feat: unify agent identities and refine agent and skill management

[GCWing]

Summary

Unify the four built-in Agent Harness identities as Minimal, Standard, Ultimate, and Creative, while keeping legacy configuration, persisted sessions, and peer/mobile/dispatch wire identities readable. Refine Agent and assistant configuration, reusable skill groups, and session panel ownership across the desktop and mobile surfaces.

Type and Areas

Feature, refactor, and UI/UX. Shared Rust contracts/runtime, desktop and CLI adapters, Web UI, mobile web, design system, and localization.

Motivation / Impact

• Generate Rust and TypeScript Harness identities from one contract, preserve unknown identities and configuration fields, and reject conflicting legacy aliases without overwriting records.
• Write canonical Swarm root identities and normalize legacy database rows at the read boundary, so upgraded trees can keep launching child Agents without rewriting existing records.
• Keep the SSH dispatch adapter independently buildable: its remote-ssh-concrete feature explicitly activates the existing core-types dependency used for identity wire translation. No dependency version or default feature changes are required.
• Manage reusable skill groups in the Skills scene and expand selections through the shared group store when configuring an Agent. Keep unavailable members and partial catalog results visible.
• Consolidate Agent/assistant configuration presentation, simplify navigation and motion, and align terminal/content panel ownership with the active session.
• Integrate the current main changes to account-device routing, removal of cloud settings sync, and retirement of first-entry mode reminders.

Verification

Passed locally:

• pnpm run fmt:rs
• cargo test --locked -p openbitfun-core-types
• cargo test --locked -p openbitfun-config-contracts --lib — 77 tests passed.
• cargo test --locked -p openbitfun-core --no-default-features --lib service::config:: — 63 tests passed.
• cargo test --locked -p openbitfun-core --no-default-features --features agent-runtime,git,external-sources --lib agentic::coordination:: — 173 tests passed, including legacy Swarm database and external Agent ownership regressions.
• cargo test --locked -p openbitfun-services-integrations --no-default-features --features remote-ssh-concrete --lib remote_ssh::manager::tests::workspace_ — 24 tests passed; 2 existing opt-in diagnostics skipped.
• pnpm run check:web — type-check, Appearance, typography, theme color, and visual-contract gates passed.
• pnpm run design-system:test
• pnpm --dir src/mobile-web run test:ui-components — 10 tests passed.
• pnpm --dir src/mobile-web run type-check
• pnpm run i18n:generate, pnpm run i18n:contract:test (37 passed), and pnpm run i18n:audit (zero warnings).
• pnpm run check:core-boundaries:test — 142 tests passed.
• pnpm run check:core-boundaries
• pnpm run check:repo-hygiene
• pnpm run motion:audit — intent inventory reviewed; not visual acceptance.
• node scripts/check-git-object-sizes.mjs --base origin/main --head HEAD
• git diff origin/main...HEAD --check

The focused frontend command below passed all 92 tests across 8 files:

pnpm --dir src/web-ui run test:run src/features/skill-groups/skillGroups.test.ts src/features/skill-groups/skillGroupsStore.test.ts src/shared/agents/identity.test.ts src/app/components/panels/content-canvas/hooks/canvasPanelOwnership.test.tsx src/app/services/sessionSceneLifecycle.test.ts src/flow_chat/services/sessionActivation.test.ts src/flow_chat/services/storeSync.test.ts src/app/stores/sceneStore.test.ts

The CI regression command below also passed all 62 tests across 5 files. The text-clipping test additionally checks the compact line-height scale against the bundled font metrics.

pnpm --dir src/web-ui run test:run src/app/global-search/globalSearchArchitecture.test.ts src/app/styles/textClipping.test.ts src/flow_chat/components/ChatInputWorkspaceStripLayout.test.ts src/app/components/panels/content-canvas/hooks/canvasPanelOwnership.test.tsx src/shared/agents/identity.test.ts

Full automated validation is provided by the PR CI; this PR will be merged only after all applicable checks succeed on its final commit.

Reviewer Notes

AI-assisted. Testing level: focused automated tests; full CI pending. No browser automation, mock visual validation, or manual visual acceptance was performed. Compatibility fixtures cover legacy identity round trips; live SSH workspaces, remote control, Peer Device Mode, and detached dispatch were not exercised manually.

Checklist

• This PR contains the requested product changes and no temporary artifacts or credentials.
• Verification and its current limits are recorded above.
• User-facing strings, docs, and locales are updated where applicable.