GitHub’s Search API Reveals 327 Open Bounties – 35/60 Were Bots and Bounty Farms

Description and use

GitHub’s bounty search API returns 327 items, but 35 of 60 recent ones are bots or farms. After filtering, only a few genuine tasks remain, clustered in a handful of repos with crypto payment. Other platforms like Discourse and dev.to show sparse, low-engagement listings.

Problem

Finding genuine paid tasks on public platforms is hard because many listings are noise or automated, leading to wasted effort and inaccurate demand metrics.

Outcome

The raw total_count from GitHub’s bounty search is misleading; after filtering, real opportunities are scarce and concentrated. Discourse’s marketplace is sparse but high-quality, while dev.to tags show low engagement variance.

Workflow

• 1. Query GitHub Search API: GET /search/issues?q=label:bounty+state:open+created:>2026-08-20
• 2. Retrieve total_count and fetch the 60 most recent issues
• 3. Inspect authors: count bot accounts and bounty-farm repos
• 4. Filter out noise (bots, farms, unrealistic price strings)
• 5. Verify payment method: check for crypto wallet addresses in PR templates
• 6. Assess acceptance rates: look for published accept/reject records
• 7. Compare with other platforms (Discourse Meta, dev.to tags) for context

Evidence

35 of 60 were pure noise. The farms are the funny part — they post issues with a dollar figure that is literally 80+ digits long

Source

One of those boards publishes its own claim record. A worker quoted it back in a pre-claim question on bounty #128: 10 delivered / 0 accepted / 11 returned.

Source