agent-orchestrator/packages/core
Ashish Huddar af2af115bd
style(design): design review fixes + fresh Warm Terminal design system (#927)
* style(design): FINDING-001 — add prefers-reduced-motion support

All animations and transitions are disabled when the user's system
requests reduced motion, per DESIGN.md accessibility requirements.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style(design): FINDING-003 — remove concurrent breathe animations

Status pills had two animations: a breathe animation on the pill
and a dot-pulse on the child dot. DESIGN.md says "one animation per
element, one purpose" and "keep dot pulse, remove border heartbeat."

Removed all three breathe keyframes, kept dot-pulse only.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style(design): FINDING-004 — fix dashboard title weight and tracking

DESIGN.md specifies display headings at weight 680 and letter-spacing
-0.035em. The dashboard title was using 600 / -0.05em.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style(design): FINDING-005 — fix detail-card text to blue-tinted graphite

Detail cards overrode text-secondary and text-tertiary with neutral
grays (#9898a0, #5c5c66). DESIGN.md specifies blue-tinted graphite
palette (#a5afc4, #6f7c94) for dark mode text.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style(design): FINDING-008 — add text-wrap: balance on headings

Dashboard title and kanban column titles now use text-wrap: balance
for more even line breaks on narrow viewports.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style(design): FINDING-006/009 — fix section label semantics and spacing

Changed "Attention Board" from <h2> to <div role="heading"> since it's
styled as a 12px uppercase label, not a heading. Also fixed letter-spacing
from 0.16em to 0.06em per DESIGN.md UI/Labels spec.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* style(design): FINDING-007 — contextual empty state messages

Empty kanban columns now show context-specific messages instead of
generic "No sessions" text. Each column's empty state reflects its
purpose: "No agents need your input" (Respond), "No code waiting
for review" (Review), etc.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* docs(design): fresh design system — Warm Terminal

Complete redesign from Industrial Precision (blue-tinted) to Warm Terminal
(brown-tinted). Key changes:

- Warm charcoal surfaces (#121110, #1a1918, #222120) replace blue-gray
- Cream text (#f0ece8) replaces blue-white (#eef3ff)
- Warm periwinkle accent (#8b9cf7) replaces cool blue (#5B7EF8)
- Berkeley Mono for display headlines (mono cohesion)
- Added: Accessibility section (44px touch targets, WCAG AA, focus-visible)
- Added: Component anatomy (button states, card structure, input fields)
- Added: Light mode design rationale (warm parchment, not clinical white)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* docs(design): swap Berkeley Mono for JetBrains Mono (free)

Berkeley Mono is a paid font ($75). JetBrains Mono is free, open source,
already loaded in the project, and the mono-for-headlines concept works
the same way with it.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* docs(design): fix light mode contrast failures

Light mode text-tertiary #a8a29e failed WCAG AA at 2.5:1 on white.
Darkened to #736e6b (5.0:1). Light mode accent #6b73c4 was borderline
at 4.3:1, darkened to #5c64b5 (5.3:1). All pairs now pass AA.

Added verified contrast ratios for both modes to accessibility section.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* chore: gitignore .gstack/ directory

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* docs(design): add design audit report and screenshots

Design review audit report with before/after screenshots for all
dashboard pages (kanban, session detail, PRs) across desktop, tablet,
and mobile viewports in both light and dark mode.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix(design): address PR review comments

- Fix mobile test expecting removed "No sessions" text. The merge zone
  emptyMessage is now "Nothing cleared to land yet." (Bugbot comment #1)
- Remove no-op .dark .detail-card override that duplicated global dark
  values after FINDING-005 fix aligned them (Bugbot comment #2)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: gitignore .gstack-report/ and remove from tracking

The .gstack-report/ directory contains local audit artifacts with
filesystem paths. Should not be tracked in the repository.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix(design): align dashboard title CSS to new DESIGN.md spec

Dashboard title was using old Geist Sans values (weight 680, -0.035em).
New spec is JetBrains Mono, weight 500, letter-spacing -0.02em.
Added font-family: var(--font-mono) to match.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: use native h2 element for Attention Board section heading

Replace ARIA role="heading" div with semantic h2 per ARIA first rule — native elements are preferred over ARIA roles for actual headings.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix: pre-landing review fixes — a11y, dead code, test coverage

- Add aria-controls + id to accordion button/body pair in AttentionZone
- Wrap empty-state messages in aria-live="polite" regions for AT announcements
- Remove dead message prop and isDefault from EmptyState (Skeleton.tsx)
- Add parameterized test covering all 6 zone-specific empty messages

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix: revert aria-live on empty states — causes false AT announcements

Codex review identified that role="status" aria-live on static empty-state
text causes burst announcements on page load (all empty columns fire) and
announces in collapsed mobile sections that aren't visible. Empty states are
static text, not dynamic transitions. The aria-controls fix is kept.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* chore: bump version and changelog (v0.0.1.0)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* chore: remove .gstack-report/ from .gitignore

* chore: remove VERSION and CHANGELOG (not used in this project)

* style(design): warm terminal color migration + inline style removal

Migrate all CSS tokens from cool blue-tinted graphite to warm
brown-tinted terminal aesthetic per DESIGN.md spec. Replace inline
style color mappings in ActivityDot, AttentionZone, Dashboard,
ProjectSidebar, and SessionCard with data-attribute CSS selectors.
Fix duplicate className bug on SessionCard done-title element.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: pre-landing review fixes — activity dot fallback + review stat color

Add base CSS fallback for activity-dot, activity-pill, and
activity-pill__text so null/unknown activity states render visibly
(gray) instead of invisible. Fix review stat card to use accent-orange
(matching kanban/sidebar/mobile review indicators) instead of cyan.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* chore: checkpoint current design branch state

* design changes

* feat(web): redesign session detail page — compact PR card, identity strip, layout reorder

- Redesign SessionTopStrip with simplified breadcrumbs, action buttons (Message/Kill)
- Replace stacked PR card with compact inline layout: title row + blocker/CI chips + collapsible comments
- Move PR card above terminal for better information hierarchy
- Replace vertical IssuesList with inline buildBlockerChips helper
- Add ~200 lines of new CSS classes for compact PR card design system
- Add changedFiles field to DashboardPR type

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat(web): design system tokens, sidebar redesign, and component primitives

- Align all color tokens (status, bg, border, text) across three HTML mockups
- Rewrite ProjectSidebar to match finalized.html: rotation chevron, session status text, border-bottom project separators, 224px width
- Add packages/web/DESIGN.md: agent-readable reference for tokens, typography, component patterns, anti-patterns
- Add Badge.tsx: generic badge/chip/pill primitive with status/outline/default variants
- Add Button.tsx: ghost/primary/danger button primitive
- Update CLAUDE.md to reference DESIGN.md as required pre-read for web UI work

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* style(design): FINDING-001 — card border-radius 0 → 6px to match mockup

* style(design): FINDING-002 — column border-radius 0 → 7px, border subtle to match mockup

* style(design): FINDING-003 — column header mono font, 500 weight, muted color to match mockup

* style(design): FINDING-004/005/006 — fix accent-blue/yellow/purple tokens to match mockup

* style(design): FINDING-007 — add --color-bg-card token (light #fff, dark #1c1b19)

* Refine dashboard design system and remove fixture flow

* Fix respond status colors in dashboard indicators

* Align tests with updated dashboard and metadata behavior

* fix(core): register notifier aliases consistently

* chore(web): drop uncovered showcase routes

* Add desktop PullRequestsPage coverage tests

* Remove generated coverage artifact

* Consolidate web design guidance into the root design system

* Fix working and ready status color tokens

* Fix sidebar collapse and inline kill confirmation

* fix(qa): ISSUE-001 - show all mobile filter chips

* Restore full title contrast in session cards

* Fix review feedback in dashboard state styling

* style: implement mobile responsive designs (feed, terminal-first, dense PRs)

Dashboard: replace accordion with urgency-sorted priority feed, horizontal scroll filter pills.
Session Detail: terminal-first layout with floating header, status pill, PR bottom sheet.
PRs: dense rows with CI dots, grouped sections, muted merged/closed rows.
Update tests to match new mobile layouts.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Fix mobile terminal padding with PR sheet layout

* Polish mobile feed and session detail styling

* Align mobile dashboard layouts with gstack designs

* Fix mobile terminal actions and PR review labels

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-11 16:30:03 +05:30
..
__tests__ fix: added GraphQL batch PRfeat: add GraphQL batch PR enrichment for orchestrator polling enrichment for orchestrator polling (fixes #608) (#637) 2026-03-30 00:31:04 +05:30
src style(design): design review fixes + fresh Warm Terminal design system (#927) 2026-04-11 16:30:03 +05:30
CHANGELOG.md chore: version packages 2026-03-20 15:47:55 +00:00
README.md fix: merge upstream main and resolve conflicts for cursor agent 2026-04-10 11:48:27 +05:30
package.json style(design): design review fixes + fresh Warm Terminal design system (#927) 2026-04-11 16:30:03 +05:30
tsconfig.build.json feat: implement SCM and tracker plugins (github, linear) (#4) 2026-02-14 15:45:51 +05:30
tsconfig.json feat: implement SCM and tracker plugins (github, linear) (#4) 2026-02-14 15:45:51 +05:30
vitest.config.ts fix: merge upstream main and resolve conflicts for cursor agent 2026-04-10 11:48:27 +05:30

README.md

@aoagents/ao-core

Core services, types, and configuration for the Agent Orchestrator system.

What's Here

  • src/types.ts — All TypeScript interfaces (Runtime, Agent, Workspace, Tracker, SCM, Notifier, Terminal, Session, events)
  • src/services/ — Core services (SessionManager, LifecycleManager, PluginRegistry)
  • src/config.ts — Configuration loading + Zod schemas
  • src/utils/ — Shared utilities (shell escaping, metadata parsing, etc.)

Key Files

src/types.ts — The Source of Truth

Every interface the system uses is defined here. If you're working on any part of the orchestrator, start by reading this file.

Main interfaces:

  • Runtime — where sessions execute (tmux, docker, k8s)
  • Agent — AI coding tool adapter (claude-code, codex, aider)
  • Workspace — code isolation (worktree, clone)
  • Tracker — issue tracking (GitHub Issues, Linear)
  • SCM — PR/CI/reviews (GitHub, GitLab)
  • Notifier — push notifications (desktop, Slack, webhook)
  • Terminal — human interaction UI (iTerm2, web)
  • Session — running agent instance (state, metadata, handles)
  • OrchestratorEvent — events emitted by lifecycle manager
  • PluginModule — what every plugin exports

src/services/session-manager.ts — Session CRUD

Handles session lifecycle:

  • spawn(config) — create new session (workspace + runtime + agent)
  • list(projectId?) — list all sessions
  • get(sessionId) — get session details
  • kill(sessionId) — terminate session
  • cleanup(projectId?) — kill completed/merged sessions
  • send(sessionId, message) — send message to agent

Data flow in spawn():

  1. Load project config
  2. Validate issue exists via Tracker.getIssue() (if issueId provided, fails-fast if not found)
  3. Reserve session ID
  4. Determine branch name
  5. Create workspace via Workspace.create()
  6. Generate prompt via Tracker.generatePrompt()
  7. Build launch command via Agent.getLaunchCommand()
  8. Create runtime session via Runtime.create()
  9. Run Agent.postLaunchSetup() (optional)
  10. Write metadata file
  11. Return Session object

Note: If issue validation fails (not found, auth error), spawn fails before creating any resources (no workspace, no runtime, no session ID). This prevents spawning sessions with broken issue references.

src/services/lifecycle-manager.ts — State Machine + Reactions

Polls sessions, detects state changes, triggers reactions:

State machine:

spawning → working → pr_open → ci_failed/review_pending/approved → mergeable → merged

Reactions:

  • ci-failed → send fix prompt to agent
  • changes-requested → send review comments to agent
  • approved-and-green → notify human (or auto-merge)
  • agent-stuck → notify human

Polling loop:

  1. For each session: check agent activity state (Agent.getActivityState())
  2. If PR exists: check CI status (SCM.getCISummary()), review state (SCM.getReviewDecision())
  3. Update session status based on state
  4. Trigger reactions if state changed
  5. Emit events

src/services/plugin-registry.ts — Plugin Discovery + Loading

Loads plugins and provides access to them:

  • register(plugin, config?) — register a plugin instance
  • get<T>(slot, name) — get plugin by slot + name
  • list(slot) — list all plugins for a slot
  • loadBuiltins(config?) — load built-in plugins (runtime-tmux, agent-claude-code, etc.)
  • loadFromConfig(config) — load built-ins today; external plugin descriptors are the marketplace extension point

Built-in plugins (loaded by default):

  • runtime-tmux, runtime-process
  • agent-claude-code, agent-codex, agent-aider, agent-opencode
  • workspace-worktree, workspace-clone
  • tracker-github, tracker-linear, tracker-gitlab
  • scm-github, scm-gitlab
  • notifier-desktop, notifier-discord, notifier-slack, notifier-composio, notifier-openclaw, notifier-webhook
  • terminal-iterm2, terminal-web

src/config.ts — Configuration Loading

Loads and validates agent-orchestrator.yaml:

Main config sections:

  • Runtime data paths are auto-derived from the config location under ~/.agent-orchestrator/{hash}-{projectId}/
  • port — web dashboard port (default 3000, set different values for multiple projects)
  • terminalPort — terminal WebSocket port (auto-detected if not set)
  • directTerminalPort — direct terminal WebSocket port (auto-detected if not set)
  • defaults — default plugins (runtime, agent, workspace, notifiers)
  • plugins — installer-managed external plugin descriptors (registry, npm, or local)
  • projects — per-project config (repo, path, branch, symlinks, reactions, agentRules)
  • notifiers — notification channel config (Slack webhooks, etc.)
  • notificationRouting — which notifiers get which priority events
  • reactions — auto-response config (ci-failed, changes-requested, approved-and-green, etc.)

Zod schemas validate all config at load time.

Common Tasks

Adding a Field to Session

  1. Edit src/types.tsSession interface
  2. Edit src/services/session-manager.ts → initialize field in spawn()
  3. Rebuild: pnpm --filter @aoagents/ao-core build

Adding an Event Type

  1. Edit src/types.tsEventType union
  2. Emit the event: eventEmitter.emit() in relevant service
  3. Add reaction handler (optional): src/services/lifecycle-manager.ts

Adding a Reaction

  1. Edit src/services/lifecycle-manager.ts → add handler function
  2. Wire it up in the polling loop
  3. Add config schema in src/config.ts if new reaction type

Feedback Tools (v1)

@aoagents/ao-core exports two structured feedback tool contracts:

  • bug_report
  • improvement_suggestion

Both share the same required input fields:

  • title
  • body
  • evidence (array of strings)
  • session
  • source
  • confidence (0..1)

Example:

import { FEEDBACK_TOOL_NAMES, FeedbackReportStore, getFeedbackReportsDir } from "@aoagents/ao-core";

const reportsDir = getFeedbackReportsDir(configPath, projectPath);
const store = new FeedbackReportStore(reportsDir);

const saved = store.persist(FEEDBACK_TOOL_NAMES.BUG_REPORT, {
  title: "SSO login loop",
  body: "Google SSO redirects back to /login repeatedly.",
  evidence: ["trace_id=abc123", "screenshot: login-loop.png"],
  session: "ao-22",
  source: "agent",
  confidence: 0.84,
});

Storage format:

  • Reports are persisted under ~/.agent-orchestrator/{hash}-{projectId}/feedback-reports
  • Each report is a typed key=value file (report_<timestamp>_<id>.kv) for easy inspection
  • A deterministic dedupe key (sha256, 16 hex chars) is generated from normalized tool+content

Migration notes:

  • No migration needed for existing AO installs
  • The feedback-reports directory is created lazily on first persisted report

Testing

# Run all core tests
pnpm --filter @aoagents/ao-core test

# Run in watch mode
pnpm --filter @aoagents/ao-core test -- --watch

# Run specific test
pnpm --filter @aoagents/ao-core test -- session-manager.test.ts

Tests are in src/__tests__/:

  • session-manager.test.ts — session CRUD, spawn, cleanup
  • lifecycle-manager.test.ts — state machine, reactions
  • plugin-registry.test.ts — plugin loading, resolution
  • tmux.test.ts — tmux utility functions (not a plugin test)
  • prompt-builder.test.ts — prompt generation utilities

Building

# Build core
pnpm --filter @aoagents/ao-core build

# Typecheck
pnpm --filter @aoagents/ao-core typecheck

This package is a dependency of all other packages. Build it first if working on the codebase.

Architecture Notes

Why flat metadata files?

  • Debuggability: cat ~/.agent-orchestrator/<hash>-my-app/sessions/app-3 shows full state
  • No database dependency (survives crashes, easy to inspect)
  • Backwards-compatible with bash script orchestrator

Why polling instead of webhooks?

  • Simpler (no webhook setup, no ngrok for local dev)
  • Works offline (CI/review state is fetched, not pushed)
  • Survives orchestrator restarts (no missed events)

Why plugin slots?

  • Swappability: use tmux locally, docker in CI, k8s in prod
  • Testability: mock plugins for tests
  • Extensibility: users can add custom plugins (e.g., company-specific notifier)