Agent! 1.0.76.166

🧠 Apple Intelligence

• Add run_agent tool so Apple AI can launch agents via FoundationModels
• Persist accessibility agent session across turns
• Use Apple AI itself to classify follow-up prompts
• Gate Accessibility intent parsing on Accessibility permission
• Skip accessibility triage for file-path and coding prompts
• Skip triage when project folder is active (unless a system app is named)
• Skip triage when Accessibility permission is not granted
• Better app name recognition — fixes Photo Booth confusion
• Stop biasing accessibility tool toward Photo Booth
• Teach accessibility agent to discover button labels before clicking
• Refuse destructive UI actions during triage
• Skip tool + screenshot bypass to stop hallucinated clicks
• Inject previous conversation turn into rephrase prompt
• Fix isEnabled defaulting to false
• Fix misrouting non-agent tool names to run_agent
• Tighten run_agent trigger to exact phrase only
• Fix duplicate log entry for accessibility tasks

🎨 UI & Design

• 🧠 Brain icon: AngularGradient multi-color blend for active Apple AI toggles
• Tint Apple Intelligence toggles with distinct colors (red/orange/blue/green)
• Mediator: gray brain icon when disabled (was red)
• Accessibility intent toggle: teal → deep blue
• Add separate Triage greetings toggle
• Add 3px to LLMOutputBox height to prevent cursor blink jitter
• Settings: add refresh button to Claude model picker
• Mirror outgoing iMessage into Messages tab

📸 Images & Attachments

• Paste: popup to pick image resolution, remembers last choice
• Screenshot button: apply same resolution popup as paste
• Cache hi-res originals of pasted/screenshot attachments
• Auto-repair U+202F narrow-no-break-space in screenshot paths
• Add copy_image action to the file tool

🔧 Bug Fixes

• Fix accessibility handler: accept appBundleId key in manage_app, click_menu_item, set_window_frame
• Fix bump_version/get_version to use selected project instead of Xcode open documents
• Fix iMessage send lockup: use osascript subprocess instead of in-process NSAppleScript
• Fix double echo: skip appendLog for shell commands that already streamed output
• Fix agent script run routing: script tabs run on same tab, main tab spawns separate tab
• Fix: Token Compression toggle now guards every compression entry point
• Fix: respect Token Compression toggle for ALL tiers, not just Apple AI summarization
• Replace force-unwrap with if-let in diff_and_apply rangeNote
• Fix tool output formatting: newline prefix and :0:0 warning cleanup

✏️ Diff & Apply

• diff_and_apply: actually use the LLM-provided source snippet
• Fix diff_apply truncation: splice source into file instead of replacing entire content
• Fix LLMs sending diff-formatted input to diff_apply
• edit_file: show line numbers in activity log status line
• edit_file: log only the status line, skip D1F diff and full file dump
• FilePath on its own line at both log sites
• Put blank line inside the diff block

📖 Help System

• Add macOS Help Book bundle (Agent.help) with full Help Viewer compatibility
• Adopt CSS-variable pattern + color-scheme meta for dark mode
• Replace <kbd> with <code class="key"> to match Logos pattern
• Fix Help "content unavailable" — switch HPDBookType from remote (1) to local (3)
• Rebuild Agent.helpindex with hiutil
• Flatten Agent.help structure to match Logos (fixes invisible keyboard shortcuts)
• Switch target to hand-authored Info.plist so CFBundleHelpBookFolder/Name actually land

📚 Documentation

• Document Launch Daemon as optional, user-controllable
• Remove outdated Write Protection claim from security docs
• Document SMAppService re-signing empirical proof in SECURITY.md
• Explain why no manual setCodeSigningRequirement is needed
• Update founder's title, badge colors, formatting, copyright, and legal notices
• Remove poem section and founder cancer message from README
• Remove ReleaseNotes folder and contents

🔨 Maintenance & Chores

• Bump version to 1.0.76 (build 166)
• Pin AgentTools 2.51.6 + AgentD1F 1.0.3 + AgentSwift 1.1.1 + AgentTerminalNeo 1.37.2
• Add FUNDING.yml for GitHub Sponsors (macOS26)

Agent! 1.0.75 (build 164)

Agent! 1.0.75 (build 164)

🧠 Apple Intelligence Mediator

• Refresh master Enable Mediator toggle on popover appear so sub-features re-evaluate against current state
• Added separate Triage Greetings toggle
• Tinted Apple Intelligence toggles with distinct colors
• Brain icon now blends colors of active Apple-AI toggles via AngularGradient
• Smoother multi-toggle blend with interleaved midpoint colors
• Gray brain icon when Mediator disabled (was red)
• Accessibility intent toggle recolored teal → deep blue
• Gate Accessibility intent parsing on Accessibility permission
• Apple AI: skip tool + screenshot bypass to stop hallucinated clicks
• Apple AI: refuse destructive UI actions during triage
• Fix two bugs behind "Apple AI claimed Safari opened" false positive

📘 Help Book

• Added full macOS Help Book bundle (Agent.help) with Help Viewer compatibility
• Switched Agent! target to hand-authored Info.plist so CFBundleHelpBookFolder/Name land correctly
• Fix Agent Help "content unavailable" — HPDBookType 1 → 3 (local)
• Removed remote URL from Help book (local-only Pull Help)
• Flattened Agent.help structure to match Logos (fixes invisible keyboard shortcuts)
• Replaced <kbd> with <code class="key"> to match Logos pattern
• Adopted CSS-variable pattern + color-scheme meta for dark mode
• Fixed Keyboard Shortcuts help readability
• Rebuilt Agent.helpindex with hiutil

🛠️ Diff & Apply

• diff_and_apply: actually use the LLM-provided source snippet
• Put the blank line inside the diff block
• Added blank line between diff block and status line
• filePath onto its own line at both log sites
• Added copy_image action to the file tool

🗜️ Token Compression

• Token Compression toggle now guards every compression entry point
• Respect Token Compression toggle for ALL tiers, not just Apple AI summarization

📸 Screenshots & Attachments

• Paste: popup to pick image resolution, remembers last choice
• Screenshot button: apply same resolution popup as paste
• Cache hi-res originals of pasted/screenshot attachments
• Auto-repair U+202F narrow-no-break-space in screenshot paths

💬 Messaging & Tabs

• Fix iMessage send lockup: use osascript subprocess instead of in-process NSAppleScript
• Mirror outgoing iMessage into Messages tab
• Record tool steps on tab tasks so Steps list renders
• Mark tab tool steps red when tool_result content indicates failure
• Detect JSON-shaped tool failures in Steps error heuristic

📊 Usage & Settings

• Usage popover: record subscription billing per-call, show Included
• Settings: added refresh button to Claude model picker
• Added 3px to LLMOutputBox height to prevent cursor blink jitter

🐚 Shell

• Shell tool: include exit code in tool_result content on failure

📚 Docs & Repo

• Point repo URLs at macos26/agent
• README: explain why no manual setCodeSigningRequirement is needed
• SECURITY: document SMAppService re-signing empirical proof
• Repin AgentD1F 1.0.3, AgentSwift 1.1.1, AgentTerminalNeo 1.37.2, AgentTools 2.51.6

Agent! v1.0.72 (Build 161)

✨ Features

• feat: add MiniMax provider support
• Claude auth: accept Claude Code OAuth tokens alongside API keys
• Claude OAuth: prepend Claude Code identity system block
• Claude models: OAuth-aware fetch, add Opus 4.7 to defaults

🐛 Bug Fixes

• Fix edit_file diff syntax highlighting in activity log
• Fence edit_file diff log with language for syntax coloring
• Skip Apple triage in TaskExecution when Accessibility is off
• Skip Apple triage when Accessibility permission is not granted
• Claude auth: strip whitespace/newlines from pasted credentials

🚀 Improvements

• Remove per-tool-result and per-message caps — full output to LLM
• read_file: remove 50K char cap — return full file to LLM
• Remove hardcoded default model lists — fetch is the only truth
• index_create/recreate/append/continue: return full JSONL body in same turn

📚 Documentation

• revert: restore full README (undo 6250603 short version)
• docs: clarify Launch Agent is not privileged
• Docs: add CLAUDE_CODE_OAUTH.md — how to mint + paste a claude setup-token

🔖 Release

• Bump version to 1.0.72 (161)

🔀 Merges

• Merge pull request #6 from octo-patch/feature/add-minimax-provider
• Merge branch 'main' of https://github.com/macOS26/Agent

🔧 Recent Fixes

• Add Codex LLM provider (CodexService.swift) + provider setup wiring
• New docs: docs/CODEX.md, docs/CODEX_OAUTH_RESEARCH.md
• TaskExecution refinements: Setup, ToolBatch, Guards, ErrorHandler, FileTools
• AgentViewModel: StuckGuard, LLMServices, ScriptTabs, ModelFetching, SubAgent, Logging improvements
• Apple Intelligence mediator + Claude service updates
• UI: FallbackChainView, SettingsView, ToolsView, NewMainTabSheet, ThinkingIndicatorView polish
• LaunchAgent/LaunchDaemon plist tweaks

🦾 Agent! 1.0.70 (Build 159)

What's Changed

OpenAI Model List Improvements

• Stop filtering OpenAI models — now returns every model the API lists, no more hidden models (a3cb7be)
• Dropped legacy GPT-3.5 and dated GPT-4 snapshot filtering (1a8e980)
• Future-proofed model filter with blocklist for non-chat models (b3404e2)

Apple Intelligence

• Skip accessibility triage when a project folder is active unless a system app is explicitly named — reduces unnecessary UI queries (f1c0959)

Full Changelog: https://github.com/toddbruss/Agent/compare/v1.0.69.158...1.0.70.159

v1.0.69.158

What's New in v1.0.69.158

🧠 Project Index & Memory

• Project Index tool — portable JSONL index per file (lines, doc block, top-level symbols)
• Memory tool — persistent /memories filesystem with global and project scopes
• Consolidated per-project hidden dirs under .agent/{index,memory,worktrees,plans}
• Prompt version now derived from marketing + build number (no more hardcoded revision)

✏️ Editing & Diff

• diff_apply fix — splice source into file instead of replacing entire content
• Fix diff_apply truncation and diff-formatted input handling
• Edit reliability: drop truncation guard, add cache invalidation + no-op detect
• Stuck-edit nudge: lower threshold to 3, make actionable

🍎 Apple Intelligence

• Apple AI: persist accessibility agent session across turns
• Apple AI: use Apple AI itself to classify follow-up prompts
• Apple AI: add run_agent tool for launching agents via FoundationModels
• Apple AI: disable shell and applescript tools — accessibility only
• Fix Apple AI misrouting non-agent tool names to run_agent
• Fix Apple Intelligence isEnabled defaulting to false

🎨 UI Polish

• Long Cmd+V pastes captured as removable chips with first+last char preview
• LLM output: stop last-word flickering with cursor
• LLM output: don't snap to bottom on mouse-exit when not streaming
• Tab labels: main tab shows current model, LLM tabs preserve raw model case
• ToolStepsView: persist expanded state across Cmd+B and new steps
• Multi-line log messages drop onto their own line below timestamp

🔧 Fixes & Cleanup

• Fix LLM tab provider/model clobber + per-tab usage breakdown
• Fix tool output formatting: newline prefix and :0:0 warning cleanup
• Replace force-unwrap with if-let in diff_and_apply
• Pin AgentTools 2.50.10 — adds PROJECT INDEX prompt section
• Efficient-action system prompt rules to curb redundant file re-reads
• Anti-confabulation: dedup tool names, prune aliases, harden read guard
• Summarize comments across all major Swift files
• Wrap long lines with SwiftFormat across the project

📦 Dependencies

• AgentTools 2.50.10
• AgentAccess 2.9.5 (screenshots no longer block main thread)

---

DMG to follow.

🦾 Agent! 1.0.68 (Build 156)

✨ 🦾 Agent! What's New in 1.0.68?

🔧 diff_apply Reliability

• Fixed truncation bug: source is now spliced into the file instead of replacing entire content
• Fixed LLMs sending diff-formatted input to diff_apply
• Replaced force-unwrap with safe if-let in range note parsing
• Updated AgentTools to 2.50.7 with truncation fix
• Bumped prompt revision to 86 for improved diff_apply prompting

🧠 Apple Intelligence Fixes

• Fixed non-agent tool names being misrouted to run_agent
• Tightened run_agent trigger to exact phrase match only
• Fixed duplicate log entries for accessibility tasks

🖥️ UI Improvements

• Paste chips: Long Cmd+V pastes now appear as removable preview chips instead of flooding the text field
• Tab labels: Main tab shows current model name; LLM tabs preserve raw model casing
• Per-tab usage: Added individual usage breakdown to each LLM tab
• Log formatting: Multi-line messages now drop below the timestamp on their own line

🧹 Code Quality

• Consolidated vision model detection into shared isVisionModel() (removed two duplicate private methods)
• Summarized verbose comments across 10 source files
• Skipped suggestion matching for long inputs to prevent main-thread stalls
• Fixed tool output formatting (newline prefix and :0:0 warning cleanup)
• Added iteration cap enforcement

🚀 Agent! v1.0.65 (Build 153)

What's Changed

Version Management

• Bump version fixes — bump_version/get_version now correctly uses the selected project instead of Xcode open documents
• Version bumps: 1.0.64 (152) → 1.0.65 (153)

Security & Safety

• action_not_performed guardrail — three-layer defense against false-action claims (app layer, prompt layer, documentation)
• README + SECURITY.md updated to document the action_not_performed defense layers

Apple Intelligence Hardening

• Verify Apple AI tool outputs before claiming success — forwards to LLM on failure
• Log Apple AI tool actions to activity log for transparency
• Harden Apple AI triage — never claim completion on failure or uncertainty

Documentation

• README: Tool Execution Gating added to Defense Layers section
• README: Tip Jar section added

Full Changelog: https://github.com/toddbruss/Agent/compare/1.0.61.149...1.0.65.153

🚀 Agent! v1.0.61 (Build 149)

🦾 Agent! Agentic AI for your  Mac Desktop (or Laptop)

Agent! Requires macOS 26.4 or later. M series preferred for Apple Intelligence. We don't know if Apple AI works on Intel. Also Apple AI is not supported inside a Mac VM. It's optional, but does help move things along by doing several tasks locally, saving you some tokens.

🔧 Bug Fixes

• Fix double echo — skip appendLog for shell commands that already streamed output
• Remove tool result cache — eliminate read loop blocker that caused stale responses
• Remove file read dedup cache — LLMs work around stub responses anyway; cleaner tool flow

🧠 Apple AI Improvements

• Skip accessibility triage for file-path and coding prompts — faster task routing
• Add ! bypass prefix — force Apple AI to skip triage when you know what you want
• Add AppleScript + shell tools alongside accessibility for Apple AI agent

🖥️ UI & UX

• Propagate parent tab project folder to spawned script tabs — scripts inherit the right working directory
• Add provider tooltip to LLM Usage popup model rows — see which provider serves each model

📖 Documentation

• Add Defense Layers section to README covering security features
• Add build-without-developer-account instructions + build.sh script
• Add emojis to toolbar icon table, remove SF Symbol text names
• Fix GLM model sizes — 744B/754B MoE, not 32B
• Clarify GLM-5/5.1 are cloud-only (may be possible to run FP8 locally); GLM-4.7-Turbo (32B) runs locally

⬆️ Upgrade Notes

• If upgrading from v1.0.60, no migration needed — settings and tabs are preserved.

🚀 Agent! 1.0.59 (Build 147)

🚀 Agent! 1.0.59 (Build 147)

LLM / Provider Improvements

• 429 Fallback Chain Overhaul: Fallback now triggers after 2 failures (down from 3) with a flat 10s retry, so the chain fires within ~30s instead of retrying forever
• 429 handling: Every 429 failure is now recorded against the fallback chain, not just after 3 consecutive failures
• 429 debugging: Log the actual API response body on 429 errors and add clearRetryAfter helper
• BigModel: Same dual-API URLs as Z.ai — code and vision endpoints registered separately
• Z.ai: Proper dual-API URLs for code and vision endpoints
• Apple AI triage: Don't intercept shell commands (open ., open /path, ls, git, etc.) — let them pass through to the shell handler
• Per-tab provider: Sync Settings picker to the active tab's LLMConfig so each tab can use a different provider
• Model refresh consolidation: Route all model-refresh calls through a single fetchModelsIfNeeded(for:force:), replacing 3 separate functions
• FallbackChainView: Show model display name (with -Code suffix for Z.ai)

Tool / Infrastructure

• batch: Fix duplicate output and guard empty git branch name
• batch_commands: Remove per-step delimiters, run as a single script for cleaner output
• Remove read guard snap-out: LLMs need unlimited reads for project research

Version

• Version bump to 1.0.59 (Build 147)

Agent! 1.0.41 (Build 126)

Agent! 1.0.41

🛠 Tool System

• Consolidated tool naming — all tools renamed with _tool suffix, auto-stripped after first turn
• mkdir and cd actions added to file_manager
• Condensed tools — strip _tool suffix after first turn, auto-revert on failure
• AgentTools 2.31.0 — synced all tool action lists with new naming
• Run agent — support both run agent X and agent run X for direct execution
• Fix run_agent — compile then dlopen/run dylib instead of just compiling

📺 LLM Output Overhaul

• NSTextTable inline rendering — tables render inline with proper padding and sizing
• AgentTerminalNeo integration — native auto-scroll, smooth streaming, cursor blink
• Auto-grow output — LLM Output grows with content, capped at 50% window height
• Draggable resize handle — resize LLM Output panel with drag, respects max height
• Live token counts — estimate input/output tokens during streaming
• Always-blinking cursor — seamless streaming to idle, AGENT! prompt blinks too
• Line-count-based height — eliminates NSTextTable overestimation and jitter
• Fix streaming jitter — height only grows, never shrinks

🧠 Fallback Chain

• Fallback Chain UI — configure backup models in toolbar, auto-switch on failure
• Model picker dropdowns — shows fetched models from all providers
• Eye icon for vision models in pickers and fallback entries
• Wire fallback chain into error handler — auto-switch on provider failure

🎨 UI Improvements

• Status icons and colors — 🧠/blue Thinking, ▶/green Running, 🖥/orange Executing, 🛡/red Root
• Coding Preferences icon turns green when any feature is on
• Agents menu retry — longer initial delay, 10 insertion attempts
• Window frame persistence — retains size after tiling
• Fix double overlay — tab selected shows only its own indicator
• Cmd+B toggles overlay on all tabs including running ones

🔧 Fixes

• batch_commands — run all in single shell session so variables persist
• Fix token accuracy — live output estimate, no double-counting
• Fix timer — 0.25s refresh for live tokens, freeze on cancel
• Fix elapsed time — freeze at actual value when done
• Fix duplicate recordFailure — was double-counting failures
• Fix drag jitter — global coordinate space for handle gesture
• Fix window resizing on drag — prevent window resize from LLM Output handle

🤖 Z.ai / Model Providers

• Dynamic model fetching — Z.ai models 100% from OpenAPI spec, zero hardcoding
• Two endpoints — coding (no vision) and general (vision), URL routing with :v tag
• Fetch models for ALL providers — Z.ai, Gemini, Grok, Mistral, etc.
• Coding endpoint models tagged with -Code suffix

🏗 Architecture

• Xcode-only gating — block xc tools when no .xcodeproj detected
• Verify gate — build must pass before task_complete allowed
• Remove Claude Code references from source comments
• AgentTools pinned to 2.31.0, prompt revision 62