The Flows, Memory, and Voice Power‑Up

This release is packed with deep workflow-building upgrades, a leaner kernelized core, and big reliability wins across memory, inference, and voice—plus a bunch of UI polish that makes everything feel snappier and sturdier. 🎉

Highlights

Flows & Workflow Copilot, leveled up 🧩

Workflow authoring and Copilot got a major usability and reliability pass: clearer step labels, a “Save & enable” path, safe live test-runs behind approvals, a Stop button, better run-rail behavior, improved compile/runtime correctness, and a visual refresh across the canvas. There’s also more robust store behavior (corrupt-row handling, transactional upserts), better localization coverage, and tighter run/approval boundaries so what you approve is what runs.

(#5088, #5089, #5090, #5097, #5099, #5111, #5112, #5114, #5115, #5120, #5128, #5135, #5148, #5150, #5153, #5176, #5205, #5212, #5223, #5227, #5228, #5248, #5264, #5265, #5285, #5286, #5287, #5289, #5290, #5292, #5293, #5294, #5296, #5297, #5302, #5303, #5304, #5366) — Thank you @graycyrus and @senamakel!

Memory & agent profiles that scale 🧠

Memory took a big step forward: TinyCortex consolidation and tool-memory cutovers continue, identifiers are normalized more reliably, and a new warning helps prevent “silent” memory stoppage when embedding budgets are exhausted. Flows also gained more intentional memory behaviors (flow namespaces + digest, dedicated memory node support, and read-only memory helpers), while profiles gained “hermes-style” homes that neatly organize per-profile workspace, memory, skills, and cron attribution.

(#5043, #5101, #5118, #5140, #5175, #5176, #5205, #5220, #5227, #5237, #5244, #5246, #5275, #5280, #5282, #5368, #5398, #5402, #5427) — Thank you @senamakel, @graycyrus, @YellowSnnowmann, @CodeGhost21, and @mysma-9403!

Voice, mascot, and realtime experiences 🎙️

Voice capture and TTS reliability improved (including better sample-format handling, Piper binary fixes, and cleaner provider behavior), and realtime voice mode is now surfaced by default. There’s also a big new integration: ElevenLabs Voice Agents with core RPC, realtime sessions, and a local relay handler—paired with UX improvements like sentence-by-sentence streaming to reduce perceived delay.

(#4537, #5216, #5367, #5404, #5407, #5428, #5441) — Thank you @mysma-9403 and @YellowSnnowmann!

Chat, conversations, and day-to-day UI polish 💬

Chat and conversation UX got smoother and sturdier: long URLs now wrap correctly, a render loop was fixed, template rejections are clearer, and the thread-goal footer no longer auto-scrolls unexpectedly. The “Human” experience also evolves: the mascot can now live docked inside chat, supports uploaded custom avatars, and avoids lipsync-related UI lockups—plus you can name a thread in three words.

(#5030, #5210, #5274, #5309, #5369, #5412, #5419, #5420) — Thank you @jgentes, @graycyrus, and @YellowSnnowmann!

Inference, embeddings, search & cost correctness ⚡

Inference and embeddings behavior is more predictable and safer: better model routing (including vision-capable models), more actionable errors (and redacted secrets), improved local model reliability, and clearer handling for BYOK keys and budgets. Search expands with Exa support (managed + BYOK), and attribution is improved so usage is correctly linked to resolved providers.

(#4885, #5103, #5117, #5206, #5207, #5215, #5218, #5219, #5222, #5253, #5254, #5262, #5363, #5371) — Thank you @M3gA-Mind, @YellowSnnowmann, and @samrusani!

Core kernelization, feature gates, and performance 🧱

The core continues getting leaner and more modular: big dependency trimming, kernelization work, DomainGroup realignment, and more functionality gated behind explicit compile-time features. Performance and orchestration also improved with batching, shared config arcs, checkpoint/resume for delegations, and runtime process pooling across agents and skill runs.

(#5029, #5040, #5052, #5061, #5062, #5068, #5078, #5079, #5129, #5178, #5202, #5233, #5240, #5314, #5328, #5332, #5364, #5365, #5396, #5405) — Thank you @senamakel, @YellowSnnowmann, @oxoxDev, and @mysma-9403!

Agents, approvals, and orchestration safety 🤖

Agents and approvals got sharper edges removed: replacement routes are preserved during cleanup, path scrubbing is safer, an opt-in “auto-approve-all” setting is available, and custom registry agents now run with their real tools. Orchestration reliability improves in several places too, including cron correctness, removing dead hosted-steering polling, and making delegation/agent behaviors fail loud instead of silently dropping results.

(#4786, #4998, #5039, #5042, #5094, #5121, #5123, #5211, #5306) — Thank you @graycyrus, @samrusani, @mysma-9403, @oxoxDev, @sanil-23, and @M3gA-Mind!

Desktop packaging, config/auth resilience & build health 🛠️

Startup and sign-in flows are more resilient: unreadable configs can be healed and explained, transient store reads won’t reset profiles, and session expiry remains classifiable. Desktop packaging got targeted fixes (AppImage paths, userns relaxation, D‑Bus session bus), Tauri window behavior is improved for mixed-DPI multi-monitor setups, and CI/test coverage/stability got meaningful upgrades.

(#4716, #4772, #5081, #5083, #5171, #5181, #5189, #5190, #5221, #5241, #5252, #5255, #5267, #5276, #5277, #5311, #5336, #5337, #5340, #5341) — Thank you @YellowSnnowmann, @senamakel, @M3gA-Mind, @samrusani, @myi1, and @senamakel!

Medulla, SDK routing, and backend-facing plumbing 🌐

OpenHuman is increasingly positioned as the Medulla core, with new facades/RPC surfaces and workflow-plane serving to the hosted orchestrator. The backend client is now routed through a vendored tinyhumans SDK, managed backend traffic is cut over, and several Medulla/MCP catalog and URL-resolution fixes help keep things consistent end-to-end.

(#5105, #5229, #5232, #5233, #5234, #5238, #5245, #5260, #5266, #5361) — Thank you @senamakel and @graycyrus!

Tiny Place & agent-world improvements 🗺️

Tiny Place and agent-world capabilities continue to mature with directory search + pagination, ledger filters (with copyable tx-ids), better handle transfer wiring, and clearer framing/documentation of Tiny Place World as a local sim with a relay seam. Some identity/feed media behaviors are now explicitly web-only where appropriate.

(#4998, #5192, #5193, #5271, #5278, #5295) — Thank you @CodeGhost21 and @M3gA-Mind!

Dev tooling, bench harnesses, and docs 📚

Developer experience improved with new benchmarks and profiling environments, a logs-first tabbed TUI, and broader CI validation. Documentation also got meaningful updates: CITATION metadata, install fixes (macOS Homebrew cask), contributing guidance for Windows desktop setup, and updated search docs for Exa.

(#5060, #5107, #5131, #5142, #5145, #5152, #5177, #5215) — Thank you @YellowSnnowmann, @senamakel, @mwakidenis, @NgoQuocViet2001, and @M3gA-Mind!

UI foundations, cleanup, and quality-of-life ✨

There’s a lot of thoughtful cleanup and UI groundwork here: reusable React foundations, removal of unused/orphaned modules and files, and multiple passes to slim the core and relocate desktop features appropriately. You’ll also see small but delightful touches like a dismissible in-app GitHub star CTA and safer theme/gradient teardown behavior.

(#5095, #5102, #5110, #5113, #5124, #5125, #5130, #5141, #5143, #5185, #5186, #5188, #5206, #5273, #5279, #5310, #5343) — Thank you @graycyrus, @senamakel, @sanil-23, @M3gA-Mind, and @YellowSnnowmann!

Additional highlights 🔗

A few more focused contributions round out this release with targeted fixes and improvements that are worth calling out. (#5154, #5217, #5235, #5236, #5239, #5283, #5288, #5335, #5338, #5370, #5401, #5411) — Thank you @senamakel, @M3gA-Mind, @graycyrus, and @YellowSnnowmann!

New Contributors

• Thank you @Horst1993 — for tightening agentbox configuration handling with better base URL normalization. (#5091)
• Thank you @jgentes — for a conversations UX fix that stops the thread-goal footer text from auto-scrolling. (#5030)
• Thank you @mwakidenis — for adding citation metadata via CITATION.cff. (#5142)
• Thank you @myi1 — for a Windows keyring build fix (msg mutability under cfg(windows)). (#4772)
• Thank you @NB213 — for helping improve the project (first contribution noted in this release range).

Hope you’ll join the Discord—there are exclusive roles and contributor rewards waiting for you!

Contributor Credits

Thanks to everyone who shipped this release: CodeGhost21, Cyrus Gray, github-actions[bot], Horst1993, James Gentes, Mega Mind, Muhammad Ismail, mwakidenis, mysma-9403, nb213, NgoQuocViet2001, oxoxDev, Sam, sanil-23, Steven Enamakel, YellowSnnowmann.

Full Compare

v0.63.12...v0.63.7

The Smooth Release Bump Upgrade

v0.63.12 is here—a clean, streamlined release update to keep everything moving forward. 🎉

Highlights

Release automation & housekeeping ⚙️

This update is focused on release hygiene: bumping the version to v0.63.12 via the automated release workflow, with no feature or bugfix PRs included in this range. — Thank you @github-actions[bot]!

Contributor Credits

Thanks to @github-actions[bot] for shipping this release update.

Full Compare

v0.63.11...v0.63.12

OpenHuman v0.63.11

187 commits — 52 features, 90 fixes, 45 refactors, docs and CI. This release carries everything merged to main since v0.63.9, including the kernel restructure, ElevenLabs Voice Agents, and a large reliability pass across memory, inference and voice.

Note on the previous notes: an earlier version of this page listed only the two release-promotion PRs. That was an artefact of how release is updated, not a description of the release. The full contents are below.

Highlights

Voice

• ElevenLabs Voice Agents — core RPC, realtime session, and local-agent relay handler (#5407)
• Send the relay identity token where the backend can read it (#5428)
• Stop the Piper voice id leaking into non-Piper TTS providers (#5367)
• Repair the execute bit on extracted Piper binaries (#5216), handle U16 sample format in the fallback capture path (#4537)

Memory & embeddings

• Warn before the embedding budget silently stops memory (#5402)
• Stop a signed-in user's embeddings failing, and unstick the parked queue (#5427)
• Classify Ollama-unavailable embed failures actionably (#5398)
• Drain the coding-session backlog across bounded passes (#5368)
• Canonicalize memory identifiers symmetrically (#5275); stop paging on caller ingest-payload schema errors (#5280)

Inference & model routing

• Route Codex OAuth chat to the subscription backend (#5371)
• Error on a misconfigured vision model instead of substituting (#5262); route vision at vision-capable models (#5253)
• Explain background-role provider fallback and verify BYOK keys (#5254)
• Route local AI through TinyAgents RPC (#5240); local model reliability — /api/tags fallback and real Ollama model ids (#5219)

Flows

• Surface bounded loops from the tinyflows engine (#5366)
• Accept declared workflow inputs on every run path (#5304)
• Per-item fan-out — run an array of flow work concurrently (#5302)
• Canvas visual refresh — kind tiles, elevation, step numbers (#5303)

Agent & chat

• Send user image attachments as image blocks, not text (#5370)
• Preserve chat context across a Quick/Reasoning switch (#5401)
• Name a thread in three words (#5412)
• Name the chat-template rejection instead of blaming model/temperature (#5309)
• Filesystem-offload convention for worker artifacts on long-horizon tasks (#5217)

Kernel restructure

• Collapse 124 flat domains into 31 gate-aligned families (#5328) — pure moves, byte-identical wire surface
• Kernelize openhuman: cut the workflow dependency floor 418 → 280 crates (#5314)
• Realign DomainGroup with the family directories (#5332) and guard three unenforced consumers (#5364)
• Gate the macOS Contacts cohort behind contacts (#5365); host capability adapters and config seam (#5396)

Media, mascot & config

• Error instead of false success on the generation poll timeout (#5411)
• Accept PNG/GIF/JPEG/WebP custom avatars with file upload (#5420)
• Surface a user notice when a corrupt config.toml is reset (#5340); stop transient active_user.toml read faults from resetting the profile (#5337)

Channels & sessions

• Stop a missing edit route from looking like a deleted message (#5283)
• Keep the /auth/me 401 classifiable as session expiry rather than a hard logout (#5311)
• Stop out-of-credits background scouts paging Sentry (#5310)

Contributors

Thank you to everyone who shipped this release:

Full compare

v0.63.9...v0.63.11

The Release-Reliability Upgrade

This set of releases tightens up the shipping pipeline and keeps the release branch moving smoothly—so builds stay green and delivery stays fast. 🎉

Highlights

Release branch sync & conflict resolution 🔀

We kept release in lockstep with main through a series of promotions, ensuring the latest accumulated work lands cleanly where production builds are cut.

One promotion also required a careful conflict resolution pass to preserve the already-cut release versions while still bringing forward the full set of changes from main. (#5203, #5242, #5256) — Thank you @M3gA-Mind!

Linux AppImage release unblock 🐧

Linux desktop release smoke tests are back on track: an AppImage startup failure on Ubuntu 23.10+ was addressed by adjusting the user namespace sysctl behavior used during the release job, unblocking production runs. (#5251) — Thank you @M3gA-Mind!

Contributor Credits

Thank you to everyone who contributed to this release: @M3gA-Mind, @github-actions[bot].

Full Compare

v0.63.1...v0.63.9

The Composio Reliability Upgrade

This patch release tightens up Composio task execution and makes tool-calls more resilient, so your agents can get to work faster and recover more gracefully when model output gets messy. 🚀

Highlights

Composio task execution & schema validation 🧩

Composio actions triggered via integrations_agent now avoid a frustrating “first call bounce” and can execute immediately when the model’s arguments already match the resolved live contract. This prevents loops like “fetch my latest email” stalling before GMAIL_FETCH_EMAILS ever runs, especially along the text-mode per-action path. (#5132) — Thank you @M3gA-Mind!

Tool-call parsing robustness (Kimi recovery) 🛠️

Tool-call parsing has been hardened to better recover from Kimi-K2-family garbled tool-call bodies. The normalization now strips the <|\"|> argument-quote sentinels and rewrites malformed NAME{unquoted-json} into a canonical form, improving resilience when model output is slightly corrupted. (#5132) — Thank you @M3gA-Mind!

Contributor Credits

Thanks to everyone who contributed to v0.63.1: @M3gA-Mind and @github-actions[bot].

Full Compare

v0.63.0...v0.63.1

The Reliability & Control Upgrade

This release is packed with sturdier workflows, smoother chat and Agent World experiences, and a ton of under-the-hood hardening to keep everything fast and dependable. 🎉

Highlights

Workflows you can trust 🧩

The Workflows copilot and canvas got a major reliability pass: proposals consistently render (even for larger graphs), every copilot turn now ends in a real terminal outcome (no more silent trail‑offs), Accept now saves automatically, and the runs UI stays accurate and live-updating. Safety and correctness also improved with fixes for live-arming on updates, better circuit-breaking, approval-state visibility, and stronger tier/approval enforcement for Composio tool calls.

(#4881, #4886, #4887, #4888, #4889, #4890, #4891, #4893, #4904, #4911, #4932, #4933, #4941, #4942, #4943, #4945, #4955, #4956, #4999, #5000, #5007, #5011, #5012, #5013) — Thank you @graycyrus and @senamakel!

Chat that feels snappier & more controllable 💬

Chat got a big UX win with a Stop button plus ESC-to-interrupt and re-edit, and stronger protection against “hydrate” snapshots wiping fresh streaming or tool timelines. Orchestration also became more transparent: async sub-agent failures and “awaiting user” outcomes now surface correctly in chat, and delegations are now async-by-default with durable, resumable sessions and persistent workflow proposals.

(#4898, #4903, #4908, #5074) — Thank you @M3gA-Mind, @graycyrus, and @senamakel!

Memory & transcript upgrades 🧠

Memory ingestion and recall got sharper and safer: OpenHuman can ingest Codex and Claude coding sessions, Composio sync no longer keys upserts off secret-like titles, and recall avoids pulling same-session “echo” documents that drown out real facts. On top of that, transcripts are now much more robust: cold-boot resumes restore full-fidelity context and thinking, and the transcript system is moving to an append-only, derived view with raw session logs as the source of truth.

(#4863, #4953, #4991, #5077, #5086) — Thank you @senamakel, @M3gA-Mind, and @graycyrus!

Agent World (Tiny Place) feels alive 🌍

Tiny Place is now more interactive and complete: DMs and feeds update in real time, ledgers and feeds paginate beyond the first 50 items, and directory entries finally open a rich profile view. Profiles also leveled up with editing for your own profile (name/bio/avatar) and a public viewer route for arbitrary handles.

(#4951, #4952, #4988, #4989, #4990, #4994, #4996, #4997) — Thank you @M3gA-Mind!

Performance, polish, and “it just works” fixes ⚡

From accessibility and UI polish to startup and runtime performance, this release reduces friction across the app: improved dark theme contrast, paused mesh-gradient rendering when the app isn’t visible, and boot polling is consolidated and quieter. First-run setup overlays now appear only when actually provisioning, channel setup guidance is corrected in-app, pricing/checkout redirects behave properly, and the settings “Clear App Data” reset now correctly targets the signed-in user.

(#4897, #4902, #4905, #4910, #4939, #4954, #5035, #5036, #5066, #5075, #5085) — Thank you @M3gA-Mind, @YellowSnnowmann, and @senamakel!

Build flexibility, CI hardening, and new surfaces 🛠️

The core is now far more modular and easier to ship in slim builds via compile-time feature gates (media, web3, flows, skills, MCP, meet) and a new feature-gated terminal chat UI (openhuman tui). Desktop shipping and CI also got stricter and safer: voice and tokenjuice-treesitter are correctly forwarded into the desktop build, CI now gates feature-forwarding drift, macOS clippy blockers are resolved, the “gates-off” smoke lane is repaired and runs tests, and learning startup subscribers are decoupled from channels and covered with more stable tests. There’s also a core refactor moving the web chat conduit out of channels/.

(#4840, #4855, #4912, #4913, #4914, #4915, #4916, #4917, #4934, #5004, #5020, #5023, #5025, #5070, #5082, #5084) — Thank you @oxoxDev, @CodeGhost21, @senamakel, and @M3gA-Mind!

Integrations, inference, and tooling improvements 🔌

Integrations and local AI got more reliable: OpenAI-compatible local runtimes now probe /v1/models correctly, custom embedding endpoint verification errors are clearer and actionable, and streaming retries now honor Retry-After. Composio tool execution is smarter and safer with per-action tools gated on their full contracts (reducing argument guessing), and unsyncable toolkits no longer auto-register as memory sources. There’s also a new generate_document tool that emits real .docx, plus a public accessor for last-turn usage totals for downstream hosting/billing.

(#4899, #4909, #4940, #4993, #4995, #5064, #5067) — Thank you @M3gA-Mind and @senamakel!

Account, sharing, and navigation delights 🔗

You can now generate one-tap “look what my agent did” share cards for X and LinkedIn, the X/Twitter “Continue with Google” popup stays in-app, Rewards is finally discoverable via a cloud-gated sidebar entry, and the Codex auth button is properly localized. Plus: proactive “Agentic task insights” fixtures and lifecycle resets were tightened to keep the UI consistent.

(#4892, #4944, #4949, #4992, #5010, #5015, #5065) — Thank you @YellowSnnowmann, @graycyrus, and @M3gA-Mind!

Safety & prompt discipline changes 🧯

Desktop automation safety controls saw churn: an Emergency Stop was introduced and later fully removed based on product direction. Meanwhile, agent behavior improved for local LLM providers with provider-aware context/delegation guardrails, stronger structured-output enforcement, more resilient tool-call parsing for small models, and AGENTS.md instructions are now loaded into the system prompt—with follow-up security hardening to prevent symlink/path escape and to redact debug logs when AGENTS.md is present.

(#4600, #4900, #4935, #4938, #5073, #5087, #5096) — Thank you @CodeGhost21, @M3gA-Mind, @sanil-23, @yh928, and @senamakel!

Additional highlights 🔗

A few more focused contributions round out this release with targeted fixes and improvements that are worth calling out. (#4872, #4877, #4906, #5008) — Thank you @senamakel, @graycyrus, and @M3gA-Mind!

New Contributors

• Thank you @yh928 for improving tool-call robustness by recovering pipe-garbled <tool_call> tags from small models. (#4938)

Hope you’ll join us on Discord—there are exclusive roles and contributor rewards waiting.

Contributor Credits

Thank you @CodeGhost21, @cyrus Gray, @mega Mind, @oxoxDev, @sanil-23, @steven Enamakel, @YellowSnnowmann, and @yh928!

Full Compare

v0.61.8...v0.63.0

The Smooth Shipping Upgrade

v0.61.2 → v0.61.8 is all about clean release momentum—promoting the latest work from main into release so staging/prod builds can pick up the newest improvements. Onward and upward. 🚀

Highlights

Release pipeline & promotion 🚢

This release primarily reflects a clean promotion of main into release, bringing the accumulated changes forward with a fast-forwardable merge (no divergence), ready for staging and production cuts.

(#4936) — Thank you @M3gA-Mind!

Contributor Credits

Thanks to everyone who contributed to this release: @M3gA-Mind, @github-actions[bot].

Full Compare

v0.61.2...v0.61.8

The Orchestration & Workflows Leap

This release brings a big step forward in orchestrating agents, building reliable workflows, and tightening up privacy, stability, and observability across the app. Onward! 🚀

Highlights

Orchestration hub & agent fleet control 🧠

Orchestration matured into a true control center: a TinyPlace-focused hub and dedicated tab, an attention/“Needs you” zone for approvals and blocked runs, typed v2 harness-session streams, and a Master chat control channel for managing sessions (including better DM ingest/replies and safer handling of undecryptable messages).

This also expands what can be orchestrated—co-located local CLI agents, Cursor/Windsurf harness support, hosted-brain client work, and new device tool execution flows (with trust gating and approvals) to keep powerful actions both usable and safe. (#4588, #4589, #4592, #4609, #4623, #4627, #4652, #4660, #4693, #4709, #4710, #4712, #4713, #4720, #4722, #4725, #4738, #4753, #4773, #4775, #4777, #4778, #4837) — Thank you @oxoxDev, @sanil-23, @senamakel, @M3gA-Mind, @CodeGhost21, and @YellowSnnowmann!

Workflows (tinyflows) from foundations to polished UX 🔁

Workflows evolved rapidly: the core engine seam and domain landed, then triggers/resume/history, approvals and trust, run inspection, list + navigation, canvas views, and a full Workflows page overhaul with discoveries and runs. The builder gained major quality-of-life upgrades—live catalog grounding, clearer input handling, safer defaults and gating, better diagnostics, and a “Fix with agent” path for failed runs.

A long run of correctness fixes followed: condition and routing validation, schedule rendering, tool output path handling (including schema-less probing), better run honesty (including warnings), deduped output rendering, persisted copilot chat, and protection against accidentally arming scheduled/side-effecting flows. (#4433, #4447, #4449, #4450, #4471, #4472, #4474, #4477, #4552, #4561, #4570, #4573, #4574, #4578, #4580, #4586, #4587, #4590, #4591, #4602, #4605, #4606, #4607, #4608, #4628, #4630, #4636, #4645, #4647, #4651, #4659, #4662, #4663, #4664, #4665, #4668, #4670, #4672, #4673, #4674, #4698, #4702, #4707, #4750, #4822, #4823, #4826, #4834, #4835, #4836, #4858, #4865, #4869, #4870, #4874, #4876) — Thank you @graycyrus, @senamakel, @CodeGhost21, and @YellowSnnowmann!

TinyAgents + agent harness reliability (and better answers) 🧩

A major wave of harness and TinyAgents migration work landed: the agent_graph state-machine harness with checkpointing/HITL/observability, migration completion, structured policy-block messaging, better “final answer” guarantees, tool-failure classification surfaced to the UI, and a strengthened memory read→dedupe→write→index contract (plus a long tail of fixes around unavailable tools and no-progress nudges).

A lot of robustness work also went into preventing wedged turns and improving orchestration behaviors—timeouts, parallel fan-out, safer delegation paths, clearer tool failure rendering, prompt posture adjustments, and better scheduler/cron hardening. (#4261, #4362, #4372, #4386, #4399, #4400, #4413, #4419, #4425, #4430, #4431, #4434, #4443, #4444, #4445, #4473, #4476, #4478, #4479, #4480, #4481, #4482, #4483, #4485, #4495, #4502, #4504, #4507, #4509, #4512, #4524, #4539, #4542, #4601, #4603, #4658, #4682, #4683, #4686, #4741, #4748, #4751, #4752, #4757, #4764, #4768) — Thank you @senamakel, @M3gA-Mind, @oxoxDev, @sanil-23, @sanil-23, and @CodeGhost21!

Memory engine upgrade: TinyCortex migration 🧠📚

Memory got a foundational upgrade with a staged TinyCortex migration: adapter seams, type unification, connection foundations, chunk store flips (read paths, batching, listing), and finally the full engine migration—backed by parity harness work and updated coverage.

Along the way, several important quality fixes landed: safer rendering shims, better memory-store self-healing, clearer error classification, and improved reliability signals between Vault health and Sync status. (#4363, #4526, #4529, #4532, #4534, #4538, #4546, #4547, #4551, #4555, #4559, #4781, #4785, #4787, #4788, #4789, #4790, #4794, #4820, #4699, #4700, #4703, #4340, #4410) — Thank you @senamakel, @YellowSnnowmann, @CodeGhost21, @M3gA-Mind, and @oxoxDev!

Privacy Mode & local-only enforcement 🔒

Privacy became first-class: Privacy Mode and local-only inference enforcement, plus an egress descriptor + disclosure events, privacy status surfaces, per-action disclosure UI, and PII/identification-risk detection.

This work also extends local-only enforcement across integrations and tools, making it easier to understand what will (and won’t) leave your machine before you run something. (#4446, #4812, #4813, #4849, #4850) — Thank you @oxoxDev!

Observability & Langfuse tracing (deeper + safer) 🔭

Tracing and observability expanded dramatically: richer Langfuse metadata, one-trace-per-turn, full-content tracing across generations/tools/subagents with real costs, and export of graph/flow runs (including Agent-Graph-view metadata). Journal telemetry now feeds Langfuse usage/run insights, and trace parenting for spawned sub-agents is correctly linked.

Stability improvements round it out: chunked ingestion batches, demoted noisy rejection paths, and several trace correctness fixes (including an upstream observation-id collision fix via dependency bump). (#4498, #4499, #4506, #4562, #4568, #4676, #4678, #4864) — Thank you @senamakel, @M3gA-Mind, and @oxoxDev!

Desktop stability, startup recovery, and UX polish 🖥️

Desktop reliability improved across platforms: proactive stale process cleanup on Windows startup, macOS/Linux stale-lock reaping and installer pre-install kills, better Linux linking for error handling/builds, and CEF/Windows GPU routing via SwiftShader (plus tighter scoping for Windows pre-CEF recovery).

On the UX front, this range adds remote-core connection settings with live status, window geometry persistence and minimized-window restore behaviors, better sign-in error messaging, and a set of app-wide header/landing standardizations. (#3793, #4405, #4406, #4643, #4644, #4649, #4817, #4818, #4851, #4857, #4859, #4860, #4805) — Thank you @sanil-23, @M3gA-Mind, @senamakel, and @NgoQuocViet2001!

Channels & communications: email, Telegram/WhatsApp, meetings ✉️

OpenHuman gained native IMAP/SMTP email integration, plus continued work on channels via tinychannels migrations and runtime boundaries. Telegram pairing and tool-calling reliability improved, history ingest got back-pressure, and WhatsApp corruption handling now recovers instead of flooding Sentry.

Meetings leveled up with live transcripts, improved correlation IDs, Recall Calendar auto-join and UI improvements, richer bot “speak” tagging for mascot pose, and even dual mascots with per-mascot voice + speaker alternation. (#4367, #4414, #4569, #4549, #4557, #4349, #4350, #4354, #4391, #4585, #4715, #4740) — Thank you @oxoxDev, @senamakel, @graycyrus, and @YellowSnnowmann!

Inference + core runtime modernization ⚙️

Inference moved toward a cleaner, crate-native and TinyAgents-driven architecture: migrating callers onto tinyagents ChatModel, decoupling per-turn state, adopting ModelRouter routing, and multiple “Motion” steps to cut over BYOK cloud slugs and OpenAI/codex paths to crate-native implementations—while also addressing SSE hang scenarios and ensuring raw BYOK model IDs are preserved.

On the platform side, the core runtime gained pluggability and a “fleet host,” plus DomainSet composition and a compile-time voice feature gate. There’s also progress on managed file storage agent tools and surfaced storage transfer tools. (#4393, #4614, #4625, #4629, #4769, #4780, #4782, #4783, #4784, #4298, #4791, #4808, #4833, #4576, #4577) — Thank you @senamakel, @oxoxDev, @sanil-23, and @YellowSnnowmann!

Quality: CI, tests, and release plumbing 🧪

This range invests heavily in build and test confidence: two-lane CI (lite/full), changed-files-only PR lanes and full-suite release lanes, test audits and phased cleanup/backfill, faster Rust test builds, and many E2E stabilizations across platforms and release “tails.”

Release mechanics also got attention: workflow parse fixes, full dist artifact restoration, release-lane unblocks and timeouts, branch/limit adjustments (including a cap removal), and CodeQL alert sink hardening. (#4486, #4488, #4489, #4492, #4494, #4497, #4500, #4501, #4511, #4519, #4521, #4527, #4530, #4531, #4533, #4540, #4543, #4548, #4553, #4554, #4556, #4579, #4646, #4666, #4671, #4701, #4721, #4759, #4827, #4830, #4838, #4866, #4867, #4871) — Thank you @senamakel, @M3gA-Mind, @CodeGhost21, and @graycyrus!

Product experience: chat, persona, rewards, settings, and docs 🎁

Chat and conversations got significant upgrades: per-turn process history and a multi-phase timeline refactor, improved event merging in reducers, more reliable tool-result rendering, timeline anchoring/dedup fixes, and a 1M context window adjustment for chat-v1. There are also UX refinements like tool-call failure explanations in the processing timeline and safer cancellation scoping.

Elsewhere: morning briefing personalization, guided persona builder role templates, achievement token rewards visibility, font-size fine-tuning, standardized analytics interaction tracking, plus a major docs refresh including setup/recovery guides and reorganized install/workflows/orchestration documentation. (#4373, #4416, #4424, #4612, #4604, #4622, #4626, #4765, #4711, #4696, #4873, #4407, #4508, #4514, #4515, #4792, #4493, #4510, #4513, #4516) — Thank you @M3gA-Mind, @senamakel, @oxoxDev, and @YellowSnnowmann!

Smaller-but-mighty fixes across the app 🛠️

A broad set of targeted fixes improved everyday reliability: stopping API usage flooding, rebinding state correctly on active-user switch, handling /announcements/latest 404s gracefully, better offline-provider behavior to prevent Sentry floods, and safer boot behavior in cloud mode.

There are also improvements in learning and agentworld utilities, connection cleanup correctness, MCP catalog additions, shell execution routing on Windows, and composio approval parking behavior. (#4379, #4417, #4415, #4402, #4484, #4518, #4697, #4717, #4723, #4758, #4584, #4704, #4819, #4848, #4298) — Thank you @oxoxDev, @mysma-9403, @CodeGhost21, @M3gA-Mind, @sanil-23, and @YellowSnnowmann!

Language-based workflows (RLM/Rhai) & “Codex” vendor work 🧠🧰

Language-based workflows took shape with a first-class Rhai .ragsh REPL tool and follow-on domain renaming, backed by planning docs and broader vendor strategy work.

A set of “codex” and vendoring PRs also landed: TinyJuice/TokenJuice work, TinyChannels runtime pieces and provider adoption, tiny.place SDK vendoring and pairing flow, and various submodule updates to keep the stack aligned. (#4528, #4571, #4523, #4400, #4535, #4549, #4557, #4560, #4575, #4620, #4491, #4563, #4815) — Thank you @senamakel!

Additional highlights 🔗

A few more focused contributions round out this release with targeted fixes and improvements that are worth calling out. (#4475, #4525, #4550, #4558, #4564, #4582, #4599, #4611, #4621, #4624, #4642, #4694, #4807, #4816, #4824, #4831, #4832, #4839, #4842, #4875) — Thank you @sanil-23, @senamakel, @CodeGhost21, @YellowSnnowmann, and @M3gA-Mind!

New Contributors

• Thank you @nazarli-shabnam for improving resilience by making /announcements/latest return null on 404 instead of erroring. (#4417)

Hope you’ll join our Discord—there are exclusive roles and contributor rewards waiting for you!

Contributor Credits

Huge thanks to everyone who contributed to this release: @CodeGhost21, @cyrus Gray, @mega Mind, @mysma-9403, @NgoQuocViet2001, @oxoxDev, @sanil-23, @Shabnam, @steven Enamakel, @YellowSnnowmann, and the automation support from @github-actions[bot].

Full Compare

v0.58.7...v0.61.2

The Meetings & Connections Power‑Up

v0.58.0–v0.58.7 is packed with a sharper Meetings experience, faster + more reliable Connections (Skills/MCP), and a big wave of stability upgrades across agents, memory, and providers. 🚀

Highlights

Meetings, end-to-end polish 🎥

Meetings got a full redesign with a platform selector, a clean history master-detail layout, and smarter defaults—plus a bunch of quality fixes that make sessions feel more “alive” and less error-prone. You’ll see better UI feedback (including “Leaving…” states), automatic refresh of Recent Calls when a meeting ends, and improved handling of backend capacity messaging.

Wake phrases are now correctly forwarded for active meeting bots, meeting display names can auto-fill from your connected account, and the auto-summarize policy is honored—with a new manual summary RPC for “Ask” flows.
(#4308, #4348, #4347, #4221, #4231, #4297, #4353) — Thank you @graycyrus and @YOMXXX!

Connections: MCP & Skills that feel fast (and stay fast) 🔌

MCP took a major leap forward: the MCP Servers page is now “complete” with a curated catalog, lifecycle improvements, and E2E coverage. Under the hood, MCP tool calls return as soon as the reply frame arrives (huge latency win), stdio servers spawn with the user’s real shell PATH (hello npx/uvx), and schema migrations are now concurrency-safe.

Auth and install UX also improved across the board: clearer MCP 401 explanations (with safer messaging), better “installed / installing” card state, and Skills improvements like activation persistence, registry caching + pagination, and latency instrumentation.
(#4300, #4220, #4295, #4219, #4321, #4224, #4288, #4120, #4182, #4188, #4238) — Thank you @YellowSnnowmann, @M3gA-Mind, @oxoxDev, and @YOMXXX!

Chat experience: attachments, visuals, and smoother long turns 💬

Chat now has a first-class attachment button for images, videos, and documents—with drag-and-drop and paste support. Code blocks gained syntax highlighting across markdown surfaces, and the composer/footer layout now dynamically reserves space so queued follow-ups (and other expanding footer content) no longer overlap the conversation.

On the reliability side, long reasoning turns stay alive past the old 2-minute silence timeout thanks to a heartbeat progress signal.
(#4292, #4335, #4334, #4282) — Thank you @oxoxDev and @graycyrus!

Agents & tool runtime: fewer loops, clearer stops 🧠

Agent orchestration is now much harder to derail: identical tool-call loops are detected and halted, stuck sub-agents can return as explicitly incomplete, and session tool calls now respect timeouts instead of hanging forever. The MCP Setup Agent no longer gets trapped in a re-spawn loop when pausing for user clarification, and goal completion instructions were strengthened to improve auto-complete behavior.

The built-in researcher agent was also tightened up with a bounded tool surface and capped per-turn output.
(#4230, #4262, #4303, #4310, #4313, #4314) — Thank you @sanil-23, @YOMXXX, @oxoxDev, @salarkhannn, and @senamakel!

Memory & intelligence: better continuity and safer cross-host behavior 🗂️

Memory got meaningful durability upgrades: summary-only wiki history is now tracked via git, cross-thread search is faster by ranking before cloning heavy hits, and the Trees UI has clearer feedback and a stable toolbar layout even when vault guidance opens.

For multi-device setups, cross-host vault paths are now explicitly surfaced as belonging to a particular host OS—so a frontend won’t try to open macOS paths on Linux (or vice versa). The Brain memory graph reliably reloads after re-login, and the Intelligence page scrolls correctly inside the Brain card.
(#4168, #4311, #4331, #4342, #4330, #4223, #4332) — Thank you @senamakel, @mysma-9403, @graycyrus, @oxoxDev, and @M3gA-Mind!

Provider, inference, and observability hardening 🧰

This release includes a broad set of fixes that turn noisy “errors” into correctly-classified expected states, reduce retry storms, and improve resilience across providers and local environments. Highlights include: safer model-catalog parsing, better embeddings verification (including auto-detected dimensions and Gemini model-id normalization), improved handling of quota/limit states, and more robust OpenAI-compat parsing.

Backend routing is also safer: backend API calls won’t accidentally route to a BYO cloud inference host. For channels, OAuth “Connecting” states now recover cleanly when deep-link roundtrips fail. And for cron jobs, missing API keys halt early (with better user-visible error surfacing).
(#4157, #4160, #4184, #4186, #4185, #4207, #4209, #4210, #4161, #4189, #4187, #4208, #4286, #4306, #4166, #4169) — Thank you @oxoxDev and @M3gA-Mind!

Tiny Place: identity + trading fixes, smoother Linux rendering 🪙

Tiny Place saw a strong round of fixes: correct pricing tiers, USDC SPL mint acceptance for handle registration, longer RPC timeouts for bounty creation, and a more polished UX for destructive actions via an in-app confirm modal. Trading is more usable too—balances are read from the correct settlement RPC, Buy re-enables properly after a purchase attempt, and users can switch their primary handle across multiple identities.

On Linux, the world renderer now gets a working WebGL context via SwiftShader (without reverting to the old “disable GPU” blunt instrument).
(#4215, #4211, #4212, #4214, #4218, #4217, #4216, #4213) — Thank you @M3gA-Mind!

Voice, Human, devices, and desktop reliability 🎙️

Voice on macOS is more robust: in-process Whisper STT can work without relying on an external binary, and the install/test UI behaves more predictably. Human UI also gets a practical fix: the mic device selector stays within the viewport instead of clipping off-screen.

Desktop reliability improvements include a CEF GPU-disable startup override for tough environments, Debian installs preferring .deb packages (with proper runtime deps), and device pairing now using the correct Socket.IO ACK flow.
(#3915, #4343, #4309, #4263, #4355) — Thank you @YellowSnnowmann, @graycyrus, @sam, and @YOMXXX!

New creative tools, mascots, and model routing ✨

There’s a whole new media-generation domain with image/video agents and GMI media tools that download outputs into stable local paths. Mascots can now load from a GitHub-backed manifest, making it easier to expand and evolve the Human experience over time. On the routing side, the new burst-v1 tier is registered and used for appropriate low-context workers, including the SuperContext scout and other tool-driving agents.

There’s also improved auditability for channel replies by recording provider/model metadata on processed channel events.
(#4244, #4312, #4260, #4315, #3682) — Thank you @senamakel, @senamakel-droid, and @ron-liu!

Smaller (but mighty) fixes across the stack 🛠️

A collection of sharp fixes landed across token accounting, learning, security, people, artifacts, and CI ergonomics. TokenJuice avoids UTF-8 char-boundary panics and correctly attributes compaction savings to the per-turn model; usage gauges no longer exceed 100% due to sub-agent tokens; learning stability uses the correct facet half-life; chat files hydrate from the artifact ledger on cold start; and the people store is correctly seeded at core boot.

Security hardening also tightened several untrusted-input sinks, and CI coverage runs are stabilized by raising RUST_MIN_STACK to avoid llvm-cov stack overflows.
(#4170, #4229, #4327, #4328, #4232, #4345, #4158, #4346) — Thank you @mysma-9403, @rainbowpuffpuff, @oxoxDev, @sam, @obchain, and @graycyrus!

Documentation updates 📚

Docs received a substantial GitBook/README overhaul and a set of correctness follow-ups to tighten terminology, scope, and RPC naming.
(#4225, #4243) — Thank you @senamakel!

Additional highlights 🔗

A few more focused contributions round out this release with targeted fixes and improvements that are worth calling out. (#4140, #4163, #4183, #4287, #4293, #4296, #4318) — Thank you @senamakel-droid, @oxoxDev, @M3gA-Mind, @YellowSnnowmann, and @YOMXXX!

New Contributors

• Thank you @rainbowpuffpuff for fixing TokenJuice savings attribution so compaction savings are credited to the per-turn model. (#4229)
• Thank you @salarkhannn for strengthening the goal completion directive to improve agent auto-complete behavior. (#4310)

Hope you’ll join us on Discord—there are exclusive roles and contributor rewards waiting for you!

Contributor Credits

Thanks to everyone who contributed to this release: Cyrus Gray, Mega Mind, oxoxDev, YellowSnnowmann, Steven Enamakel, YOMXXX, mysma-9403, Sam, Ron Liu, sanil-23, obchain, rainbowpuffpuff, Salar, and Steven Enamakel's Droid.

Full Compare

v0.58.0...v0.58.7

Super Context is Here! 🎉

v0.57.53 to v0.58.0 brings 124 PRs across 133 commits, with upgrades across memory, voice, agents, reliability, and developer foundations. Thank you to everyone who contributed to this release. ✨

Highlights 🚀

🧠 Memory & intelligence

Memory and intelligence flows are faster, clearer, and more resilient, from source sync controls and richer ingest to batched embeddings and delegated memory retrieval. (#3947, #3948, #3949, #3951, #3959, #3960, #3999, #4008, #4049, #4051, #4054, #4063, #4066, #4067, #4070, #4083, #4086, #4087, #4100, #4106, #4107, #4133, #4138) — Thank you @senamakel, @M3gA-Mind, @oxoxDev, @graycyrus, @mysma-9403, @Subharup-31, @sanil-23, and @senamakel-droid!

✅ Tasks, chat & agent workflows

Task, chat, and agent workflows become easier to steer and inspect, with better task boards, persistent run state, safer delegation, and clearer model/thread controls. (#3918, #3925, #3933, #3965, #3966, #3970, #3977, #3978, #3983, #3985, #3987, #3990, #3994, #4003, #4007, #4011, #4012, #4016, #4017, #4018, #4019, #4021, #4027, #4030, #4031, #4050, #4060, #4061, #4062, #4080, #4081, #4082, #4084, #4085, #4102, #4103, #4105, #4114, #4128, #4129, #4130, #4132, #4134, #4136, #4139, #4141, #4156, #4173) — Thank you @Felyx-Fu, @graycyrus, @sanil-23, @M3gA-Mind, @nazarli-shabnam, @ly-wang19, @YellowSnnowmann, @oxoxDev, @senamakel, @hgunduzoglu, and @senamakel-droid!

🔐 Reliability, security & platform polish

Reliability and platform behavior are tightened across security policy, credential handling, session health, CEF startup, Windows filesystem edge cases, and release-only build paths. (#3958, #3969, #3974, #3976, #3992, #4004, #4009, #4028, #4058, #4064, #4065, #4068, #4069, #4076, #4078, #4101) — Thank you @M3gA-Mind, @oxoxDev, and @senamakel!

🧩 Integrations, UI & developer foundations

Integrations, UI polish, and developer foundations also move forward, including clearer connected-account surfaces, meeting-agent wiring, presentation/artifact improvements, and cleaner docs/code organization. (#3718, #3941, #3946, #3973, #3975, #3981, #3982, #3984, #3986, #3988, #3989, #3995, #3998, #4002, #4026, #4032, #4037, #4052, #4053, #4057, #4071, #4072, #4079, #4108, #4109, #4110, #4113, #4115, #4123, #4124, #4125, #4126, #4127, #4131, #4137, #4162, #4171) — Thank you @obchain, @martian56, @M3gA-Mind, @nazarli-shabnam, @graycyrus, @senamakel, @oxoxDev, @sanil-23, @YellowSnnowmann, and @senamakel-droid!

New Contributors 🌟

• Welcome Martian! Thank you for #3941 fix(composio): pin toolkit_versions=latest in direct-mode tool listings (#3932); #3946 fix(tauri): identify foreign process on core port + consent-gated force-quit (#3331).
• Welcome Shabnam! Thank you for #3970 fix(sentry): re-register inboundFiltersIntegration so ignoreErrors runs; #3984 fix(settings): render Managed provider as always-on badge, not a locked toggle.

Hope you join the Discord for exclusive roles and contributor rewards!

Contributor Credits 🙌

• Cyrus Gray (#3925, #3990, #3994, #3995, #3998, #4002, #4007, #4008, #4018, #4019, #4037, #4052, #4053, #4054, #4057, #4103, #4109, #4110, #4156, #4162) — thank you!
• Felix (#3918, #4012) — thank you!
• github-actions[bot] — thank you!
• Hüsam (#4050) — thank you!
• ly-wang19 (#3978) — thank you!
• Martian (#3941, #3946) — thank you!
• Mega Mind (#3951, #3958, #3959, #3960, #3965, #3966, #3973, #3975, #3977, #3981, #3982, #3983, #3985, #3986, #4064, #4065, #4066, #4067, #4068, #4069, #4070, #4071, #4078, #4171) — thank you!
• mysma-9403 (#4049, #4063) — thank you!
• obchain (#3718, #3988, #3989) — thank you!
• oxoxDev (#3969, #3974, #3976, #3992, #3999, #4003, #4004, #4009, #4011, #4030, #4031, #4032, #4058, #4060, #4061, #4072, #4076, #4101, #4102, #4108, #4114, #4115, #4127) — thank you!
• sanil-23 (#3933, #4016, #4079, #4080, #4082, #4083, #4128) — thank you!
• Shabnam (#3970, #3984) — thank you!
• Steven Enamakel (#3947, #3948, #3949, #4017, #4021, #4026, #4027, #4028, #4081, #4084, #4085, #4086, #4087, #4100, #4105, #4106, #4107, #4123, #4124, #4125, #4126, #4129, #4130, #4131, #4132, #4133, #4134, #4136, #4141, #4173) — thank you!
• Steven Enamakel's Droid (#4137, #4138, #4139) — thank you!
• Subharup Nandi (#4051) — thank you!
• YellowSnnowmann (#3987, #4062, #4113) — thank you!

Full Compare 🧭

v0.57.53...v0.58.0