Add a complete Scheduled Tasks guide with real form and chat examples, six framed screenshots, schedule fields, task conversations, controls, and catch-up behavior. Link it from the docs index, Usage, and A0Cleaner, and correct the Cleaner timezone instructions.
Verified both creation paths on localhost:32081, ran the form-created task, checked edit/reset/delete controls, and removed the demo tasks. Validated local links, matching website text, image dimensions, and rendered screenshots. Document the current lack of a Stop button for scheduler runs.
Add a practical guide to opening A0Cleaner, automatic cleanup defaults,
notification timing, missed schedules, manual review, custom folders, and
cleanup history. Explain the difference between one-time selection and Auto
switches, and clarify the limits around Linux /tmp and uploaded files.
Include five real UI captures with complete panel framing and neutral
surroundings. Reproduce the native notification from the saved maintenance
result and leave the custom-folder example unsaved. Optimize the PNG set
from 311 KB to 119 KB. Link the guide from the docs index and Usage, and
record ownership alongside the aligned website guide.
Validation:
- Verified the entry point, defaults, and review/cancel flow at localhost:32081.
- Checked 33 local links, anchors, components, and assets across both guides.
- Confirmed matching guide text and no em dashes.
- Inspected optimized screenshots and the rendered Core documentation preview.
- git diff --cached --check passed.
The shared Chromium, Xvfb and Xpra runtime could remain alive indefinitely
after chats stopped using it. Suspend it after one hour of inactivity by
default, preserving saved tab URLs and the shared sign-in profile. Protect
running and paused owners, subordinate parent tasks, connected viewers,
browser calls and downloads; passive polling must not restart the runtime
or extend its lifetime.
Add Stop internal browser to Browser Settings, release the runtime on
global plugin disable, and clean stale display directories only when no
live owner or process reference remains. Document the global timeout,
opt-out and loss of unsaved page state on suspension.
Validation: 206 focused tests and 20 real-Chromium security tests passed.
Live Docker checks verified periodic and manual shutdown from 13 child
processes to zero, download protection, tab/cookie recovery, the WebUI
control and CSRF rejection. The reported GPU busy-loop was not reproduced;
this change bounds idle lifetime without changing graphics flags.
Refs #1892, #2029.
Add a practical Telegram guide covering BotFather setup, private polling,
access and administrator lists, commands, attachments, project/profile
ownership, groups, webhooks, and preservation of existing configurations.
Explain distinct bot accounts per saved scope, inheritance, duplicate-token
recovery, and the limits of connection checks and default project routing.
Link the guide from the main README, documentation index, and plugin
reference. Add four optimized screenshots using neutral unsaved drafts,
with complete dialog frames and comfortable spacing. Record that screenshot
framing requirement in the documentation contracts.
Validation:
- Checked guide links, anchors, and assets alongside the website counterpart.
- Verified UI labels and screenshots against the live Docker instance.
- Confirmed Telegram configuration and activation files stayed unchanged.
- git diff --check passed.
Replace the compact Runtime Ready capture with a full-width installation
walkthrough screenshot. Preserve the complete scene and resolution, remove
the superseded asset, and update the image description and ownership docs.
Optimize the PNG from 324,982 to 122,479 bytes. The Launcher repository uses
an identical copy.
Validation: checked image paths and placement, PNG decoding, matching asset
hashes, local desktop/mobile rendering, and git diff --check.
Add the Runtime Ready onboarding screenshot directly below Desktop setup.
Present recommended downloads by platform, with Apple Silicon first for
macOS and the alternative clearly labeled Intel Macs only. Keep processor
labels concise and preserve all existing download URLs.
Update the documentation contract for the screenshot and Mac download order.
Validation: checked links, screenshot loading, Apple Silicon placement,
local desktop/mobile rendering, and git diff --check.
Present Launcher downloads by platform and processor, and recommend A0
Install in a visible Servers and shell access section. Share first-message
setup across both paths and keep direct Docker and troubleshooting under
Other install options.
Link scripted setup to the installation guide, preserve existing anchors
and download URLs, and update the documentation contract.
Validation: checked 19 Quick Start links and anchors, local desktop/mobile
rendering, accordion behavior, unchanged Docker command, and git diff --check.
Replace the Plugin Hub screenshot with the updated community plugin browser capture. Optimize the PNG while preserving its dimensions and existing README reference.
Use a plain-language introduction and a complete Launcher quickstart that
carries the first message through model setup and back to the same chat.
Simplify feature descriptions while keeping the existing demonstrations,
and add clear migration, provider-plugin, Host access and license links.
Align Core installation and quickstart with the website journey. Update
Launcher downloads to v1.9, pair local Docker commands with their browser
URLs, distinguish persistent storage from backups, and remove unsupported
model-size recommendations. Reuse the verified kickoff-plan screenshots.
Validation: local links, assets and anchors checked; README rendered
locally; git diff --check passed. No fresh OS installation was performed.
Add a screenshot guide covering ChatGPT's device-code permission, Account
onboarding, both Connect account actions, OpenAI authorization, and Main
and Utility model selection. Explain how Finish setup resumes the held
message and Continue closes the wizard without creating another chat.
Include twelve optimized captures with account details and device codes
removed, reuse the current ready screen, and link the guide from the README,
docs index and Onboarding. Replace the README's outdated account-flow copy.
Validation: Opus review, source and official authentication guidance checks,
76 link/image/dimension checks, matching website prose, opaque redactions,
and whitespace checks. The supplied completed sign-in captures were used;
no new account authorization or inference run was needed.
Suggest gpt-6.1-terra for Main and gpt-6-luna for Utility through Codex
provider metadata, while preserving explicit model lists and the live
account catalog.
Change the final onboarding action to Continue and close the wizard without
creating another chat. The existing model-gate event resumes the held message
in its original chat. Update the onboarding guide and ready-screen capture.
Add regressions for recommended and explicit model selection, configured
catalog order, held-message dispatch and completion without a pending chat.
Validation: 244 focused Python tests, three onboarding Node tests and the
OAuth reconnect check passed. Verified on the dedicated instance that
Continue closes the wizard and preserves the selected chat and chat list.
Walk new users from their first message through the Cloud provider grid,
filtered Plugin Hub, OpenCode Go installation, browser refresh and model
selection. Link the guide from the README, docs index, Onboarding and Usage,
and add seven optimized screenshots of the real setup flow.
Use the verified Go model IDs glm-5.3 and glm-5.3-flash, explain the free-model
restriction, and distinguish checked setup from untested model responses.
Validation: real installation and model selection on the dedicated instance;
local links, anchors, image dimensions, privacy and whitespace checked.
No subscription credentials were entered or paid inference performed.
Walk through Pi, Hermes Agent and Claude Code exports, imports, separate
working-file uploads and continued tasks using real CLI sessions. Add
built-in chat and message search instructions and optimized neutral captures.
Preserve Claude custom titles and keep tool-result blocks in historical
tools instead of duplicating them as user messages. Retain unmatched results
and error markers, including coverage through the shared OpenClaw parser.
Validation: 28 migration tests and 11 subtests; six live imports into three
projects; three completed continuations; two booking tests; documentation
links, image privacy checks and rendered previews.
Import Pi JSONL session versions 1–3 through the existing preview and category-selection workflow. Preserve separate retained branches, tool arguments and results, errors, public shell history, titles, and project context while excluding system prompts, reasoning, extension state, and image bytes. Validate session trees and bound branch expansion before processing.
Persist migration provenance in durable chat data, redact imported titles, and accept Unicode line separators inside JSONL strings. Add the Pi source selector and a user guide with two optimized screenshots from a real preview and import.
Validation: 25 migration tests and 11 subtests passed; the live Web UI imported two real Pi chats into one project with saved provenance and historical tool records; source-session hashes were unchanged and private reasoning was excluded. Checked desktop/mobile layout, 487 local documentation references, JavaScript syntax, image metadata/privacy, and git diff --check.
Rework Quick Start around sending a first message and opening its saved result. Explain held-message retry during onboarding, Welcome layout recovery, running-chat controls, and goals using current product behavior.
Update Agent Profiles for individual tool actions and Model Presets for current controls. Replace stale views with twelve real, neutral screenshots shared with the website, optimized from 605118 to 251382 bytes. Refresh the docs index and ownership guidance, and repair three VPS table-of-contents anchors.
Validation: 478 local documentation references passed; image integrity, metadata, and privacy checks passed; representative captures and rendered Quick Start were inspected; git diff --check passed. Changes are documentation and assets only.
Replace the onboarding terminal preview and CLI guide screenshots with optimized full-colour captures of the current vanilla CLI. Use neutral workspace paths, remove image metadata, and keep shared captures identical across the documentation repositories.
Update host-picker wording, automatic connection guidance, chat tabs and model controls, and Browser mode scope. Version the onboarding image URL to refresh browser caches and document capture ownership.
Validation: reviewed onboarding at desktop and mobile sizes, verified documentation links and image integrity, and passed git diff --check.
Explain how host folders map into Docker containers, how bind mounts differ
from named volumes, and what persists when an instance is removed. Add CLI,
Docker Desktop, and Compose examples for user data and separate work folders.
Describe Launcher storage choices, generated versus exact folder paths,
workspace defaults, persistence migration, and keep/delete controls. Link the
new guide from installation, troubleshooting, and the documentation index.
Validation: documentation links and anchors, shell syntax, Compose YAML, and
nine existing Launcher storage tests passed. Live container recreation and
Compose execution were not tested; the Compose plugin is unavailable locally.
Avoid starting Chromium for empty tab listings while preserving saved-tab
restoration. Honor navigation readiness without redundant load waits or
fixed pauses, and let WebUI navigation return at document commit.
Skip automatic history screenshots for inspection, evaluation, and batches
containing only those actions. Read page metadata in one evaluation and
report the actual loading state. Reuse the native viewer display size when
no resize is requested, preventing page zoom from enlarging and clipping
frames during navigation.
Update browser documentation and regression coverage. Validated 194 browser
regression/output tests and 29 host-browser tests in separate runs, plus
live WebUI navigation, pointer input, screenshots, and repeated reloads.
Local test-page medians improved from 474 to 83 ms for navigation and from
82 to 19 ms for content reads.
Consolidate plugin creation, review, and contribution behind a0-create-plugin with focused references and brief troubleshooting. Honor explicit local or community intent and ask once when unclear.
Keep a0-manage-plugin as the discovery and lifecycle entrypoint, with Plugin Index filtering, candidate assessment, and framework API operations. Retire redundant skills and update documentation and discovery tests.
Validation: 38 targeted tests passed in a clean snapshot. Verified both entrypoints and six references in the live runtime and exercised Index discovery at localhost:32081 without installing plugins or running security scans.
Run Chromium headful through Patchright on a private Xvfb display while keeping Agent Zero's page helpers in Patchright's isolated world.
Reconcile the pinned Patchright package and matching architecture-specific Chromium from Browser plugin hooks so fresh Docker images and self-updated installations converge on the same runtime.
Replace the legacy guided-profile instructions with screenshot-led Easy and Advanced editor workflows, project availability, and A0 CLI creation.
Document sparse YAML, Markdown, and JSON overrides for identity, prompts, Tool and MCP policy, Skills visibility, and model presets. Refresh adjacent guides and remove obsolete screenshots.
Place the internal Browser proxy fields in a native disclosure and keep its styling borderless. Refresh the guide screenshot and cover the config markup.
Add project-scoped HTTP and SOCKS proxy settings for the internal Docker browser, including bypass and optional authentication. Pass them through Playwright's native proxy option, restart active Browser runtimes when they change, and update regression coverage and user documentation.
Make model presets a single global collection and treat presets as reusable setups (main, utility, embedding). Persisted scope configs now store only a preset name (Default is always present and immutable); project-scoped preset definitions were removed and attempts to edit/reset project presets are rejected. Add embedding-model support throughout: include embedding slot in presets, surface embedding in the switcher and override APIs, and notify extensions when the effective embedding model changes. Implement rename/retire propagation to update plugin configs, saved chats, and live AgentContext objects when preset names change or presets are removed. Update CLI/integration commands to use an "inherit" action for returning to the scoped preset and to report the effective scoped-or-chat preset. Add startup migration/bootstrap to initialize presets from remote or bundled fallback and numerous docs/UI text updates to reflect the new preset model. Extra: refactor config resolution helpers and add safety checks (validation, atomic writes) when saving presets.
Stop PrintStyle from creating per-process HTML files under /a0/logs and remove the obsolete tracked folder scaffolding.
Delete existing legacy log directories during startup migration so self-updated and manually upgraded installations are cleaned automatically. Point plugin debugging guidance to Docker output and add regression coverage for repeatable cleanup.
Preserve explicit endpoint case, accept host:port and HTTP discovery addresses in Browser settings, and keep live-runtime guidance independent of any fixed localhost port.
Reduce Browser Settings host-browser choices to automatic and advertised debug endpoints, with a validated Custom endpoint input and faster inventory refresh.
Update Browser docs and regressions to document the custom endpoint workflow and CLI websocket selection.
Add Browser Settings guidance for supported Chromium-family remote debugging pages, including Opera and explicit remote-debugging endpoint fallback.
Update BYOB docs and troubleshooting to clarify that browser inventory comes from the connected A0 CLI and requires reconnecting the CLI after authorizing a browser.
Negotiate binary and slim screencast frames so updated clients avoid base64-in-JSON and repeated viewer metadata on every frame.
Clamp screencast capture to the subscriber viewport and DPR, lower streaming JPEG quality while preserving snapshot quality, and teach the client to manage binary frame URLs safely.
Document the browser viewer performance plan and update regression coverage for the new transport contract.
Update first-run onboarding docs for the welcome screen, in-chat model gate, Cloud/AI account/Local provider paths, and explicit main and utility model choices.
Replace stale onboarding screenshots with optimized current assets and add gate, account, and local-provider captures.
Fall back to Chat Completions for provider/proxy Responses endpoint failures before any output is emitted.
Preserve streamed Chat Completions tool-call deltas as structured LLMResult function-call items, so fallback providers can still drive tools.
Document Docker host-gateway addressing for local model servers, where container localhost does not reach host loopback.
Verified with: PYTHONPATH="/home/eclypso/a0/agent-zero" conda run -n a0 pytest tests/test_stream_tool_early_stop.py tests/test_responses_architecture.py -q; PYTHONPATH="/home/eclypso/a0/agent-zero" conda run -n a0 python -m py_compile helpers/litellm_transport.py helpers/llm_result.py; git diff --check.
Update the docs index, quickstart, and installation guide to present A0 Launcher, A0 Install, and direct Docker as distinct setup paths.
Add the A0 Launcher v0.9 download matrix, document headless installer usage, and align visible architecture labels with the x86/ARM wording.
Add a Launcher guide with runtime setup, Installs, Instances, and screenshot capture notes.
Link the guide from README, quickstart, and docs index, and clarify the v1.20-to-v2.0 image upgrade path with backup/restore screenshots.
Introduce a new prompt "protocol" separate from extras: add LoopData.protocol_temporary and protocol_persistent, include protocol contents before message history during prompt construction, and clear temporary protocol data each turn. Add helper _build_context_message to render protocol/extras, update response input conversion to include protocol, and move project instructions & active/loaded skills injection into protocol. Update docs, prompts, plugin metadata, the SkillsTool message, and add tests to verify protocol placement and behavior.
Add oMLX as a first-class local chat and embedding provider using the Docker-friendly host.docker.internal:8000/v1 default and LiteLLM hosted_vllm wiring. Mark it as a no-key local provider, include it in onboarding with a bundled logo, document Apple Silicon setup including paged SSD cache, and cover the provider defaults, model search behavior, and onboarding metadata with focused tests.
Add shipped local-provider defaults for LM Studio and Ollama so Dockerized Agent Zero can reach host-running services without manual api_base edits. Map host.docker.internal in the sample Compose file for Linux Docker, document the behavior, and cover the provider defaults and compose mapping with focused regression tests.
Introduce DOX (AGENTS.md) contracts across the repository to formalize ownership and local work contracts: adds .github/AGENTS.md plus AGENTS.md files for agents, api, conf, docker, docs, extensions, helpers, knowledge, lib, plugins, prompts, scripts, skills, tests, tools, usr, webui (and several subfolders). Update root AGENTS.md (content and last-updated date) to include DOX framework guidance and a child DOX index. Update .gitignore to allow usr/ and usr/plugins AGENTS.md files. Remove legacy deep-dive files under docs/agents (banners, components, modals, plugins) and migrate frontend/plugin references to webui/ and plugins/ DOX locations. Also adjust plugins/README.md and several skills/*/SKILL.md files to align with the new DOX layout.
Split the legacy core speech stack into two built-in, independently toggleable plugins: `_kokoro_tts` for TTS and `_whisper_stt` for STT.
This refactor keeps dependency installation and bootstrap concerns in Docker/bootstrap/preload, while moving speech-specific tooling, APIs, prompts, UI, and runtime behavior into the plugins. Core now exposes engine-agnostic `tts-service` and `stt-service` brokers, with browser-native TTS preserved as the fallback when Kokoro is disabled.
Included in this change:
- add built-in `_kokoro_tts` plugin with plugin-owned synth API, config, status UI, and provider registration
- add built-in `_whisper_stt` plugin with plugin-owned transcribe API, mic runtime, device UI, prompt injection, and provider registration
- remove legacy core speech APIs/helpers/settings/UI and delete unused `webui/js/speech_browser.js`
- replace the old hardcoded speech settings section with a generic voice surface backed by plugin extensions
- update preload/docs/tests to match the new plugin-owned speech architecture
Behavioral intent:
- both plugins are built-in but not `always_enabled`
- users can now hot-switch TTS and STT independently
- browser TTS remains available when `_kokoro_tts` is off
- Whisper mic UI only appears when `_whisper_stt` is enabled
Replace the plugin list activation dropdown and advanced shortcut with a one-click ON/OFF switch. Keep project/profile-specific activation inside the plugin config flow, remove the old advanced-only modal, update plugin docs, and add regression coverage for the binary list toggle contract.
Deep-merge model preset slots with the active configuration so custom context windows, rate limits, and nested kwargs survive preset switches.
Treat legacy utility preset defaults as implicit values, allow omitted utility and embedding slots to inherit configured models, and document the partial-preset behavior.
Rename high-impact skills to task-oriented names and move plugin-owned skills into their owning plugin folders.\n\nAlign renamed skill frontmatter with the official SKILL.md standard by keeping trigger language in name/description metadata, replacing the old create-skill wizard with build-skill, and updating browser, A0 connector, computer-use, CLI setup, and scheduler skill references.\n\nTighten the recurring cross-provider guidance gaps surfaced by the evidence sweeps: memory requests now avoid promptinclude-file routing, scheduler prompts distinguish cron schedules from planned ISO dates, document questions prefer document_query, skills_tool search/read_file usage is clearer, normal notifications set info/priority 10, and local/host text editors preserve patch intent.\n\nUpdate regression tests for the renamed skills, plugin ownership, prompt budget reality, and standard frontmatter shape.
Summarize the provider evidence sweeps into a scored Agent Zero tool-efficiency chart.\n\nDocument the scoring rubric, ranked provider/model results, recurring failure clusters, applied prompt/skill improvements, and follow-up candidates while keeping the raw EVIDENCE files untracked.
Add a screenshot-led guide for the guided onboarding wizard using Agent Zero API as the example setup. Link the guide from the docs hub, quickstart, and installation setup flow, and include optimized screenshots for choosing Cloud, selecting a provider, choosing models, and reaching the ready state.
Refresh README, quickstart, and the docs index around Browser, Desktop, A0 CLI, projects, memory, skills, profiles, and model presets.
Add optimized scoped screenshots for Web UI, Browser, Desktop, and CLI workflows.
Trim architecture-heavy developer pages toward DeepWiki handoffs so local docs stay focused on practical user guidance.