langchain-ai/deepagents · error · RequestError

-32002

-32002

Error message

Resource not found: session_id

What it means

`parse_model_allowlist` validates the raw TOML value of `[models].allowed` and raises `TypeError` when the value is not a list. The allowlist must be an array of spec strings; anything else (string, table, bare value) cannot represent an allowlist.

Source

Thrown at libs/acp/deepagents_acp/server.py:412

    async def load_session(
        self,
        cwd: str,
        session_id: str,
        additional_directories: list[str] | None = None,  # noqa: ARG002  # capability is not advertised
        mcp_servers: list[McpServer] | None = None,
        **kwargs: Any,  # noqa: ARG002  # ACP protocol interface parameter
    ) -> LoadSessionResponse:
        """Restore and replay a persisted ACP session."""
        if not self._load_sessions:
            method = "session/load"
            raise RequestError.method_not_found(method)

        self._session_cwds[session_id] = cwd
        agent = self._checkpointed_agent(session_id)
        metadata = (await agent.aget_state(self._session_config(session_id))).metadata or {}
        if metadata.get(_ACP_SESSION_METADATA_KEY) is not True:
            self._forget_session(session_id)
            raise RequestError.resource_not_found(session_id)
        if metadata.get("cwd") != cwd:
            self._forget_session(session_id)
            raise RequestError.invalid_params(
                {"cwd": "must match the working directory used to create the session"}
            )

        self._session_mcp_servers[session_id] = list(mcp_servers or [])
        self._initialize_session_options(session_id)
        if self._restore_session_options(session_id, metadata):
            self._reset_agent(session_id)

        await self._replay_session(session_id, agent)
        return LoadSessionResponse(
            modes=self._session_mode_states.get(session_id),
            config_options=self._build_config_options(session_id) or None,
        )

    async def set_session_mode(

View on GitHub (pinned to a1af029e6e)

Solutions

1. Wrap the value in a TOML array: `allowed = ["anthropic:claude-sonnet-4-5"]`
2. Check the `[models]` table — ensure `allowed` is not accidentally a sub-table header
3. Validate the TOML structure with a TOML linter/parser before loading

Example fix

# before
[models]
allowed = "anthropic:claude-sonnet-4-5"
# after
[models]
allowed = ["anthropic:claude-sonnet-4-5"]

allowed = data.get("models", {}).get("allowed")
if allowed is not None and not isinstance(allowed, list):
    raise TypeError("[models].allowed must be a TOML array of strings")

def is_allowlist_value(value: object) -> bool:
    return isinstance(value, list) and all(
        isinstance(e, str) and e.strip() for e in value
    )

try:
    parsed = parse_model_allowlist(raw)
except TypeError:
    logger.error("[models].allowed must be an array; got %s", type(raw).__name__)
    parsed = ()

When it happens

Trigger: Setting `allowed = "anthropic:claude-sonnet-4-5"` (a bare string instead of `["..."]`) or `allowed = { ... }` in `config.toml`; passing a non-list programmatically to `parse_model_allowlist`, e.g. via `coerce_toml_value`.

Common situations: Misreading the config schema and writing a single spec as a scalar instead of a one-element array; TOML table accidentally placed under `[models.allowed]` making it a dict.

Related errors

• -32602
• {name} must be a table
• context.auto_approve must be a boolean or null, got {type(au
• modes can only be provided when agent is a factory
• models can only be provided when agent is a factory

AI-assisted analysis of langchain-ai/deepagents@a1af029e6e (2026-08-29). Data as JSON: /api/errors/ce5e3b7b2aea1383. Report an issue: GitHub.