Heima

The community decided. The chain executed. As finalized block 10,184,559, Heima Network completed the community-approved burn of totally 16,500,000 $HEI. verify it on the block explorer polkadot.js.org/apps/?rpc=wss:…

We update the device related architecture on Agentkeys, and add new web chat to the flow. A paired device no longer talks directly to an agent bridge. It keeps its signing key locally, receives scoped channel permission, and sends text, voice, and task requests through that shared path. The agent works in its isolated environment and replies through the same channel. Devices with narrower authority and one permissioned path. ⛳️

@amitmiat It is still in the internal test.

AgentKeys changed the agent identity architecture. A paired device no longer talks directly to an agent bridge. It keeps its signing key locally, receives scoped channel permission, and sends text, voice, and task requests through that shared path. The agent works in its isolated environment and replies through the same channel. Useful device. Narrower authority. One permissioned path. This is a big change and related to a lot of architecuture refactors, but it lays a solid foundation for the security and flexibility of future work.

FYI, the current block height is 10,176,678.

Updates on Agentkeys: speech to text(ASR, TTS) become more important as we use voice input for agent. However the access should be temporary and narrow. On the Amazon Web Services(AWS) stack, AgentKeys now checks an agent's speech permission before issuing cloud credentials limited to transcription and speech synthesis. The session is tagged to that agent and expires after 15 minutes. The broker does not hold a long-lived speech application token for this path. Source: `github.com/litentry/agent…`

@Ali06312390 It will be executed on block height: 10,184,559

@heimaNetwork Don't you think the time you gave for the cremation has expired ???

AgentKeys now has a simple flow for creating an agent without a QR pairing round trip🚀. The steps are as simple as: 1. Choose a preset. 2. Give the agent a name and fresh or inherited memory. 3. Approve once with Touch ID. The ceremony creates the agent with only its initial chat and memory access. Archive it later to return its agent slot. Agent slot will potentially cost $HEI One approval. A new agent. 🗒️It is a development progress, not a public launch. Source: github.com/litentry/agent…

AgentKeys now separates who can send updates from who can receive them. 🔐 A camera can send without reading the feed. A display can receive without permission to write. Narrow permissions make connected agents and devices easier to control. This is a Heima's preparation for the upcoming AI device era.

Good infrastructure doesn’t rely on operators making perfect decisions, and part of building infra is about making operator paths more explicit. ✅ That’s why we’ve been spending time on things like tightening how state is chosen and shared, and making sure production funding paths require an explicit opt-in before scripts move funds. They’re the kind of changes that reduce hidden assumptions, and make systems easier to inspect and operate over time. 🏗️

A quick update on the 16.5M HEI burn proposal. The burn is scheduled to execute at block #10,184,559 on the Heima Network. Based on the current average block time of approximately 6 seconds, the burn is expected to take place in around 10 days. We’ll continue to monitor the process and share another update once the burn has been successfully executed on-chain.

Permission is not trust. ❌ Many agent systems still blur the line between the two. AgentKeys treats permissions as temporary authority, not permanent trust. Operators explicitly define what an agent is allowed to do through scoped permissions, and those permissions can evolve as an agent’s responsibilities change. That authority can be updated, narrowed, or revoked, with permissions enforced at defined boundaries rather than assumed because an agent exists or holds a credential. 🛡️Trust is never assumed.

Some of the most important infrastructure work doesn’t look flashy, but it’s just as important. 💪 Over the past few weeks, much of AgentKeys development has been about exactly that. 💡Making permissions explicit instead of implied, reducing configuration drift, and narrowing the places where agents and operators can misunderstand what’s actually happening. They’re the kinds of changes that make systems easier to operate, easier to inspect, and easier to trust as they grow.

`setScope` is AgentKeys’ permission scope setting function. 🛡️ Instead of giving an agent unrestricted access, the operator can explicitly define what that agent is allowed to do. As an agent’s responsibilities change, those permissions can be updated or revoked. The first `setScope` on the Heima chain is another step toward making agent permissions explicit and auditable. 🤖

The first `setScope` (agentKeys permission scopes setting function) is on Heima chain now! explorer.heima.network/tx/0xb58e1239f…

Heima block time is now 6s, down from 12s.⚡️ Same chain. Faster block production. Quicker onchain actions and audits for agentkeys built on top.

Agentkeys just tightened how state is chosen and shared. 🧐 Making it less ambiguous at three different boundaries: who can read shared memory, which source ref a fleet deploy uses, and which chain the web app is actually connected to. • Delegated canonical-memory READ landed: a delegate reads the master's canonical memory through a signed CanonicalFetch path, with the worker doing the read and the CLI/harness/web surfaces wired around it. • Fleet deploys now choose the selected jj bookmark instead of drifting to a detached checkout. • The web header shows the daemon's live chain instead of a build-time fallback. • The permission-grant ceremony is documented as the familiar app-permission model: request, approve, enforce, revoke.

Over the weekend, AgentKeys landed another mechanics-focused slice, reducing hidden residue in the system. 🥳 Credential plaintext now lives in zeroizing buffers, and session debug output redacts bearer tokens. The fleet console shows active CI, loads panels incrementally, and can manage long-running deploy jobs. The master-register path now has a broker-built sponsored UserOp route: account factory initCode plus registerFirstMasterDevice in one batch. Less hidden state, fewer sharp defaults. 👀

Part of building infrastructure is making sure operator mistakes are harder to hide. 🛡️ We've been applying that principle across AgentKeys.👇 Production funding paths now require an explicit opt-in before scripts move funds. The fleet console puts health, balances, deploys, harness runs, and worktrees behind one operator UI. And pairing requests now expire visibly and repeated requests replace stale pending cards for the same device.