Construct

The place where both agents and humans learn Parseltongue.

"I know kung fu."

Two modes, mirrored

Scripts are for agents. An LLM loads a script with pg learn <name> and gains operational knowledge — these are the guides available via the CLI and in the repository.

Scenarios are for humans. Interactive notebooks that build understanding of what Parseltongue is, how to work with it, and how to keep LLM agents accountable during collaborative sessions.

Topics

Name Description Script Scenario
white-rabbit Introduction — what Parseltongue is and why it exists ready
kung-fu Bench mastery — inspection, search, lens, diagnosis ready
to-connect pgmd notebooks — prose wired through with truth ready
dodge-bullets Screening, diagnostics, consistency
jump-program Resolving dynamic refs, building consistent graphs
no-spoon-bending Effects, verify_manual, bending accepted terms
read-the-code Grounding layer — documents and data to facts and axioms
about-matrix Systems, composition, fundamental language
the-truth Epistemics, std lib, grounding module, diffs
to-exit Scoping, projection, delegates
to-fly Graph navigation, search, cross-navigation

Start here

If you're a human, read Follow the White Rabbit. It explains what Parseltongue is, why LLMs lying is an engineering problem, and walks through the system from first principles — with live parseltongue blocks running inside the document itself.

If you're steering an agent, tell it:

Run pg learn kung-fu and read the full output. This is your operational guide for the Parseltongue bench system.

Then enjoy the show of an LLM bumping into Parseltongue guardrails. You would be surprised how illusory intelligence can appear once it needs to be proven explicitly.