🔒All inputs are processed locally in your browser. Nothing is transmitted, stored, or logged. Inputs disappear when you close the tab.

OCG-Industries Showcase · Cluster ④ AI Agents and Autonomous Systems

Agent Memory
Lineage

When an agent does something surprising, the useful question is which policy and which remembered state produced it. This tool derives a memory descriptor that stamps both, and flags a decision whose policy ancestor is missing. Same policy, same memory state, same decision label gives the same descriptor, so a decision log can be re-derived rather than trusted.

Data: no external data

Inputs STEP 1
Any label you use to identify the agent instance
The fingerprint of the policy or system prompt in force. Leave blank to see how an orphan decision is flagged.
A short label for what the agent decided
Leave blank if this is the first memory node.
The remembered context the agent acted on. Text only, no personal data.
--
--
Trace Status
--
Whether a policy fingerprint was supplied
Trace Depth
--
0 if this is the first memory node, else 1
Derived From
--
The policy declared in force, or none
Detail
Agent identifier--
Policy fingerprint--
State key--
Memory key--
Decision key--
Decision taken--
Trace depth--
Trace status--
Derived from policy--

Clear the policy fingerprint field and run again to see the orphan case. This is a decision log you can re-derive. It says nothing about model internals, weights, or why the model chose what it chose; it records what the operator declared was in force at decision time.

execution_hash: --
Policy Mandate v2.0 · what is this?