EmailShield for AI agents

These docs are written to be read by people and by AI assistants. If you’re building an agent that cleans lists, or asking an assistant to help you integrate, point it here.

Plain-text versions

File	What’s in it
/llms.txt	Short index with the key facts and links
/llms-small.txt	The docs, compacted
/llms-full.txt	Every page in one file

Rules for an agent using the API

• Every verification spends a credit. Check GET /v1/account/credits before large jobs, and confirm with the person before spending a large share of their balance.
• Deduplicate and drop malformed addresses first. One malformed address fails the whole bulk call with 422.
• Batch. One POST /v1/verify/bulk with up to 50,000 addresses, not thousands of single calls.
• Poll gently (every 10–30 s) or use a webhook. Stop when terminal is true.
• Read details.primary_label and details.tags, not the per-row status, to decide what’s safe. Reading a result →
• Only valid is safe to send. Treat catch_all, unknown, not_verified and the role tag as risky; never mail invalid, disabled, spam_trap or disposable.
• Never guess a verdict for an address that came back unknown — re-verify it later instead.
• Keep the API key secret. Never print it, log it or put it in client-side code.
• On 429 wait for X-RateLimit-Reset; on 503 wait for Retry-After.