AI generated Meeting Summary - 2026-03-02

1. HBD Proposal Payout Under Haircut Rule

A ConvertDHFProposal change was implemented so that when HBD is near the haircut threshold, proposals pay out in Hive instead of HBD, converted at the current median price.
Discussion centered on naming a new field in the virtual operation to represent the converted Hive amount. Options discussed: converted vs converted_hive. The group leaned toward keeping it simple since the asset type is implicit.
Concern raised about downstream impacts on HAF apps and Balance Tracker, which may assume values are always in HBD.

Work has started on upgrading to PG-18, but it's low priority. New Docker images are needed because the CI pipeline currently breaks immediately.

Both features are ready for review and awaiting merge into HiveMind when the develop branch is stable enough.

Since AI can write code much faster than humans, verification (CI) became the bottleneck — previously 90+ minutes for some projects.
Significant effort over the past months to re-architect CI, bringing test times down to 10-30 minutes for major projects (HiveD, HAF, HiveMind).
HiveMind full replay time was reduced from 60 hours to ~20 hours (3x speedup), which greatly shortens the release validation cycle.

A new internal design of HAF was being tested in a parallel branch. It showed regressions and slowdowns for some apps (Balance Tracker, Block Explorer) while being slightly faster for HiveMind.
Decision deadline: Wednesday — likely outcome is sticking with the old HAF for this release to avoid risk.

Goal: tag a release candidate on Friday, test over the weekend, and release on Monday.
Only API-side/query changes will be included — anything affecting indexing performance will be deferred to a future release.
SYNCAD (testing environment) was not yet running the latest version; expected to be updated by Wednesday.

Wax library size cut by more than 2x, now ~1.7 MB — competitive with dhive.js.
Beekeeper (client-side transaction signing) cut by 8x, now ~300 KB.
Improvements came from optimizing the C++/Wasm codebase. Future gains expected from the approved Wasm 3.0 standard (better exception handling support).
New Wax version publicly released with documentation deployed to GitHub Pages.

Denser (front-end app) is close to a new release after switching to an industry-standard editor (same as HackMD). Hope to replace Condenser soon.
Block Explorer UI was redeployed and verified on a test environment.

When images.hive.blog went down, a community member swapped it for images.ecency.com as a quick fix.
The core team explained this breaks same-origin security policies and exposes users to risks (e.g., SVG-based XSS attacks). Their image proxy previously withstood such attacks due to specific security headers.
A longer-term fix is being worked on to harden the proxy and prevent future outages.

Team expanded to 9 people working on Magi.
NFT and token contracts released (composable smart contracts, ERC-20 based).
TSS (Threshold Signature Scheme) completed; first Bitcoin swap on testnet done. Native BTC on Magi launching in days.
Dash integration underway, with a partnership to bring external liquidity tied to HBD (~2 weeks out). Decred and Firo partnerships also in the pipeline.
CLP (slip-based) fee implementation for liquidity pools in progress.
UI redesign of the Ulterra app (Magi's front-end banking app) is underway.
Strategic focus: cross-chain DeFi and self-sustainability rather than social/outpost features.
A holistic roadmap was requested by another participant.

No significant core development to report. Hoping to bring on a developer in the coming months, currently limited by budget.

Integrated Hive-Engine tokens into Denser so anyone can create an outpost front-end using just an ENV file.
Supports three different reward contract types: legacy Steem Engine Scotbot, Hive-Engine Scotbot, and the newer Outpost contract.
Covers all ~52 existing token tribes. All 1,391 tests pass.
Changes are additive (minimal modifications to core code). A branch will be pushed for review, with detailed documentation of all changes.
Vision: Denser becomes the unified starting point for every L2 front-end, including future Magi/MetaMask login support.