Check the proof behind finished work
Check whether the proof given for a finished task is something someone else could go and check, or just a claim
Reads the proof recorded for one finished item and answers CHECKABLE when it names something a second person could look at (a command and its result, a count, a file, a measurement, a screen to open, a wall that was hit) or VAGUE when it asserts, restates the task, or describes a feeling or an intention, followed by one sentence saying why. It judges only whether the record can be trusted, not whether the work is good.
What it asks for
- evidence (Needed, words): The proof recorded for the finished item, in its own words
Use it from your AI
Connect the AI you already use to Trillion once. Then ask in your own words; your AI finds this tool by what it does and runs it.
Say to your AI: “Use Trillion to check whether the proof given for a finished task is something someone else could go and check, or just a claim”
What your AI calls: run_tool {"tool_id":"tool:check-proof-of-finished-work"}