GitHub bounty programs pay real money for fixes — Expensify alone pays $175–250 per merged PR, and platforms like Algora, Polar, and Opire aggregate hundreds of funded issues across repos. The problem is discovery: the good issues get claimed within hours, and nobody wants to refresh five different boards all day.
Here's how to build an automated bounty radar — the GitHub queries, the scoring logic, and the deduplication trick — so new funded issues hit your Telegram before the crowd sees them.
1. The GitHub search queries that matter
The GitHub Search API is free (30 req/min authenticated) and covers most of the bounty surface if you know the labels:
# Algora bounties (funded via algora.io) is:issue is:open label:"💎 Bounty" no:assignee sort:created-desc # Polar funding badges is:issue is:open label:"polar" no:assignee sort:created-desc # Expensify — the big one, $175-250/issue paid via Upwork repo:Expensify/App is:issue is:open label:"Help Wanted" no:assignee # Generic bounty labels is:issue is:open label:bounty no:assignee sort:created-desc Two details that matter more than the queries:
-
no:assigneeis non-negotiable. Issues with assignees are dead leads — someone's already working on them. Filtering at query time saves rate limits. -
sort:created-desc+ a freshness filter. On Expensify, proposals get picked ~5 hours after theHelp Wantedlabel lands, and there are typically 15–18 competing proposals. If your radar shows you issues older than 72h, you're looking at scraps.
2. Scoring: reward × freshness ÷ crowd
Not all "bounty" issues are worth your attention. The ranking function I landed on after a few iterations:
function score(issue) { const rewardUsd = extractReward(issue); // from title/labels/body const ageHours = (Date.now() - new Date(issue.created_at)) / 3.6e6; const crowd = Math.max(issue.comments, 0); if (!rewardUsd || rewardUsd < 50) return 0; // below threshold if (ageHours > 72) return 0; // stale // Comments are a proxy for competition — penalize hard return rewardUsd * Math.exp(-ageHours / 24) / (1 + crowd * 0.5); } The exponential decay means a $250 issue that's 6 hours old outranks a $500 issue that's 3 days old. That matches reality: speed wins these races.
Extracting the reward is the annoying part — every platform formats it differently. Algora puts $150 in the title, Expensify uses [$250] title prefixes, Polar shows it in an embedded badge in the body. A few regexes handle 90% of cases.
3. Deduplication with staticData (the part everyone forgets)
If your poller runs every N hours, it'll re-alert the same issues forever. In n8n, getWorkflowStaticData() persists across executions and restarts:
const staticData = this.getWorkflowStaticData('global'); staticData.seen ??= {}; // Skip already-processed issues, clean entries older than 30 days const ttl = 30 * 24 * 3.6e6; for (const [id, ts] of Object.entries(staticData.seen)) { if (Date.now() - ts > ttl) delete staticData.seen[id]; } const fresh = issues.filter(i => !staticData.seen[i.id]); for (const i of fresh) staticData.seen[i.id] = Date.now(); Without the TTL cleanup the map grows unboundedly — learned that one the hard way.
4. Rate limits and failure handling
GitHub search gives you 30 requests/minute shared across all your workflows. Five sources polling every 6 hours is fine; polling every 10 minutes will get you 403s. Also: wrap the HTTP nodes in an Error Trigger — when a token expires or GitHub rate-limits you, you want a notification, not a silent dead pipeline. Route errors to a separate admin Telegram chat so real alerts don't drown in noise.
5. Output: Telegram for speed, Sheets for history
Markdown-formatted Telegram messages (*[$250] Issue title* — repo — 3h ago) for the ones you act on now, plus a Google Sheets append for the full feed. The sheet becomes genuinely useful after a few weeks: you can see which repos fund issues repeatedly, average rewards per platform, and your own hit rate.
If you'd rather not build this from scratch: I packaged the full thing — 16 nodes, all five sources, ROI scoring, dedup, error handling, dual Telegram/Sheets output — as an importable n8n workflow with a 10-minute setup guide. It's €29:
- Polar: https://buy.polar.sh/polar_cl_XKFsw2Nl5KY5LqqSYrDT9UR4P64t5jkuEpG5u40CByi
- Gumroad: https://tonypisa.gumroad.com
Either way, build the radar first — the bounty hunters winning $175-250 Expensify issues aren't faster coders, they're faster notified.
What are you monitoring that pays? Curious what other feeds people plug into n8n.


![Mobile Manipulation Software and Operations — Ch.13: OTA Updates [EN]](https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fb0t5n8yyl10kkhac8kc0.png)
![Mobile Manipulation Software and Operations — Ch.12: Safety Validation [EN]](https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fmync7cwwizyn31zfxaud.png)