Ask Your LLM
PIC defines a new ontology: authority is continuity, not possession. An LLM — like a human — tends to map a new model onto ones it already knows (OAuth, object-capabilities, RBAC/ABAC) and gets PIC subtly, confidently wrong. This is comprehension bias, and it is not only a machine problem: it applies to human readers too.
The box below assembles everything an LLM needs into one copyable text: the specifications, the formal model, the
machine-checked proof summary, and the interpretation rules (the LLM context file) that keep the model from filling
gaps with assumptions. The sources are fetched live from the repositories at page load, so the pack always
reflects the current main — nothing here is a maintained copy.
The documents in the pack are published under CC BY 4.0 and provided "as is", with no warranty of any kind —
the governing legal terms are those in pic-legal.md, included in the pack. The pack is a comprehension aid, not a
substitute for the sources: by using it you undertake to read and study the primary sources yourself. LLMs can
misread even a well-primed context, so no responsibility or guarantee is accepted for any model's behavior or
answers. You — the human — remain solely responsible for verifying and evaluating both the pack's contents and the
LLM's output, and for deciding what to accept and what to reject. "The LLM told me so" never transfers that
responsibility.
How to use it
- Press Copy context pack and paste the whole text into a fresh LLM conversation (or attach the downloaded
.txtfile). - The pack includes the LLM context file, which instructs the model to first prove it has understood every specification, then work through PIC's known points of scrutiny before answering.
- Ask your questions — then read the relevant sources and evaluate the answers yourself: the model helps you understand, the decision of what to accept stays with you.