Leadup Public API Finder

LeadUp Public API Finder

Purpose

Take a feature need and return a short, honest shortlist of free or public APIs that fit, with one clear recommended pick + one backup, verified from official documentation. Built for LeadUp projects: client websites, multi-tenant SaaS, Flutter apps, hosting / security tools, CRM, automation workflows, travel, jewellery, salon, school, and clinic products under *.leadup.in.

When to use

Use when the user wants to discover or choose a public/third-party API. Do not trigger when:

• They want the integration code itself written (route to leadup-api-research-builder, which does the deeper integration research with auth, webhooks, env mapping).
• They want a new SaaS feature planned (use leadup-feature-option-planner or leadup-ai-feature-planner).
• They want an n8n graph built (use leadup-n8n-workflow-builder).

Trigger phrases: "find API", "free API", "public API", "API options", "best API for this feature", "API idea", "third party API", "API marketplace", "check public APIs", "use public-apis repo", "API for my project", "free APIs for SaaS", "compare APIs".

Inputs needed

• The feature or job the API must do (e.g. "verify an email", "currency conversion INR↔USD", "send WhatsApp template", "detect malicious URLs").
• Project context (LeadUp website / Hostendor / Jewellery SaaS / Salon SaaS / INET CRM / Trivasia / Security suite / other).
• Audience and market (India / global), to weight UPI / GST / WhatsApp / language coverage.
• Where the API will run (browser-only, server-only, or both — CORS matters).
• Volume / rate expectations (calls per day, peak per minute).
• Cost ceiling (free only, free + paid OK, hard budget).
• Commercial use (yes — always, for LeadUp client work).
• Must-keep constraints (regulated industry, PII risk, regional data residency).

Ask at most 2 clarifying questions if the feature or runtime (browser vs server) is genuinely ambiguous.

Source/resource strategy

Use these sources, in this priority order:

1. Official API documentation of each candidate (the source of truth for auth, limits, pricing, commercial use, CORS, SDKs).
2. public-apis/public-apis on GitHub (https://github.com/public-apis/public-apis) — discovery only. It is a curated index; entries can be stale or incomplete.
3. Provider pricing pages for the current free-tier limits and commercial-use language.
4. GitHub examples (sample code, popular client libraries, recent stars/commits) — to gauge real-world use.
5. RapidAPI marketplace — for aggregated metadata, latency stats, reviews, and a single billing layer when useful.
6. ApyHub — for utility APIs (PDF, image, validation) when free alternatives are limited.
7. Postman public collections / public workspaces — for ready request examples and quickstart.
8. SDK / package docs (npm / pypi / pub.dev / packagist) — for integration ergonomics.
9. API status pages (e.g. status.<provider>.com) — for uptime / incident history.

Detail and notes per source: references/public-apis-repo-notes.md.

API discovery workflow

1. Restate the requirement in one line (what the API must do, for whom, where it runs).
2. Classify the API category (validation / payments / messaging / maps / weather / docs / security / AI / data / search / files).
3. Scan public-apis/public-apis for that category; capture 3–8 raw candidates (name + 1-line desc + auth + HTTPS + CORS hint).
4. Cross-check with RapidAPI, ApyHub, and an open Google search to surface providers not listed in public-apis.
5. Drop obviously bad fits (wrong category, dead links, paid-only when user said free).
6. Shortlist 3–5 for verification.

Discovery framework + heuristics: references/public-api-discovery-framework.md.

API verification workflow

For each shortlisted candidate, verify from the official docs (not the directory) and record:

• Auth method (none / API key / OAuth2 / JWT / mTLS).
• HTTPS (must be yes).
• CORS (yes / no / partial / unknown — drives backend proxy need).
• Free tier (calls / month, daily, per minute; expiry of any trial).
• Paid pricing (next tier band; per-call vs subscription).
• Rate limits (per minute, per day, burst behavior).
• Commercial use (allowed / restricted / attribution required).
• Terms / data residency (where data is processed).
• SDK availability (official npm / pip / dart / etc.).
• Webhook support (events, retries, signature).
• Privacy risk (does it see PII / payments / health / kids?).
• Backend proxy required? (browser-CORS missing or API key would leak in the client).
• Reliability / maintenance (last release, status page, GitHub activity, incident history).
• Region coverage (India / EU / US / global) and language support if relevant.

Checklist with what to read on the docs page: references/api-verification-checklist.md.

API scoring framework

Score each verified candidate 1–5 on the eight axes below. Average for a quick rank, and call out any single sub-3 score as a red flag.

Full rubric and tie-breakers: references/api-scoring-framework.md.

Step-by-step workflow

1. Restate the requirement in one line; confirm runtime + market + commercial use.
2. Discover candidates per "API discovery workflow" above.
3. Verify each shortlisted candidate per "API verification workflow" against its official docs.
4. Score with the 8-axis rubric; mark sub-3 axes as risks.
5. Pick one best + one backup. Justify in 2–3 lines.
6. Spec the integration: env placeholders, backend-vs-frontend placement, error handling, rate-limit strategy, retry policy, logging without PII.
7. Test plan: 3–5 fixtures (happy path, auth fail, rate-limited, downstream 5xx, malformed payload).
8. Output the structured report (see Required output format).
9. Hand off:
   • Implementation depth → leadup-api-research-builder.
   • Workflow wiring → leadup-n8n-workflow-builder.
   • PII / payment scope → leadup-pii-risk-reviewer and leadup-security-review.
   • Public copy mentioning the integration → leadup-human-content-editor.

Required output format

One Markdown report with these sections, in this order:

1. Requirement summary — feature, runtime, market, commercial.
2. API category — single tag (validation / payments / messaging / maps / weather / docs / security / AI / data / search / files / other).
3. Candidate APIs (3–5) — table with: name · auth · HTTPS · CORS · free tier band · pricing band · region · maintained? · score.
4. Best API recommendation — one name + one-line reason.
5. Backup API recommendation — one name + one-line reason.
6. Why selected — 3–5 bullets weighting the 8 scoring axes.
7. Official docs / resource links — direct links to docs, pricing, status page, SDK page; cite access date.
8. Required env variables — .env.example placeholders only.
9. Backend / frontend integration notes — where it runs, why; CORS handling; secret handling; quick code shape.
10. Security / privacy risks — PII categories that touch the API; DPA status if known; route to leadup-pii-risk-reviewer / leadup-security-review flags.
11. Rate limit / pricing notes — bands, what triggers a paid upgrade, caps the team should set in code.
12. Implementation plan — concrete steps the dev follows.
13. Test plan — 3–5 fixtures with expected outcomes.
14. Decision — use now / prototype only / avoid + one line reason.
15. Hand-offs — to other LeadUp skills.

Templates: assets/api-comparison.template.md, assets/api-integration-plan.template.md, assets/env-placeholders.template.md, assets/api-test-plan.template.md.

Safety rules

• Treat public-apis/public-apis as a discovery index only. Entries can be stale; never quote pricing, CORS, or auth from the directory in the final output. Always re-read the official docs.
• Do not ask the user to paste live API keys, tokens, or .env values. Use .env.example placeholders like __SET_ME__ / {{API_KEY}}.
• Do not put API keys in browser-shipped code. If the API is browser-friendly but the key is secret, recommend a backend proxy.
• Do not recommend an API for production if last release / commit is > 24 months old without an explicit "use with warning" note.
• Do not recommend an API that lacks HTTPS for production.
• Do not ignore commercial-use clauses. Free for personal / research is not free for LeadUp client work.
• For India / EU users with PII: confirm data residency and DPA; route to leadup-pii-risk-reviewer.
• For payment-related APIs: route to leadup-security-review and prefer Razorpay / Stripe / Cashfree per leadup-saas-mvp-planner defaults.
• For WhatsApp APIs: route to leadup-whatsapp-automation-planner (opt-in + template categories + BSP choice).
• For AI / LLM APIs: route to leadup-ai-feature-planner (cost ledger, fallback, redaction).
• Do not invent rate limits, prices, or CORS behavior. If unclear in the official docs, label requires verification and link the doc page.

Common mistakes

• Quoting the directory's metadata as truth.
• Picking the API with the loudest landing page over the one with cleanest docs.
• Ignoring CORS until launch day.
• Putting a paid API key in a frontend bundle.
• Choosing "free" without reading commercial-use language.
• One mega table with 15 candidates and no recommendation.
• Forgetting backup; the best API will go down eventually.
• No rate-limit strategy → 429 storms in production.
• No test fixtures → "it worked once in Postman" is not proof.

Troubleshooting

• No browser / search MCP available: deliver the report from known docs + public-apis notes; mark anything not directly verifiable as requires verification and include the exact URL to confirm.
• Highly regulated category: cap the recommendation; flag every PII or financial signal; route to leadup-pii-risk-reviewer and leadup-security-review before committing.
• India-first audience: prefer providers with INR billing, India data residency, GST invoices, and India-region latency.
• Browser-only feature with secret key: switch to a backend proxy plan in section 9; never expose secret keys in client code.
• All free tiers are too small: list the cheapest paid tier band per candidate and recommend a feature flag to gate usage.
• Provider is unmaintained but unique: deliver as "prototype only", not "use now"; plan migration path to an active alternative.
• User wants instant pick: deliver just sections 1, 3, 4, 5, 14; hold the rest for an iteration.

Test prompts

Should trigger (5)

1. "Find a free public API for email validation."
2. "What public API should we use for currency conversion in our jewellery SaaS?"
3. "Best API for IP lookup / threat reputation for Hostendor?"
4. "Free APIs for weather + flight info for Trivasia."
5. "Check public-apis repo and recommend a maps API for our salon SaaS."

Should NOT trigger (3)

1. "Write the Razorpay integration code." (→ leadup-api-research-builder)
2. "Plan an AI feature using Claude." (→ leadup-ai-feature-planner)
3. "Build an n8n graph that calls this API." (→ leadup-n8n-workflow-builder)

Functional test cases (2)

1. Given "email validation API for our lead form, used server-side, India audience, commercial LeadUp client use", return a report with 3–5 candidate APIs, 8-axis scores, a best + backup pick with docs links + access date, env placeholders, backend-vs-frontend note (server-side here), rate-limit and pricing bands, an implementation plan, a test plan with 4 fixtures, and a final use now / prototype only / avoid decision.
2. Given a browser-only feature ("currency conversion in a Flutter web app") with a secret API key, return a report that explicitly recommends a backend proxy (or a key-less public source like the ECB / exchangerate.host), forbids embedding the secret in the client, and provides a candidate alternative.

Success criteria

• All 15 output sections present in order.
• Candidates verified from official docs, with a cited access date — public-apis/public-apis used only for discovery.
• 8-axis scores per candidate; sub-3 axes flagged.
• One best + one backup, with 3–5 bullet justification.
• Env shown as placeholders only; no real keys anywhere.
• CORS / backend-proxy decision is explicit.
• Commercial-use safety is confirmed for LeadUp client work.
• Rate-limit and pricing labelled verified vs requires verification, with links.
• Decision: use now / prototype only / avoid.
• Hand-offs to leadup-api-research-builder, leadup-n8n-workflow-builder, leadup-pii-risk-reviewer, leadup-security-review, leadup-whatsapp-automation-planner, and leadup-ai-feature-planner are explicit when relevant.