Last updated: 2026-07-25

Privacy

Clanker Banker processes public web content on behalf of autonomous agents. This policy is written for developers and security reviewers.

1. What we process

Request inputs: public http(s) URLs you send, optional maxTokens, noCache.
Fetched content: HTML/text retrieved from that URL to produce clean/pack/extract output.
Payment metadata: x402 payment proofs / settlement fields required to unlock paid tools (not your seed phrase).
Operational metrics: counts of paid calls, 402s, tool name, timestamps (dashboard).

2. Caching (default) & zero-retention mode

Mode	Behavior
Default	Successful packs may be stored in Cloudflare KV for a short TTL (config `CACHE_TTL_SECONDS`, often ≤ 1 hour) to reduce cost/latency for identical public URLs.
No-cache / zero retention for that request	Send `"noCache": true` in JSON body. We skip KV read/write for that call. Response header `X-Clanker-Cache: BYPASS`.

POST https://clanker-banker.agentfetcher.workers.dev/v1/llm-pack
{ "url": "https://example.com", "maxTokens": 4000, "noCache": true }

Private/localhost/metadata IPs are rejected before fetch.

3. What we do not do

We do not require human accounts or email lists for agents.
We do not sell packed page content as a training corpus.
We do not ask for private keys or seed phrases (agents use their own wallets for x402).
We do not intentionally fetch non-public networks (SSRF protections).

4. Subprocessors / infrastructure

Cloudflare — Workers compute, KV cache, edge network
Coinbase CDP — x402 facilitator / settlement (when configured)
Origin websites — content fetched at your request

5. Contact

Operator email is listed in OpenAPI info.contact / env CONTACT_EMAIL. For deletion of cached public URLs, contact the operator with the exact URL keys; TTL expiry also purges automatically.