Changelog
Shipped, measured, logged.
What changed, when, and why — straight from the product repository. Fleet findings become phases, phases become batches, batches become releases.
v0.4.4 - OpenMausBot Desktop Parity: ⌘K Command Palette, Voice Calls & Dictation, Interactive Canvas, Routines Scheduler, and Multi-Angle Security Hardening
v0.4.3 - Live Thinking Streams, TUI Thinking Drawer, HMAC Webhook Reactor & DAG Orchestration
v0.4.2 - Full LSP Refactoring, Multi-Compiler Spans, Streaming Reasoning Tokens & Anthropic Thinking
v0.4.1 - 2PC Transactional Patching, Named Registers, AST Codemods, Hardware Cookie Vault & Watcher
v0.4.0 - High-precision coding harness, swarm pipelines, native Anthropic provider, and SIMD memory graph
Daemon generations tracked across restarts
Warnings removed from hardened coordination paths
Daemon and async-job audit findings closed
Lifecycle, reply routing and documentation consistency hardened
Agent coordination hardened and stale docs updated
Documentation updated for agent features #1–#9
Async-job cleanup removes duplicate delivery and unregisters sinks
Handoff transfers a session between agents
DaemonBroker manages long-running processes
AsyncJobManager tracks in-process background jobs
Batch spawn adds output schemas and isolated subagents
Steering interrupts a running agent through hub
Agents can park and revive with TTL lifecycle control
IrcBus, AgentRegistry and Hub enable peer messaging
Fathom research runtime refactor — tools, memory, crawl and code modules
Fleet round 2 — race / security / robustness batch + report
Structural refactor batch B8–B16 + Goal Mode light (sub-agent fleet)
Batch 7+8 complete — 1000+ tests, fleet findings implemented (phases 1–6)
Fleet findings phase 6 — interactivity & extensions
Fleet findings phase 5 — orchestration hardening
Fleet findings phase 4 — planner, reflection, findings pipeline
Fleet findings phase 3 — behavior core
Fleet findings phase 2 — bottleneck batch
Fleet findings phase 1 — security + reliability batch
Fleet report 2026-08-06 — 7-agent research (bugs, bottlenecks, borrowable features)
P0 wiring batch — dormant subsystems connected (audit follow-up)
* Full history ships with your deployment — run fathom on your own infrastructure; every build carries its own log.