Locsiclocsic.com
← Work In use

LocsicClaw

An interactive workspace for multi-agent collaboration

Built on OpenClaw, LocsicClaw lets agents work on parts of a task in parallel. A coordinating agent checks and combines their results. Trace a conclusion to a specific deliverable, or adjust a GenUI control in the conversation to see how an assumption changes the outcome.

In use2026-11-03

How agents work together

Consider two deployment options. You supply the workload and budget. One agent checks capacity while another estimates running costs. If the estimate uses the wrong request volume, the coordinator sends just that estimate back. Review notes refer to a particular version, so the verified capacity analysis still stands.

At handoff, the next agent can consult shared inputs, existing results and open questions. A completed subtask can wake the coordinator to continue. If a restart leaves a step in doubt, the agent checks saved work and execution receipts before repeating it.

Task activity: Judy is revising cost estimate v2; Atlas has completed capacity analysis v1.
Product screenshot with sample tasks. Atlas has completed capacity v1; Judy is revising the v2 estimate against the same request volume.

How people explore the result

An agent can answer a simple question in text, compare options in a table or show a trend in a chart. When the answer depends on assumptions, GenUI puts the inputs and the calculated result together in the reply.

A tested cost model uses CNY 12,000 in monthly fixed costs and CNY 0.03 per request. At 200,000 requests a month, charging CNY 0.08 produces a CNY 2,000 loss. Charging CNY 0.09 breaks even. In the workspace you can vary the volume, reload and resume from your last inputs, or copy the assumptions, formula and result together. The conversation can also host audio, video and a 3D model you can rotate and zoom.

At 200,000 monthly requests and CNY 0.09 per request, monthly profit is CNY 0.00: break-even.
Native GenUI · English interface preview with sample inputs at break-even.
Monthly requests200,000
Monthly profit (CNY)−2,000
Simplified calculator on this page. Monthly profit = requests × (unit price − variable cost per request) − monthly fixed costs.

Problem

How can agents work from consistent assumptions and trace each conclusion to the deliverable that supports it?

Approach

Agents share task inputs and return separate deliverables. The coordinator reviews each version and routes revisions to the agent responsible.

Next

Compare options in the conversation, change a cost assumption and see where the model breaks even.