Nate B. Jones / OB1
OB1 Agent Dashboard Walkthrough
A screenshot-first guide to the dashboard surfaces that make agent memory visible, reviewable, and useful for real work.
Personal continuity layer / OpenClaw launch demo
Dashboard screenshot
One operational scan

Dashboard

The dashboard answers the first operator question: what is in this OB1 and what needs attention now?
  • Total thoughts, type mix, and top topics show the current memory shape.
  • Workflow counts reveal active execution state without opening the board.
  • Recent activity confirms the newest continuity and launch notes are landing.
NBJ / OB1 01 / 07
Thoughts screenshot
Canonical memory list

Thoughts

Thoughts is the raw operating table: browse, filter, and inspect the durable records that the rest of the system builds on.
  • Filters let a user slice by type, source, or importance.
  • Nate proposal, OpenClaw registry, workflow, and audit seed data are visible together.
  • This page is useful when a user wants the memory graph, not just agent-written memory.
NBJ / OB1 02 / 07
Workflow screenshot
Task and idea continuity

Workflow

Workflow turns task and idea thoughts into a board so agents and humans can see what is new, active, in review, and done.
  • Cards preserve task context as memory instead of trapping it in a chat.
  • Priority and status make handoffs faster for the next agent or human.
  • OpenClaw launch work, screenshots, evals, and plugin distribution are represented as live work.
NBJ / OB1 03 / 07
Agent Memory screenshot
Governed write-back

Agent Memory

Agent Memory is where OB1 reviews what agents saved before those memories can influence future work.
  • Generated or inferred memories start as evidence, not instructions.
  • Review actions confirm, downgrade, restrict, stale, or reject a memory.
  • Nate continuity seed data shows how public context, repo map, and launch rules are kept auditable.
NBJ / OB1 04 / 07
Recall Trace screenshot
Debug retrieval

Recall Trace

Recall Trace shows what an agent asked for, what OB1 returned, and which memories the agent used or ignored.
  • This is the trust surface for debugging bad agent behavior.
  • Use policies travel with each recalled memory.
  • The request payload keeps runtime, model, project, query, and scope visible.
NBJ / OB1 05 / 07
Duplicates screenshot
Memory graph hygiene

Duplicates

Duplicates keeps repeated ideas from turning OB1 into a noisy dumping ground.
  • Similarity pairs compare both versions side by side.
  • Quality scores help decide which record should survive.
  • The walkthrough data includes real launch-adjacent duplicate patterns.
NBJ / OB1 06 / 07
Audit screenshot
Quality and safety pass

Audit

Audit surfaces low-quality thoughts so bad assumptions, unsafe storage ideas, and stale scope notes do not become hidden system behavior.
  • Low scores are easy to scan and bulk-select.
  • The seed rows intentionally include bad memory-policy ideas.
  • This reinforces the proposal's point: remembering the wrong thing is worse than forgetting.
NBJ / OB1 07 / 07