation error and fix the field mismatch.
3. Retry; structured output can fail intermittently on borderline prompts.

When it happens

Trigger: Thrown at agent/plugin_llm.py:591 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of NousResearch/hermes-agent@c896c09c42 (2026-08-14). Data as JSON: /api/errors/6a859cb02fe4c6d5. Report an issue: GitHub.