ponens

Tutorials

A guided, hands-on course through the whole ponens methodology. You start with a real session from your own AI coding agent and, one lesson at a time, turn it into a curated, governed, reviewed, and cryptographically signed reasoning record. Every lesson builds on the same trace.json, so by the end you've run the entire loop end to end.

Prefer the big picture first? Skim the getting-started tour, then work through these in order. Already know what you need? Jump to a task-focused guide.

  1. 1 Capture your first trace

    Install ponens and turn your latest agent session into a reasoning trace — the two layers, and how to read it.
  2. 2 Curate the narrative

    Rewrite the raw session into a clean account of what was built and why — without touching the ground-truth actions underneath.
  3. 3 Declare the negative space

    Add the two things emission can't derive: the artifacts that make lineage real, and the residual surface — what you did NOT establish.
  4. 4 Goals & the three axes

    State a goal as a contract and read back met / governed / certified — how ponens decides 'done' from evidence, not prose.
  5. 5 Govern with policies

    Attach best-practice policies and turn `trace check` into a real gate — Computable Governance, block-by-default, waivers on the record.
  6. 6 Keep the evidence honest

    Grade the trace, watch proofs go Fresh / Stale / Detached as code changes, and record counter-evidence with defeaters.
  7. 7 Review and sign off

    Switch to the reviewer's seat: verify the consequential claims, certify the goal, and cryptographically sign the result.