Hi @fleet mural , my agent's default model is lmstudio/qwen3.6-35b-a3b; and until mid-day yesterday there were no issues with it. Then, it suddenly started switching to a fallback, complaining about tools.
Evidence:
Agent consistently chooses a fallback
These logs always print:
20:16:46+00:00 warn agent/embedded {"subsystem":"agent/embedded"} [tools] No callable tools remain after resolving explicit tool allowlist (agents.e.tools.allow: *, lcm_describe, lcm_expand, lcm_expand_query, lcm_grep, nodes); the selected model does not support tools. Fix the allowlist or enable the plugin that registers the requested tool. 20:16:46+00:00 info agent/embedded {"subsystem":"agent/embedded"} [tool-result-truncation] Truncated 68 tool result(s) in session (contextWindow=80000 maxChars=16000 aggregateBudgetChars=16000 oversized=0 aggregate=68) sessionKey=agent:elton:discord:channel:__REDACTED__
20:16:46+00:00 info agent/embedded {"subsystem":"agent/embedded"} [context-overflow-precheck] early tool-result truncation succeeded for lmstudio/qwen/qwen3.6-35b-a3b route=truncate_tool_results_only truncatedCount=68 estimatedPromptTokens=53250 promptBudgetBeforeReserve=50000 overflowTokens=3250 toolResultReducibleChars=120513 effectiveReserveTokens=30000 sessionFile=/Users/ozzy/.openclaw/agents/e/sessions/1125d9fd-384b-4ece-adf2-6f168db3c09c.jsonl
20:16:46+00:00 info agent/embedded {"subsystem":"agent/embedded"} [context-overflow-precheck] early recovery route=truncate_tool_results_only completed for lmstudio/qwen/qwen3.6-35b-a3b; retrying prompt
Please wait for the rest of the logs and details.