Weekday Query Tool

LLMs are frozen at their training cutoff. Ask one what today's date is, or what day of the week it falls on, and it'll either refuse, hedge, or confidently make something up - because it genuinely has no way to know. Weekday Tool is LLM Party's fix for that one specific, extremely common failure: it hands your agent a real, live "what day/weekday is it" function it can call instead of guessing.

Like Weather Tool and Work WeChat Tool, it's one of the pack's _tool-suffixed nodes, which means it doesn't answer anything on its own - it packages itself into a function-calling definition and wires into the tool input of an agent/LLM node that supports tool use. The model decides when to call it (usually because the user asked something date-relative, like "is the market open today" or "how many days until Friday"), and the actual lookup runs against your machine's clock, not the model's training data.

The two fields

• timezone (default Asia/Shanghai) - a plain string, and it's worth treating it as a real IANA timezone name (America/New_York, Europe/London, and so on) rather than a free-text label, since that's the format the default follows. Get this wrong and you're not going to get a helpful error dialog - you'll just get a wrong or broken answer out of the tool at call time, so it's worth double-checking the spelling before you move on.
• is_enable - a plain boolean here (unlike Weather Tool, which uses an enable/disable enum for the same job). Toggle it off and the tool vanishes from the agent's available functions with no error raised anywhere - if your agent used to answer "what day is it" and suddenly stops, this is the first thing worth checking.

The output is a single tool (STRING) socket, same shape as its siblings - it's a tool definition, not a date string, so it only does something useful plugged into a node that actually consumes tool-calling inputs.

Where it sits in the wider pack

The category path on this one - nested under "工具(tools)/实用(Utility)" - is basically the author's own admission that this is a boring utility node, not a headline feature. That's fair, and it's also exactly why it's worth using: date/time awareness is one of the most reliable places an otherwise-capable LLM agent falls over in front of a real user, and it's cheap to fix. If you're building anything conversational with LLM Party - a customer-facing assistant, a scheduling helper, anything where "today" matters - this is a five-second addition that closes a real gap rather than a nice-to-have.

Installing it

It ships inside the same single package as every other LLM Party node. Easiest path is ComfyUI Manager - search comfyui_LLM_party, install, restart. By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/heshengtao/comfyui_LLM_party

then pip install -r requirements.txt inside your ComfyUI environment (through python_embeded\python.exe if you're on the portable Windows build) and restart. This is a large, actively maintained pack that bundles chat, RAG, local model loading, and social-platform integrations all in one repo - so a fresh install can pull in a fair amount of dependency weight even though this particular node needs almost none of it. If pip install -r requirements.txt throws version conflicts, the pack also ships a requirements_fixed.txt you can fall back to.

Common issues

The two real ones are the same shape as the other tool nodes in this pack: make sure is_enable is actually true (it's a checkbox here, not a text toggle - don't go hunting for an enable/disable dropdown that this particular node doesn't have), and make sure the tool output is wired into a node that's actually built to accept tool-calling inputs - plugging it into a plain chat-only LLM node just won't connect, or won't do anything if it does. Beyond a bad timezone string, there's nothing else in the schema to misconfigure - this is about as low-surface-area as a node in this pack gets.