Neuruh · Public Commons · Runnable Demo

Public Commons Chain Demo: Mission to Receipt

One governed mission run end to end, on a synthetic example. Observation to evidence to decision to authority to governed execution to receipt to outcome to calibration.

DocumentChain demo, v0.1

StatusSimulation in your browser

DataSynthetic example only

Live systemsNone touched

Honest status. This page is a browser simulation of the Public Commons mission-to-receipt chain. It runs no live system, calls no API, needs no API key, and touches no real data. Every parcel, document, payment, name, and number on this page is synthetic. The starter's own synthetic example file was not found on disk in this workspace, so the example below was built from the starter pattern as documented in the estate build inventory of 2026-11-02: seven core libraries, 33 public package repos, index repo Apache-2.0, starter listed at 90 tests. Do not treat this demo as the starter itself, and do not treat a demo receipt as a production receipt.

1. The chain this demo runs

The Public Commons chain, per its README, is: observation to evidence to decision (ALLOW, DENY, or ESCALATE) to authority to governed execution to receipt to outcome to calibration. Seven core libraries form the runnable path. This demo walks one mission through every link, and seals each step with a hash-chained receipt.

Library	Role in this run
`capability-registry`	Lists what the agent is allowed to do at all.
`policy-gate`	Decides ALLOW, DENY, or ESCALATE from evidence and policy.
`inference-health`	Checks the evidence is fresh and complete enough to decide on.
`governed-exec`	Runs only the action the gate approved, inside authority bounds.
`agent-receipt`	Seals each step into a hash-chained receipt.
`agent-run-manifest`	Records the whole run as one manifest: mission, steps, receipts, outcome.
`sovereign-agent-starter`	The runnable starting point. Its synthetic example pattern is what this demo follows.

2. Pick a synthetic mission

Three missions. Each one exercises a different gate outcome, so you can see ALLOW, ESCALATE, and DENY in one place.

Mission

Ready. No steps run yet.

3. Policy gate simulator

Try the gate on its own. The policy-gate decides from two conditions. DENY takes precedence over ESCALATE.

Evidence is stale (older than the 30 day freshness limit) Requested action exceeds the authority bound ($500)

Decision: ALLOW

Evidence is fresh and the request is inside authority. The gate allows execution.

4. Receipt log (hash-chained)

Each receipt seals one step. Each receipt's hash covers its step number, mission, inputs, outputs, and the previous receipt's hash, so changing any earlier receipt breaks every receipt after it. Hashing uses SHA-256 through the Web Crypto API when the browser provides it; otherwise a deterministic FNV-1a 64-bit fallback is used, and the page says which one ran. Current method: not yet computed.

No receipts yet. Run the mission first.

Receipt JSON

[]

5. What this proves, and what it does not

Proves: the chain shape works end to end in a browser: a mission can be decomposed, gated, executed inside bounds, receipted with a verifiable hash chain, and closed with an outcome and a calibration note.
Does not prove: that the Python libraries behave this way in production, that any real record was verified, or that a demo receipt is admissible evidence anywhere. The libraries' own test counts (starter: 90 tests, core libraries: 7 to 20 tests each) are the README's claims, not re-verified here.
Next step for the real thing: per-repo license audit across the 33 package repos, confirmed release tags, and a landing page linking the chain. That is the open-source release cleanup, not this demo.