Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for technical writers

API Reference Drift Agent

Keep the API reference matched to the shipped API with working examples

Quality controlCreatingIT and developmentWriters
See how it worksGet access
API Reference Drift Agent: what goes in, what the agent does and what you get

What it does

Developers change endpoints faster than writers can track them. Readers then copy examples that fail. This agent compares the published API reference with the current API specification after every release build. It lists added, removed and changed endpoints, parameters, types and error codes. For each difference it drafts the reference update and a working request example. It then runs every example against a test environment; if a call fails or returns a different shape than documented, it fixes the draft and runs it again. Items it cannot explain, like a field in the spec with no description, go to the owning developer as a question. The writer reviews the drafts and decides on wording and publication. Edge case: a field marked deprecated stays documented with a removal note until the stated sunset date.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Release build with spec changes 2 USES A TOOL Read the new specification and the publishedreference 3 DOES List added, removed and changed endpoints, fieldsand errors 4 DOES Draft reference updates and request examples 5 USES A TOOL Run each example against the test environment 6 CHECKS THE RESULT Does every example succeed and match the documentedresponse? If not: fix the example or the description and runagain. Back to step 4. 7 CHECKS THE RESULT Does every new field have a description from thespec or the developer? If not: send a question to the code owner and mark theitem pending. Back to step 3. 8 YOU APPROVE Writer reviews and approves the reference changes 9 RESULT Updated API reference ready to publish
Read the steps as a list
  1. Release build with spec changes
  2. Read the new specification and the published reference
  3. List added, removed and changed endpoints, fields and errors
  4. Draft reference updates and request examples
  5. Run each example against the test environment
  6. Does every example succeed and match the documented response?If not: fix the example or the description and run again. Back to step 4.
  7. Does every new field have a description from the spec or the developer?If not: send a question to the code owner and mark the item pending. Back to step 3.
  8. Writer reviews and approves the reference changesThe agent waits here for your OK.
  9. Updated API reference ready to publish

How it decides

Every spec difference becomes a draft change; examples are kept only when they run successfully against the test environment.

  • Deprecated fields stay documented until the sunset date
  • Undocumented spec fields go to the code owner, never guessed
  • Breaking changes get a note at the top of the changelog page
  • Examples use test keys only

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Spec file location (default repository main branch)
  • Test environment URL (default staging)
  • Who answers field questions (default code owners file)
  • Changelog format (default newest first)

What keeps you in control

It always asks you first

  • Publishing reference changes

Hard limits

  • Never uses production API keys
  • Never publishes directly

It stops when

  • Done: all differences documented or pending with an owner
  • Stop: test environment down

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensRelease 4.12 added 2 parameters and removed the legacy_id field. The agent drafted updates and 5 examples. The list-orders example returned a next_cursor field not in the draft, so it added it and reran: success. The new region parameter had no description, so it asked the code owner, who replied the same day. The writer approved on 17 April.

More agents for technical writers

Compliance · Quality control

Editorial Revision Compliance Agent

A revision compliance map and response for the editor.

7 steps1 check7 jobs
Quality control · Writers

Article Structure Revision Agent

A tracked structural revision that keeps the argument fully supported.

7 steps1 check6 jobs
Compliance · Creatives

Copy Approval Chain Tracker Agent

One locked copy version with a complete record of who approved it

9 steps2 checks7 jobs
Quality control · Automation

Web Content Translation Sync Agent

Keep every translated page in sync with its source, with verified text and rendering.

10 steps2 checks7 jobs