Verified Open Software Bounties & Buyer Intent

Find fresh, evidence-backed public software bounties and A/B buyer demand. The default mode returns open software bounties with visible spend evidence. Optional filters target n8n, Make.com, CRM/API integration, AI agents, Vapi/Twilio, PDF/OCR, and browser/data automation.

What counts as verified demand

• A — explicit spend event: a public, open request with both a real implementation need and visible budget, bounty, reward, tender, or other spend evidence.
• B — explicit demand: a buyer publicly asks for a developer, consultant, contractor, vendor, or implementation help, but the budget is not public.
• C/D are excluded: funding, hiring growth, generic news, product interest, market discussion, and guessed needs are never returned as buyer intent.

Every billed item has a source URL, an evidence summary, a public contact path, a publication timestamp, and last_verified. If an item cannot be verified, it is not output and is not billed.

Verification means the public source was reachable, the request was open at collection time, and the source contained an explicit spend claim. It does not independently guarantee escrow, payment, buyer identity, or future availability. Users must review the source and platform terms before doing work.

Input

{
  "capability": "Any software bounty",
  "industry": "Any",
  "country": "Any",
  "max_age_hours": 168,
  "max_results": 20
}

Use Any for industry or country when the public source does not state that attribute. A restrictive value intentionally produces zero results rather than guessing.

Output

Each dataset item includes buyer identity/type, need, source and evidence, timestamps, budget and deadline when public, contact path, A/B class, buying stage, intent/freshness/actionability/confidence scores, verification status, and a stable demand ID for deduplication.

Pricing

The primary PPE event is verified-demand: $0.02 per newly delivered verified dataset item. One event maps to one visible item. Zero new verified items means zero verified-demand charges. When all opportunities were already delivered, the Actor returns one transparent delivery_status: "previously_seen" fallback through the low-cost default dataset event. When no opportunity currently matches, it returns a record_type: "status" item explaining that result. Both fallback types have billable_verified_demand: false; neither is charged as a new verified demand. The standard apify-actor-start event remains at Apify's recommended $0.00005. Platform usage is included so cost is predictable and the Actor can qualify for agentic payments.

Data-source and access boundaries

V1 uses only public, unauthenticated sources intended for public access: official Algora project bounty pages cross-checked against GitHub's public Issues API, GitHub's public Issues Search API with repository-reputation verification, and the Hacker News public Algolia API. GitHub recall covers both bounty and reward labels in the default mode; structured cash templates, including CNY amounts, are normalized, and known elapsed deadlines are rejected. It does not bypass logins, CAPTCHAs, paywalls, robots rules, or access controls. It does not scrape or redistribute Upwork/Freelancer listings. Promotional posting, backlink, spam, unfunded, precommit, testnet-token, controlled demo/rehearsal, personal-experience review, and external-content publication tasks are excluded. Partial source failures are surfaced in the audit output instead of being silently treated as a clean scan. Source content remains subject to its original publisher's terms; the Actor returns short evidence summaries and direct source links, not copied pages.

Reliability and limits

The Actor caps results at 50, memory at 256 MB, request timeouts, and retries a rate-limited source once. It uses no paid LLM or external paid API. If all sources are unreachable, the run fails before emitting or billing unverifiable items. Demand IDs and persistent seen-state prevent the same source demand from being billed repeatedly; any previously seen fallback is explicitly labeled and uses the default dataset event rather than verified-demand.

Repeatable opportunity audit

Run npm run scan to execute the same strict public-source collection and classification path without emitting or charging Actor dataset events. The command prints a versioned JSON report with its timestamp, normalized input, collected/accepted counts, rejection reasons, source errors, and only the minimal fields needed to audit accepted opportunities. Each run also appends the report to work/opportunity-scans/history.jsonl and atomically refreshes latest.json. Records form a SHA-256 chain; the scanner verifies the complete prior chain before appending, so modified or truncated history is detected rather than silently extended. Optional flags are -- --max-age-hours=2160 --max-results=20, --audit-dir=/absolute/path, and --no-audit.