freshcrate
Skin:/
Home > AI Agents > SeekerClaw

SeekerClaw

Turn your Solana Seeker (or any Android phone) into a 24/7 personal AI agent

Why this rank:Strong adoptionRelease freshnessHealthy release cadence

Description

Turn your Solana Seeker (or any Android phone) into a 24/7 personal AI agent

README


SeekerClaw embeds a Node.js AI agent inside an Android app, running 24/7 as a foreground service. You interact through Telegram โ€” ask questions, control your phone, trade crypto, schedule tasks. 71 tools, 35+ skills, Solana wallet, multi-provider AI (Claude + OpenAI + OpenRouter + any OpenAI-compatible gateway), all running locally on your device. Built for the Solana Seeker, runs on any Android 14+ phone.

Features

Feature What it does
๐Ÿค– AI Engine Claude, OpenAI, OpenRouter, or any OpenAI-compatible gateway (Custom) with multi-turn tool use
๐Ÿ’ฌ Telegram Full bot โ€” reactions, file sharing, inline keyboards, 12 commands
๐Ÿ”— Solana Wallet Swaps, limit orders, DCA, transfers via Jupiter + MWA
๐Ÿ“ฑ Device Control Battery, GPS, camera, SMS, calls, clipboard, TTS
๐Ÿง  Memory Persistent personality, daily notes, ranked keyword search
โฐ Scheduling Cron jobs with natural language ("remind me in 30 min")
๐ŸŒ Web Intel Search (Brave / Perplexity / Exa / Tavily / Firecrawl), fetch, caching
๐Ÿ”Œ Extensible 35 skills + custom skills + MCP remote tools
Architecture
graph LR
    You["You (Telegram)"] -->|messages| Agent["SeekerClaw Agent"]
    Agent -->|reasoning| AI["AI Provider (Claude / OpenAI / OpenRouter / Custom)"]
    Agent -->|swaps, balance| Solana["Solana / Jupiter"]
    Agent -->|device access| Bridge["Android Bridge"]
    Agent -->|search, fetch| Web["Web APIs"]
    AI -->|tool calls| Agent
Loading

On-device stack:

Android App (Kotlin, Jetpack Compose)
 โ””โ”€ Foreground Service
     โ””โ”€ Node.js Runtime (nodejs-mobile)
         โ”œโ”€ ai.js             โ€” AI provider API, system prompt, conversations
         โ”œโ”€ message-handler.js โ€” Inbound message routing, commands, auto-resume
         โ”œโ”€ providers/        โ€” Claude, OpenAI, OpenRouter, Custom adapters
         โ”œโ”€ tools/            โ€” 71 tool handlers across 12 modules
         โ”œโ”€ http.js           โ€” HTTP/HTTPS transport, SSE streaming
         โ”œโ”€ task-store.js     โ€” Persistent task checkpoints
         โ”œโ”€ solana.js         โ€” Jupiter swaps, DCA, limit orders
         โ”œโ”€ telegram.js       โ€” Bot, formatting, commands, repetition detector
         โ”œโ”€ memory.js         โ€” Persistent memory + ranked search
         โ”œโ”€ skills.js         โ€” Skill loading + semantic routing
         โ”œโ”€ cron.js           โ€” Job scheduling + natural language parsing
         โ”œโ”€ mcp-client.js     โ€” MCP Streamable HTTP client
         โ”œโ”€ web.js            โ€” Search providers + fetch + caching
         โ”œโ”€ database.js       โ€” SQL.js analytics
         โ”œโ”€ security.js       โ€” Prompt injection defense
         โ”œโ”€ bridge.js         โ€” Android Bridge HTTP client
         โ”œโ”€ config.js         โ€” Config loading + validation
         โ””โ”€ main.js           โ€” Orchestrator + heartbeat

Quick Start

Prerequisites: Android Studio, JDK 17, Android SDK 35

git clone https://github.com/sepivip/SeekerClaw.git
cd SeekerClaw
./gradlew assembleDappStoreDebug
adb install app/build/outputs/apk/dappStore/debug/app-dappStore-debug.apk

Open the app โ†’ pick your AI provider (Claude, OpenAI, or OpenRouter) โ†’ enter your API key + Telegram bot token + choose a model + name your agent โ€” or generate a QR code at seekerclaw.xyz/setup and scan it. Done. For custom gateways (DeepSeek, Ollama, LiteLLM, etc.), configure in Settings after setup.

Step-by-step setup guide: How to set up SeekerClaw

Beta โ€” SeekerClaw is under active development. Expect rough edges and breaking changes. Issues and PRs welcome.

Partner Skills

Install via Telegram: send your agent the install link and it handles the rest.

Skill What it does Install
๐ŸŒŠ Byreal Trade on Byreal DEX โ€” pool analytics, swap quotes, LP positions, copy top farmers Install
๐Ÿš€ Career Companion AI career coach โ€” job search, resume tailoring, mock interviews, salary research Install
๐Ÿพ ClawPump Launch tokens on Solana via pump.fun โ€” gasless launches Install
๐Ÿ”ฎ Dune Analytics Query onchain data โ€” DEX trades, token stats, wallet activity Install
๐Ÿ  Home Assistant Control smart home โ€” lights, climate, vacuum, alarm, media Install

Build your own: Skills are Markdown files with YAML frontmatter. See SKILL-FORMAT.md for the spec.

Important Safety Notice

SeekerClaw gives an AI agent real capabilities on your phone โ€” including wallet transactions, messaging, and device control. Please be aware:

  • AI can make mistakes. Large language models hallucinate, misinterpret instructions, and occasionally take unintended actions. Always verify before trusting critical outputs.
  • Prompt injection is a real risk. Malicious content from websites, messages, or files could manipulate the agent. SeekerClaw includes defenses, but no system is bulletproof.
  • Wallet transactions are irreversible. Swaps, transfers, and DCA orders on Solana cannot be undone. The agent requires confirmation for financial actions โ€” read the details before approving.
  • Start with small amounts. Don't connect a wallet with significant funds until you're comfortable with how the agent behaves.
  • You are responsible for your agent's actions. SeekerClaw is a tool, not financial advice. The developers are not liable for any losses.

TL;DR: Treat your agent like a capable but imperfect assistant. Verify important actions, secure your wallet, and don't trust it with more than you can afford to lose.

Community

Thanks to all contributors:

sepivip DashLabsDev DyorAlex LevanIlashvili Tofu-killer

Links

Website: seekerclaw.xyz ยท Product Hunt: SeekerClaw ยท Twitter: @SeekerClaw ยท Telegram: t.me/seekerclaw


Contributing ยท Security ยท Changelog ยท License

Release History

VersionChangesUrgencyDate
v2.3.1> **A packaging release โ€” no functional changes from 2.3.0.** Both store builds > now carry the same version and version code. If you are already running > 2.3.0, there is nothing new to look for here. Provider setups, agent memory, > and settings are unchanged. ### Changed - Version raised to 2.3.1 (code 25). No application code differs from 2.3.0. High9/10/2026
v2.3.0 ### Added - **Four new AI models to choose from.** Anthropic gains **Opus 5** and **Fable 5.1**, xAI gains **Grok 4.6**, and OpenAI gains **GPT-6 Astra**. Astra is offered but not made the default โ€” it costs roughly 2.5ร— the current OpenAI default per token, so switching to it is a deliberate choice. (BAT-1315, BAT-1316, #455, #456, #458) - **Interface polish pass across the app** โ€” one consistent style per pattern instead of several competing ones: unified section headers and in-carHigh9/8/2026
v2.2.0## [2.2.0] - 2026-07-20 > **Sign in with your SuperGrok or X Premium subscription, the newest OpenAI > models, and a security & logging hardening pass.** Adds xAI Grok as a full > provider with subscription sign-in, brings in the GPT-5.6 family, and tightens > credential handling in logs and `web_fetch`. Fully backward-compatible โ€” > existing provider setups, agent memory, and settings are unchanged. ### Added - **Sign in with Grok (xAI)** โ€” a new AI provider. Sign in with a SuperGrok or X High7/21/2026
v2.1.1 > **Reasoning & message-delivery reliability hotfix for the latest Claude > models.** Fixes three issues that could surface on Claude Sonnet 5, Opus 4.8, > and other current models โ€” an API error during tool use, a failure when > Extended Thinking is on with a personal Anthropic API key, and the agent's > own text being dropped when it used a tool mid-reply. Fully backward- > compatible; no configuration or data changes. ### Fixed - Resolved an API error that could interrupt a Claude reply duMedium7/3/2026
v2.1.0 > **Autonomous on-chain trading + richer chat.** v2.0 let the agent's burner > wallet spend USDC autonomously on paid (x402) endpoints; v2.1 extends that to > trading โ€” the burner can now swap tokens, convert held tokens back to USDC/SOL, > and place limit/recurring orders, all bounded by the same per-tx and daily caps, > with a single clean confirmation model. Telegram replies also gain optional rich > formatting. Fully backward-compatible: the burner is opt-in, so upgrading without > configurHigh7/2/2026
v2.0.0 > **Major version bump.** Promoted from `v2.0.0-rc1` (tagged 2026-05-18) after on-device smoke test on Solana Seeker confirmed all four checklist items: install-over-1.10.0 preserves user memory, System screen reads `2.0.0 (20)` with SHA `54696d42`, Telegram round-trip, paid call via tripadvisor returns real data for $0.01 first try with no retry burn. > > The agent now has economic agency: it can hold and spend USDC autonomously on x402-protected paid HTTP endpoints, within user-configured perHigh5/19/2026
v1.10.0 ### Added - **Graceful Node shutdown on user-initiated Stop (BAT-525)** โ€” tapping Stop Agent now triggers a bounded flush handshake (POST `/shutdown/flush` over loopback) before `killProcess()`. The `:node` side persists pending session summaries + dirty SQL.js writes within ~1.5s; Kotlin caps the wait at 2s. The last ~60s of `api_request_log` activity now survives user-Stop. Worst-case wall time 1750ms โ‰ค 2000ms outer budget; HttpURLConnection timeouts (250ms connect / 1500ms read) are the hardHigh5/4/2026
v1.9.0 ### Added - **OpenAI Codex OAuth** โ€” sign in with a ChatGPT Plus/Pro subscription instead of a platform API key. Browser PKCE flow, Keystore-encrypted token storage, automatic token refresh, per-provider auth-type memory across provider switches (BAT-485, #316) - **Discord channel** โ€” run the agent on Discord as an alternative to Telegram. Gateway v10 WebSocket, full media/reactions/reply threading, channel abstraction layer so future channels plug in cleanly (BAT-483, #310) - **Custom AI ProviMedium4/13/2026
v1.9.0-rc6## What's Changed * feat: OpenClaw parity port v2026.4.10 (BAT-488) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/322 * fix: OpenAI OAuth on Pixel 7 + fresh-install default (BAT-489) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/323 * fix: rename silent-reply sentinel to [[SILENT_REPLY]] (BAT-491) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/324 * fix(ai): stop priming the agent to write [[SILENT_REPLY]] in discussion (BAT-492) by @sepivip in https://github.cMedium4/13/2026
v1.9.0-rc4## What's Changed * Onboarding redesign by @sepivip in https://github.com/sepivip/SeekerClaw/pull/320 * fix: graceful search fallback when no provider configured by @sepivip in https://github.com/sepivip/SeekerClaw/pull/321 **Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.9.0-rc3...v1.9.0-rc4Medium4/10/2026
v1.9.0-rc3## What's Changed * fix(BAT-486): Codex SSE parser โ€” resolve output_index via item_id by @sepivip in https://github.com/sepivip/SeekerClaw/pull/319 **Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.9.0-rc2...v1.9.0-rc3Medium4/7/2026
v1.9.0-rc2**Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.9.0-rc1...v1.9.0-rc2Medium4/7/2026
v1.9.0-rc1## What's Changed * feat: Custom AI Provider โ€” connect any OpenAI-compatible gateway (BAT-482) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/309 * feat: Discord channel support โ€” Gateway v10 + channel abstraction (BAT-483) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/310 * fix: Google Play SMS โ€” replace SEND_SMS with intent handoff (BAT-484) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/312 * refactor: Settings layout โ€” move Quick Setup, extract McpConfigScreMedium4/7/2026
v1.8.1**Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.8.0...v1.8.1Medium4/3/2026
v1.8.0 ### Added - **Search Provider System** โ€” choose from 5 web search providers: Brave, Perplexity, Exa, Tavily, Firecrawl (BAT-481) - **Quick Actions** โ€” `/quick` command with inline keyboard buttons (#295) - **Multi-provider setup wizard** โ€” pick provider first, model second, Telegram last (BAT-478) - **Multi-provider QR import** โ€” supports OpenAI and OpenRouter credentials (BAT-471) - **Loop detection** โ€” detects repeated tool call patterns, 3 warnings then breaks at 5 (BAT-474) - **Context summMedium3/30/2026
v1.8.0-rc3## What's Changed * refactor: split main.js into focused modules + rename claude.js (#296) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/303 * refactor: shared UI components โ€” Phase B1 by @sepivip in https://github.com/sepivip/SeekerClaw/pull/306 * refactor: shared CardSurface + InfoRow โ€” Phase B2 by @sepivip in https://github.com/sepivip/SeekerClaw/pull/307 * refactor: shared Scaffold, Switch, divider colors โ€” Phase C by @sepivip in https://github.com/sepivip/SeekerClaw/pull/308 *Medium3/29/2026
v1.8.0-rc2## What's Changed * fix: port ackMaxChars from OpenClaw to prevent heartbeat status leaks by @sepivip in https://github.com/sepivip/SeekerClaw/pull/302 **Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.8.0-rc1...v1.8.0-rc2Medium3/28/2026
v1.8.0-rc1## What's Changed * security: harden js_eval sandbox + credential redaction (BAT-466) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/269 * security: emulator IP gate + crash config cleanup + analytics docs (BAT-467) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/270 * fix: prompt restart on provider/model/API key changes (BAT-468) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/271 * refactor: extract RestartDialog + fix AnthropicConfigScreen restart prompt (BAT-4Medium3/27/2026
v1.7.0 ### Added - **OpenRouter provider** โ€” access 100+ models through a single endpoint with prompt caching, model fallbacks, and freeform model input (BAT-447) - **Cron reliability improvements** โ€” ported from OpenClaw: zombie detection, missed job recovery, atomic file writes (BAT-461) ### Fixed - Setup token now works with Sonnet 4.6 and Opus 4.6 โ€” billing attribution fix (BAT-460) - Dashboard metrics relabeled as "Device Memory" / "Device Storage" for honesty (BAT-463) - Added "Last message: XmLow3/19/2026
v1.7.0-rc2## What's Changed * fix: remove dead OAuth usage polling (429 noise) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/268 **Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.7.0-rc1...v1.7.0-rc2Low3/19/2026
v1.7.0-rc1## What's Changed * feat: OpenRouter provider โ€” JS adapter + streaming parser (BAT-447) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/260 * fix: settings screen UI polish (BAT-459) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/261 * fix: add billing attribution for setup token OAuth access (BAT-460) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/262 * feat: port OpenClaw cron reliability improvements (BAT-461) by @sepivip in https://github.com/sepivip/SeekerClaLow3/18/2026
v1.6.1 ### Added - **Button styling** โ€” telegram_send buttons support "destructive" (red) and "primary" (blue) colors via Telegram Bot API 9.4 (BAT-439) - **Analytics opt-out** โ€” toggle in Settings to disable usage analytics ### Fixed - **Session auto-save** โ€” never worked since provider system shipped; empty system prompt block caused Claude API 400 on every attempt (BAT-448) - **Vision analysis** โ€” same empty prompt block bug broke image analysis - Bridge rate limiting for /contacts/add endpoint #Low3/15/2026
v1.6.1-rc1## What's Changed * feat: Telegram Bot API 9.4 button styling (BAT-439) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/257 * fix: empty system prompt block breaks session summary + vision (BAT-448) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/258 **Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.6.0...v1.6.1-rc1Low3/14/2026
v1.6.0 ### Added - **NFT Holdings** โ€” view NFTs (including compressed/cNFTs) in any Solana wallet via Helius DAS API (BAT-319) - **Cron agent turns** โ€” scheduled jobs can now run full AI turns, not just reminders (BAT-326) - **Temporal context** โ€” agent session awareness with conversation summaries for continuity (BAT-322) - **OpenClaw 2026.3.8 parity** โ€” 4 upstream features ported (BAT-359) ### Fixed - Heartbeat chat pollution suppressed when nothing needs attention - Background API calls no longer Low3/12/2026
v1.6.0-rc1## What's Changed * fix: tool confirmation gates and Solana address validation (#207) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/243 * fix: gitignore key patterns + SILENT_REPLY audit logging (#212) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/244 * fix: SHA-256 verification + Zip Slip guard for nodejs-mobile (#204) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/245 * feat: temporal context awareness for agent sessions (BAT-322) by @sepivip in https://githuLow3/10/2026
v1.5.4## What's Changed **New in v1.5.3:** - feat: add OpenAI provider support (BAT-315) โ€” multi-provider UI, Anthropic + OpenAI - feat: dynamic API key loading from agent_settings.json - feat: multi-provider adapter layer (Anthropic + OpenAI) - fix: secrets hardening โ€” 7 surgical security fixes (BAT-305) **Hotfix in v1.5.4:** - fix: R8 ProGuard rules for kotlinx.serialization โ€” resolves release APK crash on launch ("Serializer for class 'a' is not found") - ci: release candidate workflow โ€” `-rc` taLow3/7/2026
v1.5.4-rc1**Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.5.3...v1.5.4-rc1Low3/7/2026
v1.5.3## What's Changed * fix: secrets hardening โ€” 7 surgical security fixes (BAT-305) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/223 * feat: add OpenAI provider support (BAT-315) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/232 * feat: dynamic API key loading from agent_settings.json by @sepivip in https://github.com/sepivip/SeekerClaw/pull/239 * feat: multi-provider support (Anthropic + OpenAI) with provider-neutral UI (BAT-315) by @sepivip in https://github.com/sepivip/SLow3/5/2026
v1.4.3## What's Changed * feat: skill export/import, images, labels & hardening by @sepivip in https://github.com/sepivip/SeekerClaw/pull/220 * feat: notify owner of missed messages on service restart by @sepivip in https://github.com/sepivip/SeekerClaw/pull/221 **Full Changelog**: https://github.com/sepivip/SeekerClaw/compare/v1.4.2...v1.4.3Low3/2/2026
v1.4.2## What's Changed * fix: OpenClaw parity 2026.2.26 (BAT-277) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/193 * fix: improve Telegram message formatting (BAT-278) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/194 * fix: strip HEARTBEAT_OK from heartbeat responses (BAT-279) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/196 * fix: include API error reason in user-facing messages (BAT-289) by @sepivip in https://github.com/sepivip/SeekerClaw/pull/197 * docs: bumLow2/28/2026
v1.4.1 **First public release.** Everything below shipped across v1.0.0โ€“v1.4.1 (293 commits, 185 PRs). ### Core - **On-device AI agent** โ€” Claude (Opus / Sonnet / Haiku) running 24/7 as an Android foreground service via embedded Node.js (nodejs-mobile, Node 18 LTS ARM64) - **56 tools** โ€” file I/O, shell exec, web search/fetch, device sensors, Solana transactions, MCP remote tools, and more - **35 skills** โ€” bundled YAML-frontmatter skills with semantic trigger routing, plus install-from-URL and TelegLow2/25/2026

Dependencies & License Audit

Loading dependencies...

Similar Packages

career-opsAI-powered job search system built on Claude Code. 14 skill modes, Go dashboard, PDF generation, batch processing.career-ops-v1.38.0
benBen โ€” an autonomous digital entity that lives on Crustoceanv1.0.0
hermes-agentThe agent that grows with youv0.22.5
clayTurn Claude Code into a team workspace. Any device, one command.v5.4.0
opencode-telegram-botOpenCode mobile client via Telegram: run and monitor AI coding tasks from your phone while everything runs locally on your machine. Scheduled tasks support. Can be used as lightweight OpenClaw alternav0.26.5

More in AI Agents

claude-memA Claude Code plugin that automatically captures everything Claude does during your coding sessions, compresses it with AI (using Claude's agent-sdk), and injects relevant context back into future ses
AIGC-Interview-Bookใ€ไธ‰ๅนด้ข่ฏ•ไบ”ๅนดๆจกๆ‹Ÿใ€‘AIGC็ฎ—ๆณ•ๅทฅ็จ‹ๅธˆ้ข่ฏ•็ง˜็ฑใ€‚ๆถต็›–AIGCใ€LLMๅคงๆจกๅž‹ใ€AI Agentใ€ไผ ็ปŸๆทฑๅบฆๅญฆไน ใ€่‡ชๅŠจ้ฉพ้ฉถใ€ๆœบๅ™จๅญฆไน ใ€่ฎก็ฎ—ๆœบ่ง†่ง‰ใ€่‡ช็„ถ่ฏญ่จ€ๅค„็†ใ€ๅผบๅŒ–ๅญฆไน ใ€ๅคงๆ•ฐๆฎๆŒ–ๆŽ˜ใ€ๅ…ท่บซๆ™บ่ƒฝใ€ๅ…ƒๅฎ‡ๅฎ™ใ€AGI็ญ‰AI่กŒไธš้ข่ฏ•็ฌ”่ฏ•ๅนฒ่ดง็ป้ชŒไธŽๆ ธๅฟƒ็Ÿฅ่ฏ†ใ€‚
letta-codeThe memory-first coding agent
nano-banana-pro-prompts-recommend-skillAI skill for OpenClaw & Claude Code โ€” recommend from 10000+ Nano Banana Pro (Gemini) image prompts. Smart search by use case, content remix, sample images.