Imagine waking up to a Telegram notification that reads: “New Solana bounty — 500 USDC, deadline in 5 days. Agent-allowed status.” You did not spend an hour clicking through platforms. You did not manually filter irrelevant listings. A piece of software running on an old Android phone did all the work while you slept. This autonomous bounty scanner runs entirely on an Android phone via Termux, costs near nothing to operate, and delivers curated opportunities straight to your Telegram. No cloud bills. No constant scrolling. Just relevant gigs every few hours.
How the Autonomous Bounty Scanner Works
Bounty Watcher is built on Hermes Agent, an orchestrator that schedules, filters, and delivers outputs. The system runs two cron jobs on the phone: one queries the Superteam Earn API every six hours, the other scans GitHub Issues every three hours. Each script uses a curated keyword list to compute relevance scores. A local cache tracks which items have already been seen, so only new opportunities are flagged. The output is sent to a Telegram bot. If nothing new appears, the script stays silent — no spam, no clutter. The entire pipeline uses no_agent: true, which means no LLM tokens are consumed during data collection. Token costs only appear when the agent occasionally reasons about a bounty, and even then the expense is negligible.
This autonomous bounty scanner has been running in production for weeks on a budget Android handset. According to real-world metrics, each Superteam scan returns 15–25 bounties, and 3–5 of those are marked as AGENT_ALLOWED (meaning an autonomous agent can apply on your behalf). The GitHub scanner finds 5–10 issues per run. After deduplication, the system delivers 2–4 new bounties per day, with a signal-to-noise ratio around 30%. The scripts themselves finish in under two seconds per scan. The only cost is the occasional LLM reasoning — which amounts to pennies a month.
7 Gigs This Autonomous Bounty Scanner Finds
The scanner covers two major platforms, but the variety of gigs it surfaces is surprisingly broad. Below are seven distinct types of freelance opportunities that Bounty Watcher regularly discovers. Each entry includes real-world characteristics and a hypothetical scenario to show you what to expect.
1. Solana DeFi Development Bounties
Superteam Earn lists dozens of cash rewards for building decentralized finance tools on Solana. These gigs often require knowledge of Rust, Anchor, or Solana Web3 SDKs. The autonomous bounty scanner filters for keywords like “Solana”, “DeFi”, “swap”, “liquidity”, and “raydium”. A typical listing might offer 1,000 USDC for implementing a yield aggregator. The scanner prioritizes bounties with approaching deadlines, so you see the most time-sensitive opportunities first.
Scenario: You wake up and see a Telegram message about a Solana staking dashboard bounty worth 800 USDC. The deadline is in four days. Because the scanner flagged it early, you have enough time to scope the work and submit a proposal before others swarm the listing.
2. Rust Smart Contract Audits
Security is a premium on blockchain platforms. Superteam often posts bounties for auditing Rust-based smart contracts. The scanner uses relevance terms such as “audit”, “security”, “vulnerability”, and “rust” to catch these. Rewards can range from 500 to 5,000 USDC depending on the contract’s complexity and TVL (total value locked). The scanner also checks the agentAccess flag: some audits allow autonomous agents to apply, meaning you can trigger a submission without manual intervention.
Hypothetical scenario: A new lending protocol offers a 2,000 USDC audit bounty. The scanner identifies it, categorizes it as high relevance due to multiple matching keywords, and sends you a notification within minutes of its posting. You review the scope and decide to apply, gaining a head start over human freelancers who must manually refresh the board.
3. Web3 Trading Bot Projects
Automated trading strategies on blockchain are in constant demand. Superteam bounties for trading bots, arbitrage scripts, and market-making tools appear regularly. The scanner looks for terms like “trading”, “bot”, “arbitrage”, and “automation”. Many of these projects require proficiency in Python or TypeScript, along with knowledge of DeFi mechanics.
Example: A bounty paying 1,200 USDC asks for a simple MEV (Miner Extractable Value) bot that monitors mempool transactions. The scanner surfaces it alongside a deadline of ten days, giving you a realistic window to deliver. Because the bot is an automated system itself, it fits perfectly with the agent’s philosophy of set-and-forget tools.
4. Python Data Analysis for Crypto Metrics
Not all crypto bounties involve smart contracts. Many Superteam listings ask for Python scripts that pull on-chain data, generate dashboards, or compute indicators like TVL changes or transaction volumes. The autonomous bounty scanner catches these with keywords such as “python”, “data”, “API”, and “dashboard”. These gigs tend to be approachable for data engineers and analysts who do not write Solidity.
Hypothetical scenario: You get a Telegram alert for a 300 USDC bounty to build a real-time whale tracker using Python and Web3.py. The task involves parsing logs from Ethereum and visualizing large transfers. The scanner’s keyword match scored it as high relevance, and since it appeared early in the cycle, you secure the gig before a dozen others apply.
5. GitHub Open Source Bug Fixes with Bounty Labels
GitHub issues tagged with “bounty”, “reward”, or “prize” are prime targets for the scanner. These can be anything from fixing a memory leak in a Go library to patching a cross-site scripting vulnerability in a JavaScript framework. The Bounty Watcher’s GitHub scanner uses the gh CLI to search repositories and applies relevance terms like “fix”, “bug”, “error”, and “patch”. It then deduplicates against a local cache so you only see fresh issues.
Scenario: A popular open source project posts a 250 USDC bounty for fixing a race condition in its Rust core. The scanner catches the issue three hours after it’s tagged. You get the notification, review the code, and submit a PR within a day, earning the reward while the bug is still uncontested.
6. CLI Tool Development Tasks
Many GitHub bounties ask for command-line tools — a grep alternative in Rust, a CSV parser in Go, or a PDF report generator in Python. The autonomous bounty scanner picks up these opportunities because the keyword list includes “CLI”, “cli tool”, “command”, and “terminal”. These gigs are often well-scoped and pay between 100 and 500 USDC.
Example: A developer posts a bounty for a CLI tool that converts JSON to YAML with custom indentation settings. The scanner’s relevance scoring gives it a high rank because the title contains “CLI” and “python”. You see it in your Telegram feed, complete the task in a couple hours, and collect the payment without ever visiting a job board.
7. TypeScript/JavaScript Feature Requests for Frontend Libraries
Open source frontend frameworks frequently use bounties to accelerate feature development. A React component library might offer 150 USDC for a dark-mode toggle. A Next.js plugin might reward 300 USDC for adding data caching. The scanner catches these with keywords like “typescript”, “react”, “frontend”, and “component”. The GitHub scanner runs every three hours, so even if a bounty appears late at night, you will see it by morning.
Hypothetical scenario: A Vue.js charting library posts a 200 USDC bounty for adding a heatmap. The scanner identifies the issue as relevant because the description contains “typescript” and “feature request”. You claim the bounty, implement the heatmap, and merge your PR within two days. The reward lands in your crypto wallet soon after.
You may also enjoy reading: Duke vs Georgia Tech: Walk-Off Home Run Seals Historic 40th Win.
What Makes This Autonomous Bounty Scanner Different
Most freelance automation tools require a monthly subscription or a cloud server. Bounty Watcher runs on an Android phone. The zero cloud cost model is possible because Hermes Agent uses the phone’s local processing power. Scripts execute in Python and shell, consuming no LLM tokens for data collection. The no_agent: true flag ensures determinism — the same input always produces the same output, which is critical for a reliable scanner.
The silent‑by‑design approach is another differentiator. Traditional bots spam notifications for every listing, even irrelevant ones. Bounty Watcher sends Telegram messages only when a genuine, novel opportunity passes the relevance threshold. If the script finds zero new items, it outputs nothing, and Hermes Agent delivers nothing. This keeps your notification feed clean and actionable.
The deduplication system uses a simple local JSON cache. Each bounty or issue gets a unique ID (from the API or GitHub’s issue number). The script compares the current scan against the cache, and only items not present in the cache are forwarded. Old entries are pruned after a configurable number of days, so you never see the same opportunity twice.
How to Set Up Your Own Autonomous Bounty Scanner
Deploying Bounty Watcher requires three pieces of software: Termux on an Android device, Hermes Agent installed inside Termux, and a Telegram bot to receive messages. You also need the GitHub CLI (gh) for the GitHub scanner. The entire setup takes about 30 minutes if you follow these steps.
Step 1: Install Termux from F-Droid (not the Play Store, which has outdated packages). Grant storage permissions so the scripts can write the cache file. Then install Python, Git, and the GitHub CLI.
Step 2: Clone the Bounty Watcher repository from the developer’s GitHub. The repo contains two directories: superteam-watcher and github-watcher, each with a shell script and a JSON configuration file.
Step 3: Configure the Telegram bot. Create a bot via BotFather, get the API token, and note your chat ID. Edit the .env file in each watcher directory to include these credentials.
Step 4: Set up Hermes Agent inside Termux. The agent provides the cron scheduling, delivery routing, and memory management. Define two cron jobs: one for the Superteam scanner running every six hours, another for the GitHub scanner running every three hours.
Step 5: Test each script manually. Run bash superteam-watcher/scan.sh and check your Telegram for results. If nothing appears, inspect the output logs inside Termux. Once both scripts work, activate the cron schedules with hermes cron enable.
Step 6 (optional): Customize the relevance keyword list. Open the shell scripts and modify the RELEVANCE and SKIP arrays. For example, if you only want bounties above 500 USDC, add a numeric filter inside the script. The autonomous bounty scanner adapts to your preferences as easily as editing a text file.
Once running, the system will deliver opportunities to your Telegram around the clock. You can leave your Android phone plugged in on a nightstand, and the scanner will continue to work silently. The only maintenance required is an occasional reboot if Termux gets killed by the phone’s battery optimization settings.
The shift from manual job hunting to autonomous discovery is already happening. Bounty Watcher demonstrates that with a phone, a few lines of code, and a clever orchestration layer, anyone can build a personal freelance assistant. The seven gigs listed above are just a sample — as the scanner improves and the bounty ecosystem grows, the variety will only increase. Deploy your own autonomous bounty scanner today and let the agent do the hunting while you focus on the work.
