WILMINGTON, Del., August 9, 2026 — SpiderPublish, the content platform from SpiderIQ, has changed how its API answers automated callers, and documented the contract at https://publish.spideriq.ai/docs/api-reference/agent-guidance. A success response can now carry a guidance block stating what a resource is for, what it is not for, which call usually comes next, and the mistakes callers actually make. It is live for every customer and needs no migration.

The work began with an incident. An AI coding agent working inside a customer's development environment asked the API for the public URL of a web form and received a booking URL instead, according to SpiderIQ's own incident record (May 2026). The response was a valid success: correct status code, parseable body, a URL that resolved. Nothing signalled it was wrong for that case. It built eight pages around that URL and published them to a live website, where they showed visitors an error until the cause was traced.

"A human clicks the link, sees a broken page and loses a minute. An agent takes the answer and builds on it. That is not solved by making the API more accurate. It is solved by making it able to say what it is not for," said Martin Shein, Founder of SpiderIQ.

Four changes followed. Forms and booking flows now share one canonical address that reads the kind of flow behind it rather than assuming; the previous address permanently redirects, so existing links keep working. Error responses carry a structured envelope stating what the caller sent, what was expected, and a suggested URL. Success responses carry the guidance block, whose vocabulary is frozen at six keys so it stays machine-parseable. A verification tool opens a published page in a real browser and reports what a visitor actually sees, so an agent can check its own work.

Across two internal verification runs of the corrected tool, the number of incorrect URLs returned was zero. That figure was measured by SpiderIQ, not by an outside party, per its published verification record (20 May 2026, re-confirmed 9 August 2026); the method was a clean installation of the client package with a search for the superseded URL pattern, plus live requests against a production tenant. One run returned a partial result, for the reason below.

SpiderIQ also disclosed a failure in its own rollout. When the work was declared finished, the guidance block was active on one server process and absent from the one serving public traffic; the code was correct on both, and one middleware registration was missing. The gap stayed in production for eighteen days after the initiative was recorded complete, according to SpiderIQ's deployment history, and was closed by a follow-up fix.

"The first failure was a tool being confidently wrong. The second was me being confidently done. I verified the code and did not verify the surface a customer actually reaches," said Shein.

Guidance is on by default for API-token callers and can be switched off per request.

About SpiderPublish

SpiderPublish is an AI-native headless CMS from SpiderIQ, built multi-tenant and multi-project so an agency can use any agent runtime to build and run client sites: websites, funnels, interactive forms, blogs, API documentation with AI-RAG search, image and file hosting, a changelog and a newsroom, across one or many domains on a single deploy. It brings vibe publishing to any agency. Learn more at https://publish.spideriq.ai.

Media contact

Martin Shein Founder, SpiderIQ martin@spideriq.ai https://publish.spideriq.ai/press