Skip to content

Agent Smith 1.0.1

Choose a tag to compare

@drewster99 drewster99 released this 01 Aug 06:05
· 399 commits to main since this release

Release 1.0.1 of Agent Smith.

Install

  1. Download the DMG asset below.
  2. Open it and drag Agent Smith to Applications.
  3. Launch the app from Applications.

Requires macOS 26.2 or later. The app is Developer ID signed and notarized by Apple.

Changes

  • Release 1.0.1 (27cd8b0)
  • App: bump SwiftLLMKit to 0.0.136 (123bfbd)
  • Capability eval: default to probing every catalogued model (908b58c)
  • Per-agent picker: surface pricing, deprecation, and effective runtime settings (d62e4f0)
  • Trailing-system probe: realistic shape + run it in the GUI probe too (7004199)
  • Per-agent model picker: single source of truth via availableModels(role) (92a0335)
  • Models UI: flag extras + deprecated/free/size chips; probe trailing-system for all providers (e1c6553)
  • Pricing sheet: pin advanced number fields' locale explicitly, not via environment (5b973d2)
  • Capabilities sheet: make Display name read-only for now (c3da2a6)
  • Model override sheets: pin all number formatting to a fixed US period-decimal locale (be4fe43)
  • Orchestration settings: in-window toolbar button for per-session overrides (3fbde8a)
  • Model override sheets: /stupid audit fixes (2ad47d0)
  • CLAUDE.md: document Orchestration settings + record the security-review reversal (4ea0eda)
  • Orchestration OFF behaviors: the 3 new retrieval points (256c43c)
  • Orchestration OFF behaviors: per-emitter tool-call review (b62e4a0)
  • Models tab: two-level sort + disable sort control under a provider filter (775e58a)
  • Pricing sheet: clear the $0 warning once a base rate is overridden (07425dd)
  • Orchestration OFF behaviors: retire preflightScoping for scopeToolSetOnTaskStart (9e40932)
  • Model override sheets: fix codex-review findings (8be5f87)
  • Models tab: harden the build/filter pipeline (codex review follow-up) (b650a3e)
  • Orchestration OFF behaviors: validation skip (f57c936)
  • Orchestration OFF behaviors: summarizer + retrieval (35126b2)
  • Orchestration settings: push resolved snapshot into the runtime (7fa6b54)
  • Model sheets: Default/Override widget for numeric fields, effort levels, drop popup (f210715)
  • Orchestration settings: per-session overrides sheet + Session menu (7f346f1)
  • Models tab: multi-term search filter with match highlighting (006bbe1)
  • Orchestration settings: app-wide Settings tab (98c2685)
  • Orchestration settings: per-session override + resolver (1f0986b)
  • Orchestration settings: layered defaults storage (1e07267)
  • Add OrchestrationSettings layered-config type family + resolver (6e9dedb)
  • Model override sheets: native grouped Form, shared chrome, full pricing, per-model probe (4a295fe)
  • Models tab: render the list with LazyVStack; drop progressive reveal (f9f48bb)
  • Fix Models settings tab jank: extract, cache the list, drop per-row catalog scans (b0977a7)
  • Convert RoleModelConfigOverrideEditor's some-View properties to functions (8581395)
  • Guard assignment prune against an empty provider list (agy review) (9ecc652)
  • Prune dead OnboardingRole props left by the config-pool retirement (eef02fd)
  • Behavior-flags editor is enum-driven (all 8 flags), on SwiftLLMKit 0.0.131 (4a50888)
  • Update bundled-defaults test: assignments resolve by (provider, model) (c3f3bb3)
  • Document the config-pool retirement in CLAUDE.md (8d4a136)
  • Replace the Configurations settings tab with a per-model "Models" tab (a2c0841)
  • Retire the config pool: role assignments hold (provider, model) directly (7ec70ae)
  • Build role providers from the resolved override config, not the pool clamp (29aa12f)
  • Per-(role, model) runtime overrides: store, resolve-fresh wiring, and editor UI (0546d8d)
  • Recheck fixes: restore the chat field's inspector label; correct editor doc (6dff67f)
  • Pin swift-llm-kit 0.0.129; apply Fable 5 review fixes (100a57f)
  • Pin swift-llm-kit 0.0.128; exclude batch-only models from every role (b2a5394)
  • Pin swift-llm-kit 0.0.127; enum-drive the capabilities editor (5d99a9b)
  • Pin swift-llm-kit 0.0.126; harden role picker filter (codex review) (613313b)
  • Filter the role model picker by AgentRole.modelRequirements (99ce7dc)
  • Pin swift-llm-kit 0.0.125; finalize AgentRole model requirements (chat + tri-state) (12daf79)
  • Pin swift-llm-kit 0.0.124 (tri-state ModelCapabilities) (0363a04)
  • Pin swift-llm-kit 0.0.123; add AgentRole model requirements (2374833)
  • SwiftLLMKit 0.0.122, and pin defaults.json against silent decode failure (3060510)
  • SwiftLLMKit 0.0.121: OpenRouter :floor and :nitro reach the picker and the prober (85e8f15)