What's shipped, what's next, what we're kanoodling on.
This page is our kanban board. Every phase we've finished, every item we're actively building, every half-formed idea sitting in the parking lot. Updated as often as the code changes — with dev notes in between for the thinking that doesn't fit a task row.
50+ items shipped, 20,000+ blocks produced.
The chain is running. Five validators across three continents producing blocks under real BFT consensus. Explorer, CLI wallet, Chrome extension, desktop app, playground, SDK — all live today.
The current focus — delegate, govern, distribute.
Staking v2 minimums / caps / unbonding shipped today. Remaining: commission, delegation, governance v2 tiers, and the ASE-balance coupling that makes bond actually move tokens. All blocking Phase 3.
Ideas worth remembering, not ready to commit.
Cross-contract async calls. Dynamic fee adjustment. Block-time tuning. DeFi primitives. The parking lot is where an idea lives until we have the evidence or the time to move it up the board.
What we're building right now.
Governance v2 tiers.
Split proposals into hard and soft. Soft for parameter tuning and treasury spends. Hard for invariants, VM libraries, and economic changes — 3-week window, 2/3 supermajority, 10% stake quorum. Hard floors stay unreachable.
Commission + delegation.
Per-validator commission rate (20% cap, 24h change cooldown), delegator accounts with pro-rata rewards and shared slashing, redelegate helper. Rest of Staking v2. Pairs with ASE-balance coupling work.
Operator app packaging.
Signed .dmg for macOS, .exe for Windows, .AppImage + .deb for Linux. Auto-update from release artifacts. Unblocks the download page and gets the Operator into non-dev hands.
‣ Dev note
“A chain that claims ‘anybody can validate’ can't halt when a laptop closes. So we made the committee a function of who actually showed up in the last 30 blocks.”
— after the 23h outage
Governance v2 — hard/soft tiers
Tier field on proposals, different bond + quorum + window + supermajority per tier. Hard floors stay governance-unreachable. Needed before Phase 3.
Commission + delegation
Per-validator commission (20% cap, 24h cooldown), delegator accounts with pro-rata rewards + shared slashing, redelegate helper. The rest of Staking v2.
ASE balance coupling for bond / withdraw
Today staking tracks nominal stake only; caller’s ASE balance isn’t debited on bond or credited on matured withdraw. Same chain-integration thread that makes value transfer real.
Operator app packaging
electron-builder → signed .dmg, .exe, .AppImage, .deb. Auto-update wiring. Unblocks the website /download page.
Governance UI in Operator
Create-proposal form and vote buttons in the Governance tab. Read-only today.
Website /download page
Aggregates all install paths (Chrome extension, Operator, CLI, VPS one-liner) with status chips. Pending until Operator packaging ships.
Recovery phrases across wallets
BIP-39 24-word format with Asentum-specific Dilithium3 derivation. Same format across extension + Operator + CLI.
‣ Dev note
“Design target is 2-second blocks. Runtime is 6. The right answer isn't a guess — it's whatever p99 vote propagation tells us when we have 50 validators instead of 5.”
— on block time tuning
The work that's already running in production.
Phase by phase, oldest first. None of this is speculative — every row below is code you can run or a block you can look up on the explorer today.
BFT consensus engine
Tendermint-style with pre-vote / pre-commit / 2/3 finality. Tested against a 4-validator local devnet before any network code was written.
Dilithium3 signatures
ML-DSA-65 (NIST FIPS 204) across every signature path: votes, transactions, block proposals. Zero classical crypto anywhere.
Hardened JavaScript VM
SES Compartments for contract execution, frozen primordials, no ambient authority. Deterministic across CPU architectures.
Sparse Merkle Tree state
BLAKE3-keyed over LevelDB. State root in every block header.
Staking + slashing as contracts
Both modules are system contracts, not special chain code. Upgradeable via governance, inspectable like any other contract.
‣ Dev note
“Four validators, 200 lines of code, zero network sockets. The whole BFT engine gated on a single Promise. It just worked.”
— building consensus before networking
testnet.asentum.com live
6-second blocks, BFT finality, ML-DSA-65 everywhere. Caddy + Let's Encrypt TLS. JSON-RPC open to all origins.
Block explorer (rebrand)
Dark editorial palette replacing the phosphor-green terminal look. Block / tx / address / contract detail pages, search, validator list.
Faucet
Drip test ASE to any address from the explorer, via RPC, or via the wallet UI.
MetaMask compatibility
Standard eth_* JSON-RPC methods work unmodified. MetaMask reads the chain; the Asentum Wallet does the Dilithium3 signing.
‣ Dev note
“Spent weeks fighting libp2p mesh negotiation. Then I just POSTed votes over HTTP to every peer. Less cool to say. Way more reliable in practice.”
— on killing libp2p
CLI + curl | sh installer
Single Node-SEA binary for macOS, Linux, Windows. One-liner drops it on $PATH and pre-configures the testnet RPC.
Native Chrome extension wallet
Manifest V3 with window.asentum provider. Password-encrypted vault, per-origin permissions, per-tx approval popups, multiple accounts.
Contract playground
In-browser IDE at /playground. Three templates, wallet connect, post-deploy method introspection with auto-generated forms.
@asentum/sdk
TypeScript SDK — AsentumClient, AsentumWallet, AsentumContract. ~14 KB bundled, works in Node and browser.
Asentum Operator desktop app
Electron app spawning a real validator as a child process. GUI onboarding, chain sync, staking form, system-tray background mode.
On-chain governance v1
Single-tier proposals with bond, quorum, voting window, execution timelock. VM library injection end-to-end.
Five validators, three continents
Four Hetzner CX22 boxes across Germany + US East + US West, plus a local Mac. BFT surviving real network jitter.
Graceful quorum transition
PENDING_BLOCKS = 10. New validators enter a pending window before being counted for consensus.
Kill-one survival verified
Shut down primary, three survivors kept producing blocks, primary rejoined and synced back. Zero tombstones, zero forks.
Liveness-aware committee
Committee = registered validators who proposed ≥ 1 of the last 30 blocks. Deterministic, home-validator tolerant. A closed laptop no longer halts the chain.
Minimum self-bond (500 ASE)
bond() rejects totals below the floor. Applies to initial bonds and partial tops. Exiting below the floor must be a full unbond.
Maximum self-bond (100k ASE)
Hard cap per identity. Want more voting power? Run another validator. Soft cap with diminishing rewards is on the roadmap once rewards ship.
14-day unbonding window
unbond() queues into pending withdraw with unlockAt = block + 604,800. Stake stays slashable during the window. Closes the slash-and-run hole.
withdraw() + getWithdrawable()
Claim matured withdrawals; wallets read the pending/matured breakdown for the UI. getStakingParams() exposes the four constants for config-driven frontends.
VPS one-liner installer
curl | sh for Ubuntu operators. Interactive onboarding + an asentum-validator command with status / balance / earnings subcommands.
Chrome Web Store resubmission
Get v0.3.0 back in front of reviewers. Listing copy, screenshots, privacy declarations.
Validator cluster expansion
More public validators, community-run. Pairs with staking v2 — delegation lets small operators run meaningfully.
docs.asentum.com + AIPs
Move docs to a subdomain. Formal Asentum Improvement Proposal process — governance-adjacent off-chain discussion.
Contract IDE
Evolve the playground. Multi-file projects, in-browser testing against the VM, deployment history.
CI pipeline for SEA builds
GitHub Actions matrix → artifacts on release tag → auto-rsync to /install/. Replaces manual local cross-build.
‣ Dev note
“No delegation? No mainnet. No hard/soft governance? No mainnet. Docs misaligned with code? No mainnet. The gate is in the list, not on a date.”
— on why mainnet isn't scoped
Cross-contract E() async calls
The big one. Async message-passing, the mechanism that makes reentrancy structurally impossible. Designed, not shipped.
WebSocket subscriptions
eth_subscribe for newHeads, logs, pending txs. HTTP-only today. Not hard, just hasn't been needed.
Dynamic EIP-1559 base fee
Static at 1 wei today because block utilisation has been trivial. Adjustment algo ready to drop in when congestion appears.
Service-node reward pool
15% of base fees diverted before burn, routed to service nodes. Needs per-request tracking + periodic payouts.
Block time 6s → 2s
Design target is 2s, runtime is 6s. Drop it once we have real-data for p99 vote propagation.
Historical balance queries
eth_getBalance(addr, blockTag) — needs state snapshots per block. Archive-mode only.
Per-opcode gas metering
Finer metering would price individual JS operations more fairly but adds real VM complexity.
Bond burning on slashing
Today a slash tombstones but doesn't burn. Burn closes the slash-and-run loophole. Trivial once burn fraction is signed off.
Code-signing certs
Apple Developer ID + notarization for macOS. EV Authenticode for Windows to kill SmartScreen. Budget question.
DeFi / AMM primitives
Reference AMM, reference lending, reference stablecoin. Not part of the protocol, but part of what needs to exist.
Mainnet launch
Phase 4. Not dated. We ship when the spec is closed, economics finalised, and external audits come back clean.
‣ Dev note
“Every chain shows you the curated ‘we're building this’ list. The parking lot is where we also show what we're NOT building — yet.”
— on keeping the lot visible
A chain that's built in public, one row at a time.
Join the launch list and we'll send you a note whenever something significant lands — never anything else. Or read the thesis if you want the why before the what.