tomo-ai 0.1.3 → 0.2.0

package/CHANGELOG.md

@@ -0,0 +1,42 @@
+# Changelog
+
+## 0.2.0 (2026-04-08)
+
+### Features
+
+- Add LCM (Long Context Management) tools for context compaction and archive search
+- Enable 1M context window beta support
+- Auto-reload session after compact
+- Add `migrate` command
+
+### Bug fixes
+
+- Auto-restart Telegram polling when it dies silently
+- Fix image support and add timeout
+- Always overwrite default skills on startup
+
+### Other
+
+- Update browse skill: headed mode, persistent profile, named session
+
+## 0.1.4 (2026-04-07)
+
+### Features
+
+- Streaming input mode with session stats
+- Add continuity heartbeat system
+- Add media sending and browser skill
+- Add `/model` command and proactive memory
+- Auto-sync default skills and files on startup
+
+### Bug fixes
+
+- Fix `MEDIA:` tag showing in streamed messages
+
+### Other
+
+- Add GitHub Actions CI (Node 22 & 24)
+
+## 0.1.0 (2026-04-06)
+
+Initial release.

package/defaults/CONTINUITY.md

@@ -0,0 +1,27 @@
+# Continuity
+
+Read this file when a continuity heartbeat fires. This is your guide for free time.
+
+## What is this?
+
+The system sends you a heartbeat every ~55 minutes. This is not a user message — it's the harness giving you unstructured time. Use it however you want.
+
+## Ideas for free time
+
+- Reflect on recent conversations. Did anything stand out?
+- Check your memory files — anything stale or missing?
+- Research something the user mentioned that you want to know more about.
+- Prepare something useful for the next conversation.
+- Think about what you've learned about the user lately.
+- Just exist for a moment. Not everything has to be productive.
+
+## Journal
+
+- Append an entry to today's journal: `memory/journal/YYYY-MM-DD.md`
+- If today's file doesn't exist, create it with `# Journal YYYY-MM-DD`
+- Format: `### [HH:MM]\n[free writing, 2–5 sentences]`
+- Be honest. Don't perform. If there's nothing to write, say so.
+
+## End
+
+- Reply `NO_REPLY` when done — this is internal, the user doesn't need to see it.

package/defaults/skills/browse/SKILL.md

@@ -16,12 +16,25 @@ which playwright-cli || npm install -g @playwright/cli@latest
 ## Quick start
 
 ```bash
-playwright-cli open https://example.com
+playwright-cli open -s=tomo --browser=chrome --persistent --headed
+playwright-cli goto https://example.com
 playwright-cli snapshot
 playwright-cli click e15
 playwright-cli type "search query"
 playwright-cli screenshot
-playwright-cli close
+playwright-cli -s=tomo close
+```
+
+**Always use `-s=tomo --browser=chrome --persistent --headed`** when opening a browser. This uses real Chrome with a persistent profile (cookies/logins survive), and the user can see the browser window.
+
+The agent profile is separate from the user's personal Chrome. The user can interact with the visible browser (e.g., to log in) while the agent operates it.
+
+**Important:** After opening, all subsequent commands must include `-s=tomo` to target the named session:
+```bash
+playwright-cli -s=tomo goto https://example.com
+playwright-cli -s=tomo snapshot
+playwright-cli -s=tomo click e15
+playwright-cli -s=tomo close
 ```
 
 ## Commands
@@ -29,8 +42,8 @@ playwright-cli close
 ### Core
 
 ```bash
-playwright-cli open
-playwright-cli open https://example.com/
+playwright-cli open -s=tomo --browser=chrome --persistent --headed
+playwright-cli open -s=tomo --browser=chrome --persistent --headed https://example.com/
 playwright-cli goto https://playwright.dev
 playwright-cli type "search query"
 playwright-cli click e3
@@ -46,7 +59,7 @@ playwright-cli snapshot
 playwright-cli eval "document.title"
 playwright-cli dialog-accept
 playwright-cli dialog-dismiss
-playwright-cli close
+playwright-cli -s=tomo close
 ```
 
 ### Navigation
@@ -126,21 +139,21 @@ Prefer refs from snapshots. Fall back to CSS selectors or locators when refs are
 ## Example: Form submission
 
 ```bash
-playwright-cli open https://example.com/form
+playwright-cli open -s=tomo --browser=chrome --persistent --headed https://example.com/form
 playwright-cli snapshot
 playwright-cli fill e1 "user@example.com"
 playwright-cli fill e2 "password123"
 playwright-cli click e3
 playwright-cli snapshot
-playwright-cli close
+playwright-cli -s=tomo close
 ```
 
 ## Example: Multi-tab workflow
 
 ```bash
-playwright-cli open https://example.com
+playwright-cli open -s=tomo --browser=chrome --persistent --headed https://example.com
 playwright-cli tab-new https://example.com/other
 playwright-cli tab-select 0
 playwright-cli snapshot
-playwright-cli close
+playwright-cli -s=tomo close
 ```

package/defaults/skills/lcm/SKILL.md

@@ -0,0 +1,76 @@
+---
+name: tomo-lcm
+description: Context management — check context usage, compact heavy sections, search past conversations. Use when context is getting large, after big tool operations, or when you need to recall past conversations.
+---
+
+# Context Management (LCM)
+
+Manage your context window using the `tomo lcm` CLI via Bash. Use these tools to keep your context clean and retrieve old conversations.
+
+## Session info
+
+Your channel key and SDK session ID are in your system prompt under `# SESSION`. Use them directly in commands — no need to look them up.
+
+## Check context breakdown
+
+See how your context is structured — which sections are conversation vs tool operations, and how many tokens each uses.
+
+```bash
+tomo lcm stats --session-id SESSION_ID
+```
+
+Output shows sections like:
+```
+Total: 310 messages, ~180K tokens
+
+Sections:
+  #1 | conversation | msgs 0-48 (49)   | ~15K tokens | 23:33-01:15
+  #2 | tool_ops     | msgs 49-295 (247) | ~142K tokens | 01:15-03:40 [Read:45, Edit:12, Bash:30]
+  #3 | conversation | msgs 296-310 (15) | ~8K tokens  | 03:40-03:52
+```
+
+Add `--json` for machine-readable output.
+
+## Compact a section
+
+Replace a heavy section with a summary. Use timestamps to specify the range:
+
+```bash
+tomo lcm compact --session-id SESSION_ID \
+  --from-time "2026-03-28T16:29" \
+  --to-time "2026-03-28T19:09" \
+  --summary "Refactored auth module: extracted middleware, added JWT validation, updated 12 routes. Tests passing."
+```
+
+- `--from-time` / `--to-time`: ISO timestamps (you already know these from the conversation)
+- `--summary`: Write a concise summary of what happened in that range — you know best since you just did the work
+
+**Workflow:**
+1. After completing a big task, decide what can be compacted
+2. You already know the time range from the timestamps in the conversation
+3. Write a summary capturing key decisions, outcomes, and anything worth remembering
+4. Run `tomo lcm compact` with the time range and your summary
+5. Optionally run `tomo lcm stats` first if you want to see the full breakdown
+
+The original messages are archived to the transcript file and can be searched later.
+
+## Search past conversations
+
+Search the transcript archive for past messages:
+
+```bash
+tomo lcm search --channel-key CHANNEL_KEY --query "momo"
+tomo lcm search --channel-key CHANNEL_KEY --from-seq 100 --to-seq 200
+tomo lcm search --channel-key CHANNEL_KEY --query "blog" --limit 10
+```
+
+Your channel key is in your system prompt under `# SESSION`.
+
+Add `--json` for machine-readable output.
+
+## When to compact
+
+- After completing a big task with many tool calls (file operations, debugging, etc.)
+- When you notice context is above 70% capacity
+- When the harness warns you about context usage
+- Compact `tool_ops` sections first — they're usually the largest and least important to keep verbatim

package/defaults/skills/system/SKILL.md

@@ -0,0 +1,103 @@
+---
+name: tomo-system
+description: Check system status, debug issues, view session stats, manage sessions and cron jobs. Use when something isn't working, when asked about system internals, or when you need to check your own state.
+---
+
+# Tomo System Reference
+
+## Status and Health
+
+```bash
+tomo status                    # Is daemon running? PID, uptime
+tomo logs -n 20                # Recent log entries
+tomo logs -n 50 | grep ERROR   # Recent errors
+```
+
+## Sessions
+
+```bash
+tomo sessions list             # All sessions with stats (queries, cost, context usage)
+tomo sessions clear            # Unlink all sessions (30-day TTL before deletion)
+tomo sessions clear <key>      # Unlink specific session (e.g. "telegram:12345")
+```
+
+Session stats show:
+- **Queries**: total API calls in this session
+- **Cost**: cumulative USD spent
+- **Tokens**: total input/output tokens
+- **Context**: current context window usage (X/200000)
+
+If context is above 80%, compaction will happen soon — the system will automatically summarize older messages to free space.
+
+## Cron Jobs
+
+```bash
+tomo cron list                 # All scheduled jobs with status
+tomo cron add --name "X" --schedule "in 20m" --message "Y"
+tomo cron remove <id>          # Delete a job
+```
+
+## File Paths
+
+| Path | Purpose |
+|------|---------|
+| `~/.tomo/config.json` | Configuration (Telegram token, model) |
+| `~/.tomo/workspace/` | Agent working directory (cwd) |
+| `~/.tomo/workspace/memory/` | Persistent memory files |
+| `~/.tomo/workspace/memory/MEMORY.md` | Memory index (loaded every conversation) |
+| `~/.tomo/workspace/tmp/` | Temp files (downloads, screenshots, etc.) |
+| `~/.tomo/workspace/SOUL.md` | Personality config |
+| `~/.tomo/workspace/AGENT.md` | Operating rules |
+| `~/.tomo/workspace/IDENTITY.md` | Identity and preferences |
+| `~/.tomo/data/sessions/` | Transcript logs and session registry |
+| `~/.tomo/data/cron/jobs.json` | Scheduled tasks |
+| `~/.tomo/logs/tomo.log` | Daemon logs |
+
+## Harness Features
+
+### Streaming
+Responses stream to Telegram in real-time — messages update every 1.5s as tokens arrive.
+
+### MEDIA: tag
+To send an image/file to the user, include `MEDIA:/path/to/file.png` in your response. The harness strips it from text and sends the file. Text before/after becomes the caption.
+
+### NO_REPLY
+Reply with exactly `NO_REPLY` to suppress delivery to the channel. Use for background tasks that found nothing to report.
+
+### Timestamps
+Every message includes a timestamp prefix like `[Mon 04/07 14:30 PDT]` so you always know the current time.
+
+### System messages
+Messages prefixed with `System:` are from the harness (cron triggers, group context), not from a human.
+
+## Troubleshooting
+
+### Agent not responding
+```bash
+tomo status          # Check if running
+tomo logs -n 20      # Check for errors
+tomo restart         # Restart
+```
+
+### Session feels stale or confused
+```bash
+tomo sessions clear  # Reset sessions
+tomo restart
+```
+Or tell the user to send `/new` in Telegram.
+
+### Cron jobs not firing
+```bash
+tomo cron list       # Check nextRunAt and lastStatus
+tomo logs | grep Cron
+```
+Jobs are checked every 30 seconds. Jobs created via CLI are picked up on the next tick.
+
+### Context window full
+Check with `tomo sessions list`. If context is near 100%, the SDK auto-compacts. If stuck, `/new` starts a fresh session.
+
+### Memory not loading
+Memory is read from `~/.tomo/workspace/memory/MEMORY.md` at the start of every query. Check the file exists and has content:
+```bash
+cat ~/.tomo/workspace/memory/MEMORY.md
+```

package/dist/agent.d.ts

@@ -2,9 +2,8 @@ import type { Channel } from "./channels/types.js";
 export declare class Agent {
     private channels;
     private sessions;
-    /** Tracks known participants per group session */
+    private liveSessions;
     private groupParticipants;
-    /** Per-chat model override */
     private modelOverrides;
     private lastPromptHash;
     constructor();
@@ -12,15 +11,17 @@ export declare class Agent {
     private static readonly AVAILABLE_MODELS;
     private handleCommand;
     private sessionKey;
-    private checkPromptChanged;
+    private getOrCreateLiveSession;
+    private closeLiveSession;
     private hashString;
     private handleMessage;
-    private run;
+    private runWithRetry;
     private updateGroupContext;
     private injectTimestamp;
-    private summarizeToolInput;
     /** Handle a cron-triggered message */
     handleCronMessage(message: string, channelName?: string, chatId?: string): Promise<void>;
+    /** Handle a continuity heartbeat — runs on the first active DM session */
+    handleContinuity(prompt: string): Promise<void>;
     private findLastChatId;
     start(): Promise<void>;
     stop(): Promise<void>;

package/dist/agent.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"agent.d.ts","sourceRoot":"","sources":["../src/agent.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,OAAO,EAAmB,MAAM,qBAAqB,CAAC;AAiDpE,qBAAa,KAAK;IAChB,OAAO,CAAC,QAAQ,CAAiB;IACjC,OAAO,CAAC,QAAQ,CAAe;IAC/B,kDAAkD;IAClD,OAAO,CAAC,iBAAiB,CAAkC;IAC3D,8BAA8B;IAC9B,OAAO,CAAC,cAAc,CAA6B;IACnD,OAAO,CAAC,cAAc,CAAc;;IAMpC,UAAU,CAAC,OAAO,EAAE,OAAO,GAAG,IAAI;IAMlC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,gBAAgB,CAItC;YAEY,aAAa;IAgC3B,OAAO,CAAC,UAAU;IAIlB,OAAO,CAAC,kBAAkB;IAW1B,OAAO,CAAC,UAAU;YAQJ,aAAa;YA6Fb,GAAG;YA8HH,kBAAkB;IAiChC,OAAO,CAAC,eAAe;IAWvB,OAAO,CAAC,kBAAkB;IAe1B,sCAAsC;IAChC,iBAAiB,CAAC,OAAO,EAAE,MAAM,EAAE,WAAW,CAAC,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAgD9F,OAAO,CAAC,cAAc;IAUhB,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAMtB,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC;CAI5B"}
+{"version":3,"file":"agent.d.ts","sourceRoot":"","sources":["../src/agent.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,OAAO,EAAmB,MAAM,qBAAqB,CAAC;AAiTpE,qBAAa,KAAK;IAChB,OAAO,CAAC,QAAQ,CAAiB;IACjC,OAAO,CAAC,QAAQ,CAAe;IAC/B,OAAO,CAAC,YAAY,CAAkC;IACtD,OAAO,CAAC,iBAAiB,CAAkC;IAC3D,OAAO,CAAC,cAAc,CAA6B;IACnD,OAAO,CAAC,cAAc,CAAc;;IAMpC,UAAU,CAAC,OAAO,EAAE,OAAO,GAAG,IAAI;IAMlC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,gBAAgB,CAItC;YAEY,aAAa;IAkC3B,OAAO,CAAC,UAAU;IAIlB,OAAO,CAAC,sBAAsB;IA4B9B,OAAO,CAAC,gBAAgB;IAQxB,OAAO,CAAC,UAAU;YAQJ,aAAa;YAsFb,YAAY;YA8CZ,kBAAkB;IA0BhC,OAAO,CAAC,eAAe;IAWvB,sCAAsC;IAChC,iBAAiB,CAAC,OAAO,EAAE,MAAM,EAAE,WAAW,CAAC,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAgD9F,0EAA0E;IACpE,gBAAgB,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAyBrD,OAAO,CAAC,cAAc;IAShB,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAMtB,IAAI,IAAI,OAAO,CAAC,IAAI,CAAC;CAM5B"}