This release polishes the agent picker UI, adds context-usage gauge warning states, improves config handling, and expands sandbox template tooling.

What's New

• Adds warning and compacting states to the context-usage gauge: color escalates (orange ≥75%, red ≥95%) as usage approaches the auto-compaction threshold, and displays "compacting…" while a compaction runs (applies to both the main TUI sidebar and lean TUI status line)
• Installs the latest docker-mcp plugin in the sandbox template image so agents running in that environment have MCP tooling available out of the box
• Adds a hint when a config key requires a newer schema version, guiding users to bump the top-level version field instead of seeing a generic "unknown key" error
• Adds WithModelOptions passthrough in teamloader for provider-agnostic HTTP transport wrapping

Bug Fixes

• Fixes config parsing failure when a hook event list is written as a single mapping instead of a sequence, which previously caused aliases and settings to be silently dropped

Technical Changes

• Refactors agent picker status bars to use dialog.RenderHelpKeys, removing duplicated key-rendering logic
• Extracts fitHelpPairs and adds regression tests for details help wrapping
• Polishes agent picker visuals: more compact cards (height 7 → 5) and tightened layout
• Simplifies the Dockerfile

What's Changed

• docs: update CHANGELOG.md for v1.102.0 by @docker-read-write[bot] in #3541
• docs: sync documentation with recent code changes by @aheritier in #3544
• style(tui): polish agent picker layout, cards, and status bars by @dgageot in #3543
• feat: install latest docker-mcp plugin in sbx template by @dgageot in #3545
• fix(config): accept a single mapping for hook event lists by @Sayt-0 in #3546
• feat(teamloader): add WithModelOptions passthrough for provider-agnostic HTTP transport wrapping by @simonferquel-clanker in #3549
• feat(config): hint when a config key requires a newer schema version by @dgageot in #3550

Full Changelog: v1.102.0...v1.103.0

This release brings significant enhancements to the Kanban board (drag & drop, card management, shell access, sandbox fixes), a new ChatGPT provider, TUI layout customization, and numerous stability and usability fixes across the board, session management, and copy/paste experience.

What's New

• Adds a chatgpt provider enabling sign-in with a ChatGPT Plus/Pro/Business account via docker agent setup, without requiring an OPENAI_API_KEY
• Adds /custom slash command to open a layout customization dialog for sidebar position and section visibility, with live preview and persistent settings
• Adds a Section spacing selector to the /custom layout dialog (Compact / Normal / Relaxed)
• Adds warning and compacting states to the context-usage gauge, escalating color as usage approaches the compaction threshold and showing "compacting..." during compaction
• Adds s shortcut on the board to open an interactive shell in the selected card's worktree
• Adds digit keys 1–9 to move a board card directly to a numbered column, and adds mouse drag & drop to reposition cards
• Adds drag ghost preview showing where a dragged card will be inserted in the drop-target column
• Adds intermediate startup statuses (starting → loading → attaching) on board cards to show boot progress
• Adds project editing and reordering on the board: press e or Enter on a project to rename it, with confirm-delete for destructive removals
• Adds desktop notifications with hooks tip to the documentation, covering macOS and Linux examples for on_user_input and stop events

Improvements

• Aligns TUI board card colors with the web board's status palette; starting/loading/attaching render in blue, running in green, and paused in a neutral color
• Fades the dragged card immediately on the first mouse motion during a drag, providing instant visual feedback
• Clarifies the dirty-worktree removal prompt by listing the consequences of y and N on separate lines before asking the question
• Improves command categories in the help output
• Documents env vars needed for notify-send hooks in detached sessions (SSH, tmux, containers)

Bug Fixes

• Fixes board agent startup failures being silently discarded, leaving cards stuck in "starting" forever; failures are now surfaced on the card
• Fixes board control-plane sockets being placed inside the data dir, which prevented cards from starting in Docker sandboxes; sockets are now placed in the system temp dir
• Fixes board storing absolute paths that broke shared state in sandboxes; paths are now stored home-relative
• Fixes board not forwarding --config-dir, --data-dir, and --cache-dir overrides to spawned agents
• Fixes board path expansion, state load, and git repo check edge cases
• Fixes stale drag state and ignores wheel events mid-drag on the board
• Fixes a race condition where RunSession/recallSession clobbered the attached runtime's cancel, causing DeleteSession to abort in-flight streams incorrectly
• Fixes session DB being reset on transient open errors (e.g. Ctrl-C, SQLITE_BUSY), which could silently wipe access to all past sessions
• Fixes SQLite memory database opening before acquiring the advisory file lock, causing SQLITE_BUSY races under concurrent access
• Fixes TUI selection copy being unreliable (random one-word copies, trailing padding, UI glyphs in clipboard) and improves copy button affordances
• Fixes redundant copy toast appearing alongside the inline "copied" flash when clicking a copy button
• Fixes emphasized text (e.g. "Esc to interrupt") being invisible on light themes due to incorrect color mapping
• Fixes global config file management: resolves a TUI panic on malformed config and several silent data-loss paths in the load/save cycle
• Fixes DNS resolution happening at construction time in NewSSRFSafeTransport, which could slow agent startup or fail silently when DNS is unavailable
• Fixes board reporting why a relaunch failed when attaching to an errored card, and drops per-card controller state without racing in-flight relaunches
• Fixes docs table of contents sidebar overlapping the footer on long pages by switching from position:fixed to position:sticky

Technical Changes

• Renames $BOARD_EDITOR environment variable to $DOCKER_AGENT_BOARD_EDITOR; the old name is kept as a fallback for one release
• Folds the ChatGPT sign-in flow into docker agent setup rather than a separate command
• Removes the agent catalog example from documentation
• Adds a lint rule flagging DNS resolution (net.Lookup* / Resolver.Lookup*) in constructors

What's Changed

• docs: update CHANGELOG.md for v1.101.0 by @docker-read-write[bot] in #3506
• chore: bump direct Go dependencies by @dgageot in #3507
• fix(board): surface agent startup failures instead of silently doing nothing by @dgageot in #3508
• feat(tui): add /custom command to customize layout by @Sayt-0 in #3509
• fix(board): store paths home-relative so shared state works in sandboxes by @dgageot in #3510
• chore(board): rename $BOARD_EDITOR to $DOCKER_AGENT_BOARD_EDITOR by @dgageot in #3514
• feat(board): edit, reorder, and confirm-delete projects by @dgageot in #3513
• fix(board): run agent control-plane sockets outside the data dir so cards start in docker sandboxes by @dgageot in #3515
• feat(board): show intermediate startup statuses on cards by @dgageot in #3516
• feat(board): add s shortcut to open a shell in the card's worktree by @dgageot in #3517
• fix(board): forward directory overrides to spawned agents by @dgageot in #3518
• fix(board): clarify dirty-worktree removal prompt with explicit consequences by @dgageot in #3520
• feat(board): align TUI card colors with web board status palette by @dgageot in #3521
• feat(board): move cards to any column with digit keys and drag & drop by @dgageot in #3522
• docs: sync documentation with recent main merges by @aheritier in #3519
• Improve the command categories by @dgageot in #3523
• feat(board): fade dragged card immediately and fix stale drag state by @dgageot in #3524
• fix: don't clobber attached runtime's cancel in RunSession/recallSession by @dgageot in #3525
• docs: add desktop notifications with hooks tip by @dgageot in #3526
• feat(provider): add chatgpt provider with ChatGPT account sign-in by @Sayt-0 in #3512
• fix(docs): switch .toc-aside from position:fixed to sticky in flex .main row by @dgageot in #3527
• fix(userconfig): harden global config file management by @Sayt-0 in #3529
• fix(tui): unreadable emphasized text on light themes by @Sayt-0 in #3530
• docs: document env vars for notify-send hooks in detached sessions by @Sayt-0 in #3531
• feat(board): preview the dragged card at its insertion point by @dgageot in #3532
• feat(tui): add section spacing setting to /custom layout dialog by @Sayt-0 in #3533
• fix(session): don't reset the session DB on transient open errors by @dgageot in #3534
• fix(tui): make selection copy reliable and improve copy affordances by @Sayt-0 in #3535
• fix(memory): open sqlite db under file lock to avoid SQLITE_BUSY races by @dgageot in #3538
• feat(tui): add warning/compacting states to the context-usage gauge by @Sayt-0 in #3537
• fix(tui): drop the redundant copy toast when the inline "copied" flash shows by @dgageot in #3539
• fix(ssrf): avoid DNS resolution at construction time in NewSSRFSafeTransport by @dgageot in #3540

Full Changelog: v1.101.0...v1.102.0

This release adds a setup wizard for first-time model configuration, an opt-in auto theme that follows the terminal's light/dark background, and several usability and bug fixes.

What's New

• Adds an Open Board button (and b keybinding) to the agent picker, allowing users to launch the Kanban board directly without first running an agent
• Adds a docker agent setup wizard, offered when no usable model is configured, guiding users through setting up a cloud API key or pulling a local model
• Adds secret stores and a default config environment file source for environment configuration
• Adds an opt-in auto theme that follows the terminal's light/dark background at startup and live (via settings.theme: auto, --theme auto, or the /theme menu)
• Exports Pull for pre-confirmed model pulls in the Docker Model Runner integration

Bug Fixes

• Fixes the agent picker's board button hit-zone math and improves layout on narrow terminals
• Fixes --session-db default so it correctly resolves relative to --data-dir, making sessions stored under a custom data directory visible
• Fixes the theme file watcher that was lost in the tab-view rewrite, restoring live hot-reload of ~/.cagent/themes/*.yaml while the TUI is running

Technical Changes

• Adds a "Set Up a Model" getting-started tutorial covering both API key and local model setup paths
• Adds session DB wiring tests and documentation follow-ups for the setup wizard

What's Changed

• docs: update CHANGELOG.md for v1.100.0 by @docker-read-write[bot] in #3497
• feat(picker): add Open Board button to the agent picker by @dgageot in #3498
• feat(cli): add docker agent setup wizard, offered when no model is usable (phase 3 of #3442) by @Sayt-0 in #3499
• docs(getting-started): add Set Up a Model tutorial for API key and local paths by @Sayt-0 in #3502
• fix(cli): resolve session DB default against the data dir by @dgageot in #3503
• feat(tui): opt-in "auto" theme that follows the terminal light/dark background by @Sayt-0 in #3504
• fix(tui): rewire theme watcher lost in tab-view rewrite by @Sayt-0 in #3505

Full Changelog: v1.100.0...v1.101.0

This release adds new diagnostic and configuration capabilities, hardens the board communication protocol, and improves error messaging for missing models and credentials.

What's New

• Adds docker agent doctor command for diagnosing model provider credentials and agent readiness
• Adds actionable errors when models or credentials are missing, with guidance on next steps
• Adds hooks.d drop-in directory (<config-dir>/hooks.d/*.yaml) and DOCKER_AGENT_CONFIG_DIR environment variable override for config directory
• Adds heartbeat idle watchdog to abort hung event streams in docker agent board
• Adds session aggregate cost reporting to GET /snapshot
• Adds machine-readable error codes to 404 responses for unknown session snapshots
• Adds heartbeat keepalives to idle /events SSE streams

Bug Fixes

• Fixes turn-boundary events being silently dropped when a subscriber's buffer overflows

Technical Changes

• Extracts lean TUI engine into a separate package and reorganizes UI components (status models, inline images, tool views, transcript, screen model) into a dedicated UI layer
• Adds documentation for docker agent board CLI reference and actionable model/credential errors in troubleshooting guide
• Adds tmux usage guidance for running docker agent board in a sandbox environment

What's Changed

• docs: update CHANGELOG.md for v1.99.0 by @docker-read-write[bot] in #3487
• feat(cli): actionable errors for missing models and credentials (phase 1 of #3442) by @Sayt-0 in #3452
• feat(cli): add docker agent doctor for model and credential diagnosis by @Sayt-0 in #3489
• Lean tui UI split by @rumpl in #3492
• feat: harden control-plane protocol for reliable board communication by @dgageot in #3491
• docs: update CLI reference and troubleshooting for board and actionable errors by @aheritier in #3494
• Add tmux or running docker agent board in sbx by @dgageot in #3495
• feat: hooks.d drop-in directory and config-dir env override by @Sayt-0 in #3496

Full Changelog: v1.99.0...v1.100.0

This release adds significant new capabilities including a Kanban board TUI, NVIDIA provider support, parallel tool dispatch, and numerous compaction and context-window improvements, along with several important bug fixes.

What's New

• Adds docker agent board, a full-screen Kanban TUI for orchestrating multiple agents across pipeline stages (Dev → Review → Push → Done), each on an isolated git worktree
• Adds NVIDIA (NIM / build.nvidia.com) as a supported model provider via a new nvidia alias
• Adds /context slash command with a categorized context-window breakdown dialog, including a stacked usage bar and copy-to-clipboard action
• Adds the ability to list and drop attached files directly from the /context dialog
• Adds wait_background_job tool to the shell toolset, blocking until a background job finishes instead of requiring polling
• Adds shell background job recall, routing completed job output back to active or idle sessions
• Splits background jobs into a separate background_jobs toolset, leaving the shell toolset with only synchronous shell tools
• Adds unix:// URL support in the remote MCP client for connecting over Unix domain sockets
• Adds a Lean Mode checkbox to the agent picker (--agent-picker), toggled with l or mouse click
• Adds compaction_threshold as a configurable key per agent (and per model), replacing the previously hardcoded 0.9 constant
• Adds per-sub-agent context accounting in the TUI sidebar and agent inspector
• Runs tool dispatcher calls in parallel, serializing only interactive confirmations
• Extends the HCL file() function with an optional variables argument to render the file as a template
• Groups the /sessions browser by current workspace instead of showing a flat global list

Improvements

• Reconciles the compaction token estimator with provider-reported usage, using exact counts where available and a session-calibrated correction for unreported content
• Wraps lean TUI tool calls in a padded themed box for improved rendering
• Supports steering (sending messages while the agent is running) in the lean TUI, with pending messages displayed in muted styling

Bug Fixes

• Fixes compaction replacing session history with a stale reply when the summarization model returns an empty response
• Fixes compaction outcome reporting to accurately reflect applied/skipped/failed status on the completed event
• Fixes whitespace-only compaction summaries being applied instead of treated as a no-op
• Fixes --yolo flag not being applied correctly on session resume, causing safer_shell to prompt despite the flag
• Fixes Gemini models failing with MCP tools that declare non-string enum values in their JSON schemas
• Fixes session Message.AgentName silently dropping on load when reading files using the legacy agentName JSON key
• Fixes database migration rollback errors being silently swallowed
• Removes Anthropic Files API usage
• Updates stale Anthropic model IDs in examples to current IDs

Technical Changes

• Standardizes agent_name JSON tag on Message.AgentName, replacing the previous agentName camelCase tag
• Shares git-branch detection logic between the lean and full TUIs via a new pkg/gitbranch package
• Refactors lean TUI model into cohesive subsystems (usage/tool trackers, transcript type) with no functional change
• Replaces context-smuggled emitters with an explicit tools.Runtime handle in the tool dispatcher

What's Changed

• docs: update CHANGELOG.md for v1.98.0 by @docker-read-write[bot] in #3445
• feat(tui): group /sessions browser by current workspace by @Sayt-0 in #3446
• docs: point rel=canonical at the docs.docker.com mirrored pages by @Sayt-0 in #3447
• feat(config): make the auto-compaction threshold configurable (compaction_threshold) by @Sayt-0 in #3448
• feat(compaction): reconcile token estimator with provider-reported usage by @Sayt-0 in #3449
• feat(mcp/remote): support unix:// URLs in remote MCP client by @maxcleme in #3453
• ci: build and publish docker-agent sandbox templates by @dgageot in #3454
• feat(tui): add /context command with a categorized context-window breakdown by @Sayt-0 in #3458
• refactor(leantui): decompose the model into cohesive subsystems by @rumpl in #3455
• refactor: share git-branch detection between the lean and full TUIs by @rumpl in #3457
• fix: box lean TUI tool calls by @rumpl in #3460
• fix: remove Anthropic Files API usage by @rumpl in #3466
• feat(hcl): render file() as a template when given variables by @rumpl in #3464
• feat(shell): add wait_background_job tool by @simonferquel-clanker in #3463
• docs: sync documentation with recent changes by @aheritier in #3468
• feat: run tool dispatcher calls in parallel by @rumpl in #3467
• feat: add shell background job recall by @rumpl in #3461
• feat: add NVIDIA provider as supported model provider by @Priyanshu-sde in #3421
• fix: support steering in lean TUI by @rumpl in #3474
• Tweak the golang dev by @rumpl in #3472
• Split background jobs into separate toolset by @rumpl in #3473
• docs: document lean TUI steering support by @aheritier in #3475
• feat(tui): list and drop attached files from the /context dialog by @Sayt-0 in #3465
• docs: switch github.io site from Jekyll to Hugo by @Sayt-0 in #3459
• feat: add docker agent board, a Kanban TUI for orchestrating agents by @dgageot in #3451
• fix: update stale Anthropic model IDs in examples by @dgageot in #3480
• chore: bump direct Go dependencies by @dgageot in #3481
• fix: backfill SafetyPolicy on session resume with --yolo by @melmennaoui in #3482
• chore: standardize agent_name JSON tag in Message struct by @Piyush0049 in #3471
• fix: handle tx.Rollback error in database migrations by @Piyush0049 in #3476
• fix: compaction data loss on empty summary + honest compaction outcome reporting by @dgageot in #3470
• fix(session): accept legacy agentName JSON key when unmarshaling Message by @dgageot in #3483
• fix(gemini): stringify non-string enum values in tool schemas by @Sayt-0 in #3485
• feat(tui): per-sub-agent context accounting in the TUI by @Sayt-0 in #3486

New Contributors

• @Priyanshu-sde made their first contribution in #3421
• @Piyush0049 made their first contribution in #3471

Full Changelog: v1.98.0...v1.99.0

This release adds several new TUI features including an interactive getting-started tour, an effort picker dialog, and Lean Mode support in the agent picker, along with stability fixes for startup, shutdown, and proxy recovery.

What's New

• Adds an interactive getting-started tour offered on first run and replayable on demand, implemented as a floating card that observes the real UI's message stream
• Opens an effort picker dialog when /effort is used with no argument, instead of printing a usage hint
• Adds a Lean Mode checkbox to the agent picker panel, toggleable with the l key or mouse click
• Agent picker now discovers ~/.agents config files (.yaml, .yml, .hcl) when no explicit list is given, in addition to the built-in default and coder agents
• Adds cross-process locking and atomic writes for SQLite memory storage

Bug Fixes

• Fixes toolset Start during startup so the sidebar tools spinner can no longer animate forever when Docker is wedged
• Fixes TUI exit being blocked by a wedged resource cleanup
• Fixes --record proxy bypassing the configured models gateway, which caused HTTP 401 errors when gateway-managed credentials were expected
• Hardens gateway forwarding in record mode against credential leakage
• Fixes proxy recovery after cooldown — previously a socket error would permanently latch the transport to direct connections for the lifetime of the process
• Fixes /tools dialog showing Go type names (e.g. *skills.ToolSet) instead of human-readable names for loader-created toolsets
• Fixes telemetry incorrectly recording an error on a chat span after a successful LLM completion
• Fixes agent picker hardening: windowing, escapes, FIFO skip, sentinel, and description guard
• Keeps agent picker panel geometry stable and windowing math in sync

Technical Changes

• Fixes flaky frame synchronization in TUI test driver sendSync
• Fixes file descriptor double-close corrupting parallel tests in TestListen_FD
• Guards fdOwnershipPin with a mutex to fix a data race in server tests
• Trims comments in the fallback transport

What's Changed

• test: fix flaky tuitest frame sync and TestListen_FD fd double-close by @dgageot in #3420
• fix: bound toolset startup and never block the TUI exit when Docker is wedged by @dgageot in #3423
• docs: update CHANGELOG.md for v1.97.0 by @docker-read-write[bot] in #3424
• feat(tui): open effort picker when /effort has no argument by @Sayt-0 in #3425
• fix: route --record proxy through the configured models gateway by @dgageot in #3428
• docs: sync documentation with recent merges by @aheritier in #3430
• feat(tui): interactive getting-started tour on first run, replayable on demand by @Sayt-0 in #3427
• fix: show proper toolset names in /tools dialog for loader-created toolsets by @dgageot in #3441
• feat(memory): add cross-process locking and atomic writes for sqlite by @yunasora in #3114
• fix(remote): recover proxy after cooldown instead of latching to direct by @trungutt in #3431
• telemetry: don't record error on chat span after successful LLM completion by @jedp-docker in #3281
• feat(picker): add Lean Mode checkbox to agent picker by @dgageot in #3443
• feat: agent picker discovers ~/.agents configs when no list is given by @dgageot in #3444

New Contributors

• @yunasora made their first contribution in #3114

Full Changelog: v1.97.0...v1.98.0

This release adds global hook support in user config, expands environment variable references in hook and script-shell fields, and fixes hook field merging.

What's New

• Adds support for global hooks in user config (~/.config/cagent/config.yaml), with hooks merged in order: agent config → global → CLI
• Adds ${env.X} expansion in hook and script-shell working_dir and env fields, giving them the same treatment as other working_dir fields

Bug Fixes

• Fixes hook field merging so all HooksConfig fields are combined correctly when merging config and CLI hooks
• Fixes unset ${env.X} references to log at debug level instead of silently failing; config env now takes precedence over host env

Technical Changes

• Fixes flaky TestAttachedServer_DeleteEmitsSessionExited test caused by a race between session deletion and event log subscription

What's Changed

• docs: update CHANGELOG.md for v1.96.0 by @docker-read-write[bot] in #3417
• test: fix flaky TestAttachedServer_DeleteEmitsSessionExited by @dgageot in #3418
• feat(config): expand ${env.X} in hook and script-shell working_dir/env by @dgageot in #3416
• fix: merge all hook fields by @rumpl in #3419
• feat: support global hooks in user config by @rumpl in #3422

Full Changelog: v1.96.0...v1.97.0

This release adds session resumption across agent URL changes, TUI e2e test generation from recorded sessions, and fixes for duplicate tool names and Anthropic adaptive thinking display.

What's New

• Adds ability to resume sessions across agent URL query changes when using docker-agent serve api (e.g. when relaunching with a different agent version tag)
• Adds TUI e2e test generation from --record sessions, producing a ready-to-edit *_test.go file alongside the cassette with synchronization points derived from the recorded session
• Adds a Docker Wiki example agent with /init, /update, /status, and /help commands

Bug Fixes

• Fixes duplicate tool names being sent to providers by deduplicating in collectTools() — the first toolset wins and a warning is shown once per streak naming the conflicting toolsets (fixes Anthropic HTTP 400 "Tool names must be unique" errors)
• Fixes missing reasoning display in the TUI when using /effort or Shift+Tab thinking cycle on newer Claude models (Opus 4.7+) — the adaptive display field is now sent by default even when provider_opts.thinking_display is not explicitly configured

What's Changed

• docs: update CHANGELOG.md for v1.94.0 by @docker-read-write[bot] in #3406
• docs: update CHANGELOG.md for v1.95.0 by @docker-read-write[bot] in #3407
• docs: add docker wiki example agent by @rumpl in #3408
• feat(server): resume sessions across agent URL query changes by @gtardif in #3410
• fix(agent): dedupe duplicate tool names with once-per-streak warning by @Sayt-0 in #3409
• feat(tui): generate a TUI e2e test from a --record session by @dgageot in #3412
• docs: single-source documentation for the docs.docker.com Hugo module mount by @Sayt-0 in #3413
• fix(anthropic): default adaptive thinking display to summarized by @Sayt-0 in #3415
• docs: prepare pages for the docs.docker.com module mount by @Sayt-0 in #3414

Full Changelog: v1.95.0...v1.96.0

This release focuses on TUI rendering performance improvements for long conversations, along with a bug fix for test reliability and a configuration version update.

Improvements

• Memoizes scrollbar rendering across frames to reduce redundant computation during TUI redraws
• Memoizes content line widths in the scroll view to avoid recalculating unchanged values each frame
• Replaces lipgloss.JoinHorizontal with direct column zipping for scrollbar rendering, reducing allocations per frame

Bug Fixes

• Fixes restyled lines incorrectly using cached widths instead of re-measuring after style changes
• Fixes flaky renderer-registry tests that were racing on a shared global registry

Technical Changes

• Freezes config v11 and promotes v12 as the latest configuration version

What's Changed

• chore: bump direct Go dependencies by @dgageot in #3405
• freeze config v11 and start v12 as latest by @dgageot in #3401
• perf(tui): memoize rendering hot paths for long conversations by @dgageot in #3404

Full Changelog: v1.94.0...v1.95.0

This release significantly expands model provider support with 12 new built-in providers, adds mouse support and new commands to the TUI, introduces a SafetyPolicy primitive, and includes numerous bug fixes and performance improvements.

What's New

• Adds Baseten as a built-in model provider (provider: baseten)
• Adds OVHcloud AI Endpoints as a built-in model provider (provider: ovhcloud)
• Adds Groq as a built-in model provider (provider: groq)
• Adds DeepSeek as a built-in model provider (provider: deepseek)
• Adds Cerebras as a built-in model provider (provider: cerebras)
• Adds Fireworks AI as a built-in model provider (provider: fireworks)
• Adds Together AI as a built-in model provider (provider: together)
• Adds Hugging Face Inference Providers as a built-in model provider (provider: huggingface)
• Adds Moonshot AI (Kimi) as a built-in model provider (provider: moonshot)
• Adds Vercel AI Gateway as a built-in model provider (provider: vercel)
• Adds Cloudflare Workers AI and Cloudflare AI Gateway as built-in model providers
• Adds mouse support to the agent picker: hover highlights a card, single click selects it, double-click immediately starts the session, and scroll wheel navigates the YAML details panel
• Adds metadata tags field to agent config and surfaces tags as coloured chips in the agent picker
• Adds /effort slash command to select the model's reasoning level directly (available in both full and lean TUI)
• Adds SafetyPolicy primitive (unsafe/safer/strict) to sessions, forwarded to hooks via hooks.Input.SafetyPolicy
• Enforces lifecycle.startup_timeout for MCP/LSP toolset startup, which was previously parsed but never acted on
• Adapts safer_shell hook classification to the session SafetyPolicy
• Collapses file picker keyboard shortcuts to a single line on wide dialogs
• Strengthens default memory toolset instructions

Improvements

• Memoizes successful OCI config reads per process, halving warm startup latency for OCI agent refs
• Suppresses spurious empty-response warning for benign post-tool stops in forked-skill sub-sessions
• Agent picker dialog is larger and truncates long lines instead of wrapping; cards are 70 columns wide

Bug Fixes

• Fixes coalescing of system messages for self-hosted and OpenAI-compatible endpoints that reject multiple system messages (e.g. vLLM with Qwen)
• Fixes OVHcloud default model to Qwen3.5-397B-A17B
• Fixes macOS Option-key characters not triggering file picker alt+h/alt+i visibility toggles
• Removes duplicate "Initializing MCP servers…" spinner from the sidebar that fired on every turn even for agents with no MCP servers
• Adds 5-second timeout to graceful-shutdown calls that previously used WithoutCancel without a deadline, preventing indefinite blocking
• Fixes concurrent Connect races and goroutine leak on startup timeout for MCP/LSP toolsets
• Fixes auto-deny of tool asks in non-interactive sessions instead of blocking indefinitely
• Fixes tool-call confirmations being dropped without a Resume under --yolo in JSON mode
• Fixes Anthropic provider falling back to token thinking for effort levels on models without adaptive thinking support (e.g. Haiku 4.5)
• Fixes tool call render briefly disappearing or shrinking while JSON arguments are still streaming in the lean TUI
• Clamps max_tokens to the context window for OpenAI-compatible providers to prevent "context window exceeded" errors on self-hosted vLLM
• Fixes cost and context limit display in the lean TUI

Technical Changes

• Derives provider API-key env vars forwarded to eval containers from the provider registry instead of a hard-coded list
• Refactors provider client boilerplate into shared helpers across openai, anthropic, gemini, bedrock, and dmr clients
• Consolidates OpenAI-compatible alias provider tests into a single shared test
• Extracts shared calleeObject helper in lint cops to remove duplication
• Adds OTelTracerName lint cop to enforce scoped OpenTelemetry tracer names
• Replaces 44 time.Sleep calls in the test suite with deterministic synchronization primitives

What's Changed

• docs: update CHANGELOG.md for v1.93.0 by @docker-read-write[bot] in #3339
• docs: sync documentation with recent main merges by @aheritier in #3340
• feat: add Baseten provider support by @dgageot in #3341
• feat: add OVHcloud AI Endpoints provider support by @aheritier in #3343
• chore: bump github.com/anthropics/anthropic-sdk-go to v1.55.0 by @dgageot in #3345
• feat: add Groq as a supported model provider by @Sayt-0 in #3358
• fix(openai): coalesce system messages for self-hosted and open-model endpoints by @Sayt-0 in #3357
• feat: add DeepSeek as a supported model provider by @Sayt-0 in #3361
• docs: group and collapse left navigation sections by @aheritier in #3360
• docs: list providers alphabetically by name in nav by @aheritier in #3364
• fix(docs): restore inline code contrast in light mode by @aheritier in #3365
• feat: collapse file picker shortcuts to one line on wide dialogs by @dgageot in #3366
• fix(tui): remove duplicate MCP-init spinner from sidebar by @dgageot in #3367
• fix: add 5s timeout to graceful-shutdown calls that used WithoutCancel by @dgageot in #3370
• feat: add Cerebras as a supported model provider by @Sayt-0 in #3368
• fix(tui): accept macOS Option-key chars for file picker toggles by @Sayt-0 in #3333
• feat: add Fireworks AI as a supported model provider by @Sayt-0 in #3369
• test: consolidate OpenAI-compatible alias provider tests by @dgageot in #3374
• refactor(config): derive provider API-key env vars from config by @dgageot in #3375
• feat: enforce lifecycle.startup_timeout for MCP/LSP toolset startup by @dgageot in #3373
• feat: add Hugging Face Inference Providers as a supported model provider by @Sayt-0 in #3379
• docs: tighten comment guidance in AGENTS.md by @dgageot in #3380
• fix: unblock headless tool-Ask paths (dispatcher + JSON runner) by @melmennaoui in #3377
• feat(session): add SafetyPolicy primitive and hook pass-through by @melmennaoui in #3378
• feat: add Together AI as a supported model provider by @Sayt-0 in #3376
• fix(anthropic): use token thinking for effort levels on models without adaptive support by @Sayt-0 in #3381
• fix(tui): keep tool call render stable while args stream by @rumpl in #3382
• feat(hooks/safer_shell): adapt classification to session SafetyPolicy by @melmennaoui in #3383
• feat: add Moonshot AI (Kimi) as a supported model provider by @Sayt-0 in #3384
• feat: add Vercel AI Gateway as a supported model provider by @Sayt-0 in #3385
• feat: add Cloudflare Workers AI and AI Gateway as model providers by @Sayt-0 in #3389
• test: make models-list tests hermetic and parallelizable by @dgageot in #3388
• feat(lint): add OTelTracerName cop to enforce scoped tracer names by @dgageot in #3386
• feat: add tags field to agent config metadata by @dgageot in #3390
• feat: add mouse support to the agent picker (hover, double-click, wheel) by @dgageot in #3391
• fix(runtime): suppress spurious empty-response warning for benign post-tool stops by @dgageot in #3392
• feat(memory): strengthen default toolset instructions by @dgageot in #3395
• fix: clamp max_tokens to the context window for OpenAI-compatible providers by @Sayt-0 in #3393
• test: make the test suite faster and sleep-free by @dgageot in #3399
• perf(config): memoize successful OCI reads to halve warm startup latency by @dgageot in #3397
• fix: show lean tui cost and context limit by @rumpl in #3400
• refactor: extract shared calleeObject helper in lint cops by @dgageot in #3396
• feat: show metadata tags as coloured chips in the agent picker by @dgageot in #3394
• refactor(provider): deduplicate provider client boilerplate by @dgageot in #3398
• feat(tui): add /effort command to select the model's reasoning level by @Sayt-0 in #3403

Full Changelog: v1.93.0...v1.94.0