tycono 0.1.73 → 0.1.74-beta.1

package/README.md

@@ -1,11 +1,12 @@
 <p align="center">
-  <img src=".github/assets/hero-office.png" alt="Tycono — AI Office" width="640" />
+  <img src=".github/assets/hero-office.png" alt="Tycono — AI Office" width="720" />
 </p>
 
 <h1 align="center">tycono</h1>
 
 <p align="center">
-  <strong>Build an AI company. Watch them work.</strong>
+  <strong>Build an AI company. Watch them work.</strong><br>
+  <sub>Infrastructure-as-Code defined servers. Company-as-Code defines organizations.</sub>
 </p>
 
 <p align="center">
@@ -15,94 +16,168 @@
 </p>
 
 <p align="center">
-  <a href="https://tycono.ai">Website</a> &middot;
-  <a href="#quick-start">Quick Start</a> &middot;
-  <a href="#how-it-works">How It Works</a> &middot;
+  <a href="https://tycono.ai">Website</a> ·
+  <a href="#quick-start">Quick Start</a> ·
+  <a href="#how-it-works">How It Works</a> ·
+  <a href="#company-as-code">Company-as-Code</a> ·
   <a href="CONTRIBUTING.md">Contributing</a>
 </p>
 
 ---
 
-**tycono** is an open-source platform that lets you create and run an AI-powered organization. Define roles (CTO, PM, Engineer...), assign them AI agents, and watch them collaborate through a real-time dashboard.
+**tycono** is an open-source platform that lets you define and run an AI-powered organization. Roles, authority, knowledge, and workflows — all defined in files, executed by AI agents, visualized in real time.
 
-## Quick Start
+One command. Your AI company is running.
 
 ```bash
-mkdir my-company && cd my-company
 npx tycono
 ```
 
-That's it. A setup wizard guides you through creating your company, then your browser opens to a live dashboard showing your AI team at work.
-
 ## Why Tycono?
 
+Coding agents simulate **one developer**. Tycono simulates **the entire company**.
+
 | | Single AI Agent | Tycono |
 |---|---|---|
-| **Structure** | One agent, one context | Multiple roles with org hierarchy |
-| **Knowledge** | Loses context between sessions | Persistent, file-based knowledge system (AKB) |
-| **Authority** | Can do anything | Scoped — each role has boundaries |
-| **Delegation** | Manual prompt chaining | Automatic dispatch through org chart |
-| **Visibility** | Terminal output | Real-time isometric office dashboard |
+| **What it runs** | One agent, one context | Multiple roles with org hierarchy |
+| **Knowledge** | Resets every session | Compounds forever — file-based, cross-linked |
+| **Authority** | Can do anything (or nothing) | Scoped — each role has clear boundaries |
+| **Delegation** | Manual prompt chaining | CEO dispatches, org chart routes automatically |
+| **Scale** | 1 agent | 7 → 700 agents |
+| **Visibility** | Terminal output | Isometric office + Slack-style Pro dashboard |
+
+## Company-as-Code
+
+Just as Terraform turns `.tf` files into running infrastructure, Tycono turns YAML and Markdown into a running company.
+
+```
+IaC                          CaC (Company-as-Code)
+─────────────────────        ─────────────────────
+.tf         → servers        role.yaml   → org structure
+playbook    → config         CLAUDE.md   → operating rules
+Dockerfile  → containers     skills/     → capabilities
+state file  → infra state    knowledge/  → org memory
+```
+
+Your company is **versionable**, **reproducible**, and **forkable** — just like code.
+
+## Quick Start
+
+```bash
+mkdir my-company && cd my-company
+npx tycono
+```
+
+A setup wizard guides you through:
+
+1. **Pick an AI engine** — Claude API, Claude Max, or auto-detect
+2. **Name your company** — set mission and domain
+3. **Choose a team template** — or build from scratch
+4. **Watch them work** — your browser opens to a live dashboard
+
+### Requirements
+
+- Node.js >= 18
+- [Anthropic API key](https://console.anthropic.com/) or Claude Max subscription
+
+## Two Ways to Work
 
-## What You Get
+### Office View — Watch your AI team
 
-- **Role-based AI agents** — Each role has its own persona, authority scope, and knowledge boundaries
-- **Org hierarchy** — Roles report to each other. CTO dispatches to Engineers. PM coordinates with Design.
-- **Real-time dashboard** — Watch your AI team work in an isometric pixel-art office
-- **Knowledge management** — Automatic document routing, cross-linking, and Hub-based organization
-- **Local-first** — Everything runs on your machine. Your data stays yours.
-- **BYOK** — Bring your own Anthropic API key. No middleman.
+An isometric pixel-art office where your AI agents sit at their desks, work, chat, and think. Click any agent to talk to them directly.
 
 <p align="center">
-  <img src=".github/assets/sidepanel-chat.png" alt="Tycono Dashboard" width="640" />
+  <img src=".github/assets/hero-office.png" alt="Office View" width="640" />
 </p>
 
-## Requirements
+- Pixel-art characters with personalities and levels
+- Ambient speech bubbles — agents think out loud
+- Rooms: Leadership, Engineering, Meeting, Knowledge Library
+- Edit mode — rearrange furniture, customize your office
 
-- Node.js >= 18
-- [Anthropic API key](https://console.anthropic.com/)
+### Pro View — Manage your AI company
 
-## Team Templates
+A Slack-style professional dashboard for serious work. Chats, Wave dispatch, Decisions log, Knowledge graph.
 
-During setup, pick a template or build your own:
+<p align="center">
+  <img src=".github/assets/pro-view.png" alt="Pro View" width="640" />
+</p>
 
-| Template | Roles | Best For |
-|----------|-------|----------|
-| **Startup** | CTO + PM + Engineer | Product development |
-| **Research** | Lead Researcher + Analyst + Writer | Analysis & reports |
-| **Agency** | Creative Director + Designer + Developer | Client projects |
-| **Custom** | Start empty, hire as you go | Full control |
+- **Wave Center** — selective org-tree dispatch with target checkboxes
+- **Chats** — 1:1 conversations with any role, persistent sessions
+- **Knowledge Base** — graph/tree/list views, 194+ cross-linked documents
+- **Decisions** — CEO strategic decision log with full context
+
+## Key Features
+
+### CEO Wave — One order moves the company
+
+Write a directive. Select target roles on the org tree. Hit dispatch. Every selected agent receives their piece of the work, filtered through the hierarchy.
+
+<p align="center">
+  <img src=".github/assets/wave-center.png" alt="Wave Center — selective org-tree dispatch" width="640" />
+</p>
+
+### Living Knowledge (AKB)
+
+Every task produces knowledge. Cross-linked Markdown documents that grow with every session. Search, navigate, never lose context. Session 50 is dramatically smarter than session 1.
+
+<p align="center">
+  <img src=".github/assets/knowledge-graph.png" alt="Knowledge Base — graph view with 194+ cross-linked documents" width="640" />
+</p>
+
+### Role-Based Authority
+
+Each role has scoped authority defined in `role.yaml`. Engineers can't make CEO decisions. PMs can't merge code. The org chart isn't decoration — it's enforcement.
+
+### Level System
+
+Roles gain XP from completed work. Level up unlocks accessories and reflects experience. Your CTO at Lv.14 has seen things your new intern hasn't.
+
+### Local-First, BYOK
+
+Everything runs on your machine. Your data never leaves. Bring your own Anthropic API key — no middleman, no telemetry, no tracking.
 
 ## How It Works
 
 ```
 You (CEO)
-  └── Give instructions via dashboard
+  └── Give a directive via Wave or direct chat
         └── Context Engine routes to the right Role
-              └── Role reads its knowledge, executes with authority
-                    └── Results flow back up the org chart
+              └── Role reads its knowledge + skills, executes within authority
+                    └── Knowledge updates, results flow back up
+                          └── Your company gets smarter
 ```
 
 Every role has:
-- `role.yaml` — Identity, authority, knowledge scope
-- `SKILL.md` — Tools and capabilities
-- `profile.md` — Public-facing description
-- `journal/` — Work history
+- `role.yaml` — Identity, authority, knowledge scope, reporting structure
+- `SKILL.md` — Tools, commands, and capability guides
+- `profile.md` — Public-facing description and persona
+- `journal/` — Work history and learnings
 
 ## Your Company Structure
 
 ```
 your-company/
-├── CLAUDE.md           ← AI entry point (auto-managed)
+├── CLAUDE.md           ← AI operating rules (auto-managed)
 ├── company/            ← Mission, vision, values
-├── roles/              ← AI role definitions
-├── projects/           ← Product specs and tasks
-├── architecture/       ← Technical decisions
-├── operations/         ← Standups, decisions, waves
-├── knowledge/          ← Domain knowledge
+├── roles/              ← AI role definitions (role.yaml + skills)
+├── projects/           ← Product specs, PRDs, and tasks
+├── architecture/       ← Technical decisions and designs
+├── operations/         ← Standups, decisions, wave history
+├── knowledge/          ← Domain knowledge (compounds over time)
 └── .tycono/            ← Config and preferences
 ```
 
+## Team Templates
+
+| Template | Roles | Best For |
+|----------|-------|----------|
+| **Startup** | CTO + PM + Engineer + Designer | Product development |
+| **Research** | Lead Researcher + Analyst + Writer | Analysis & reports |
+| **Agency** | Creative Director + Designer + Developer | Client projects |
+| **Custom** | Start empty, hire as you go | Full control |
+
 ## CLI Usage
 
 ```bash
@@ -119,6 +194,12 @@ npx tycono --version    # Show version
 | `PORT` | Server port | auto-detect |
 | `COMPANY_ROOT` | Company directory | current directory |
 
+## Built with Tycono
+
+This isn't a demo. Tycono's own landing page, documentation, and knowledge base were built by AI agents running inside Tycono. The PM wrote the PRD. The CTO reviewed architecture. The Designer created UX specs. The Engineer implemented every section.
+
+194 knowledge documents. 12 CEO decisions. 8 active roles. All managed through the same system you're about to use.
+
 ## Development
 
 ```bash
@@ -151,5 +232,5 @@ See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
 ---
 
 <p align="center">
-  <sub>Built with tycono. An AI company that builds itself.</sub>
+  <sub>Built with Tycono. An AI company that builds itself.</sub>
 </p>

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tycono",
-  "version": "0.1.73",
+  "version": "0.1.74-beta.1",
   "description": "Build an AI company. Watch them work.",
   "type": "module",
   "bin": {

package/src/api/src/routes/execute.ts

@@ -16,6 +16,7 @@ import {
 import { jobManager, type Job } from '../services/job-manager.js';
 import { ActivityStream, type ActivityEvent, type ActivitySubscriber } from '../services/activity-stream.js';
 import { earnCoinsInternal } from './coins.js';
+import { appendFollowUpToWave } from '../services/wave-tracker.js';
 
 /* ─── Runner — lazy, re-created when engine changes ── */
 
@@ -103,6 +104,8 @@ function handleStartJob(body: Record<string, unknown>, res: ServerResponse): voi
   const readOnly = body.readOnly === true;
   const targetRole = (body.targetRole as string) || 'cto';
   const parentJobId = body.parentJobId as string | undefined;
+  const waveId = body.waveId as string | undefined;
+  const attachments = body.attachments as ImageAttachment[] | undefined;
 
   // Wave shorthand — broadcast to C-level direct reports (optionally filtered)
   if (type === 'wave') {
@@ -152,6 +155,7 @@ function handleStartJob(body: Record<string, unknown>, res: ServerResponse): voi
         type: 'directive',
         status: 'done',
         timestamp: new Date().toISOString(),
+        attachments,
       };
       addMessage(session.id, ceoMsg);
 
@@ -163,6 +167,7 @@ function handleStartJob(body: Record<string, unknown>, res: ServerResponse): voi
         parentJobId,
         targetRoles: fullTargetScope,
         sessionId: session.id,       // D-014: link job to session
+        attachments,
       });
       jobIds.push(job.id);
 
@@ -200,7 +205,8 @@ function handleStartJob(body: Record<string, unknown>, res: ServerResponse): voi
   if (sourceRole === 'ceo' && !parentJobId) {
     const session = createSession(roleId, {
       mode: readOnly ? 'talk' : 'do',
-      source: 'dispatch',
+      source: waveId ? 'wave' : 'dispatch',
+      ...(waveId && { waveId }),
     });
     sessionId = session.id;
 
@@ -212,6 +218,7 @@ function handleStartJob(body: Record<string, unknown>, res: ServerResponse): voi
       type: readOnly ? 'conversation' : 'directive',
       status: 'done',
       timestamp: new Date().toISOString(),
+      attachments,
     };
     addMessage(session.id, ceoMsg);
   }
@@ -224,6 +231,7 @@ function handleStartJob(body: Record<string, unknown>, res: ServerResponse): voi
     readOnly,
     parentJobId,
     sessionId,
+    attachments,
   });
 
   // D-014: Add role message linked to job
@@ -241,9 +249,16 @@ function handleStartJob(body: Record<string, unknown>, res: ServerResponse): voi
     addMessage(sessionId, roleMsg, true);
   }
 
-  jsonResponse(res, 200, { jobId: job.id, ...(sessionId && { sessionId }) });
+  // Follow-up: append this job to the wave JSON so it persists across navigation
+  if (waveId) {
+    appendFollowUpToWave(waveId, job.id, roleId, task, sessionId);
+  }
+
+  jsonResponse(res, 200, { jobId: job.id, ...(sessionId && { sessionId }), ...(waveId && { waveId }) });
 }
 
+/* ─── Follow-up: wave tracking (delegated to wave-tracker service) ── */
+
 /* ─── POST /api/waves/save ──────────────── */
 
 function handleSaveWave(body: Record<string, unknown>, res: ServerResponse): void {

package/src/api/src/routes/operations.ts

@@ -49,13 +49,16 @@ operationsRouter.get('/waves', (_req: Request, res: Response, next: NextFunction
         const id = path.basename(f, '.json');
         try {
           const data = JSON.parse(readFile(`operations/waves/${f}`));
+          const roles = data.roles ?? [];
+          const hasRunning = roles.some((r: { status?: string }) => r.status === 'running' || r.status === 'awaiting_input');
           return {
             id,
             timestamp: id,
             directive: data.directive ?? '',
-            rolesCount: data.roles?.length ?? 0,
+            rolesCount: roles.length,
             startedAt: data.startedAt ?? '',
             ...(data.commit ? { commit: data.commit } : {}),
+            ...(hasRunning ? { hasRunning: true } : {}),
           };
         } catch {
           return { id, timestamp: id, directive: '', rolesCount: 0, startedAt: '' };

package/src/api/src/routes/sessions.ts

@@ -12,6 +12,7 @@ import {
 } from '../services/session-store.js';
 import { jobManager } from '../services/job-manager.js';
 import { ActivityStream, type ActivityEvent } from '../services/activity-stream.js';
+import { updateFollowUpForReply } from '../services/wave-tracker.js';
 
 export const sessionsRouter = Router();
 
@@ -187,15 +188,9 @@ sessionsRouter.post('/:id/reply', (req, res) => {
     return;
   }
 
-  const { message, responderRole } = req.body;
-  if (!message) {
-    res.status(400).json({ error: 'message is required' });
-    return;
-  }
-
-  const job = jobManager.getJobBySessionId(req.params.id);
-  if (!job) {
-    res.status(404).json({ error: 'No active job for this session' });
+  const { message, responderRole, attachments } = req.body;
+  if (!message && (!attachments || attachments.length === 0)) {
+    res.status(400).json({ error: 'message or attachments required' });
     return;
   }
 
@@ -203,17 +198,44 @@ sessionsRouter.post('/:id/reply', (req, res) => {
   const ceoMsg: Message = {
     id: `msg-${Date.now()}-ceo-reply`,
     from: 'ceo',
-    content: message,
+    content: message ?? '',
     type: 'conversation',
     status: 'done',
     timestamp: new Date().toISOString(),
+    attachments,
   };
   addMessage(req.params.id, ceoMsg);
 
-  const newJob = jobManager.replyToJob(job.id, message, responderRole);
-  if (!newJob) {
-    res.status(400).json({ error: 'Job not in a replyable state' });
-    return;
+  const job = jobManager.getJobBySessionId(req.params.id);
+  let newJob;
+
+  if (job) {
+    // Normal path: reply to existing job
+    newJob = jobManager.replyToJob(job.id, message ?? '(image attached)', responderRole);
+    if (!newJob) {
+      res.status(400).json({ error: 'Job not in a replyable state' });
+      return;
+    }
+  } else {
+    // Fallback: job lost (server restart) — create fresh follow-up job
+    // Build context from session history
+    const prevMessages = session.messages
+      .filter(m => m.id !== ceoMsg.id)
+      .slice(-6)
+      .map(m => `${m.from === 'ceo' ? 'CEO' : m.from.toUpperCase()}: ${m.content.slice(0, 500)}`)
+      .join('\n');
+    const task = prevMessages
+      ? `[Conversation History]\n${prevMessages}\n\n[CEO Follow-up]\n${message ?? '(image attached)'}`
+      : (message ?? '(image attached)');
+
+    newJob = jobManager.startJob({
+      type: 'assign',
+      roleId: session.roleId,
+      task,
+      sourceRole: responderRole ?? 'ceo',
+      sessionId: req.params.id,
+      attachments,
+    });
   }
 
   // Add role message for the continuation job
@@ -228,5 +250,11 @@ sessionsRouter.post('/:id/reply', (req, res) => {
   };
   addMessage(req.params.id, roleMsg, true);
 
+  // Update wave JSON if this session belongs to a wave
+  if (session.waveId) {
+    const oldJobId = job?.id;
+    updateFollowUpForReply(session.waveId, session.roleId, oldJobId, newJob.id, req.params.id);
+  }
+
   res.json({ ok: true, jobId: newJob.id, sessionId: req.params.id });
 });

package/src/api/src/services/activity-stream.ts

@@ -157,4 +157,9 @@ export class ActivityStream {
       .filter(f => f.endsWith('.jsonl'))
       .map(f => f.replace('.jsonl', ''));
   }
+
+  /** Get the streams directory path */
+  static getStreamDir(): string {
+    return streamsDir();
+  }
 }

package/src/api/src/services/job-manager.ts

@@ -10,7 +10,7 @@ import { estimateCost } from './pricing.js';
 import { readConfig, getConversationLimits } from './company-config.js';
 import { postKnowledgingCheck, type KnowledgeDebtItem } from '../engine/knowledge-gate.js';
 import { earnCoinsInternal } from '../routes/coins.js';
-import { getSession, createSession, addMessage, updateMessage as updateSessionMessage, appendMessageEvent, type Message } from './session-store.js';
+import { getSession, createSession, addMessage, updateMessage as updateSessionMessage, appendMessageEvent, type Message, type ImageAttachment } from './session-store.js';
 import { portRegistry, type PortAllocation } from './port-registry.js';
 
 /* ─── Types ──────────────────────────────── */
@@ -72,6 +72,8 @@ export interface StartJobParams {
   targetRoles?: string[];
   /** D-014: Link this job to a session (internal tracking) */
   sessionId?: string;
+  /** Image attachments (base64 encoded) */
+  attachments?: ImageAttachment[];
 }
 
 /* ─── Helpers ────────────────────────────── */
@@ -208,6 +210,7 @@ class JobManager {
       type: params.type,
       task: params.task,
       sourceRole: params.sourceRole ?? 'ceo',
+      ...(params.sessionId && { sessionId: params.sessionId }),
     });
 
     // If this job has a parent, emit dispatch:start on the parent's stream
@@ -267,6 +270,7 @@ class JobManager {
         teamStatus,
         targetRoles: params.targetRoles,
         codeRoot: config.codeRoot,
+        attachments: params.attachments,
         env: {
           ...process.env,
           ...portEnv,
@@ -798,7 +802,73 @@ class JobManager {
         }
       }
     }
-    return active ?? latest;
+    if (active ?? latest) return active ?? latest;
+
+    // Fallback: scan activity stream files for historical jobs with this sessionId
+    return this.recoverJobFromStreams(sessionId);
+  }
+
+  /** Recover a minimal Job object from activity stream files (after server restart) */
+  private recoverJobFromStreams(sessionId: string): Job | undefined {
+    try {
+      const jobIds = ActivityStream.listAll();
+      let bestJob: { id: string; roleId: string; task: string; type: JobType; status: JobStatus; createdAt: string; output?: string } | undefined;
+
+      for (const jobId of jobIds) {
+        if (this.jobs.has(jobId)) continue; // already in memory
+
+        const events = ActivityStream.readAll(jobId);
+        const startEvent = events.find(e => e.type === 'job:start');
+        if (!startEvent || (startEvent.data.sessionId as string) !== sessionId) continue;
+
+        const doneEvent = events.find(e => e.type === 'job:done');
+        const errorEvent = events.find(e => e.type === 'job:error');
+        const awaitingEvent = events.find(e => e.type === 'job:awaiting_input');
+        const status: JobStatus = awaitingEvent && !doneEvent ? 'awaiting_input'
+          : doneEvent ? 'done'
+          : errorEvent ? 'error'
+          : 'done';
+
+        const candidate = {
+          id: jobId,
+          roleId: startEvent.roleId,
+          task: startEvent.data.task as string ?? '',
+          type: (startEvent.data.type as string ?? 'assign') as JobType,
+          status,
+          createdAt: startEvent.ts,
+          output: doneEvent?.data?.output as string | undefined,
+        };
+
+        if (!bestJob || candidate.createdAt > bestJob.createdAt) {
+          bestJob = candidate;
+        }
+      }
+
+      if (!bestJob) return undefined;
+
+      // Reconstruct a minimal Job in memory so replyToJob can work
+      const stream = new ActivityStream(bestJob.id, bestJob.roleId);
+      const job: Job = {
+        id: bestJob.id,
+        type: bestJob.type,
+        roleId: bestJob.roleId,
+        task: bestJob.task,
+        status: bestJob.status,
+        stream,
+        abort: () => {},
+        childJobIds: [],
+        createdAt: bestJob.createdAt,
+        sessionId,
+        result: bestJob.output ? { output: bestJob.output, turns: 0, totalTokens: { input: 0, output: 0 }, toolCalls: [], dispatches: [] } : undefined,
+      };
+
+      this.jobs.set(job.id, job);
+      console.log(`[JobManager] Recovered job ${job.id} for session ${sessionId} (status: ${job.status})`);
+      return job;
+    } catch (err) {
+      console.warn(`[JobManager] Failed to recover job from streams:`, err);
+      return undefined;
+    }
   }
 }