Do your developers use AI coding assistants or agents on our code? Which ones?
Clients also ask it like this:
- Do your engineers use AI tools such as Copilot, Claude Code or Cursor on our project?
- List every AI coding tool or agent that touches our code or data.
- How do you know which AI tools your developers use?
What the client is checking
The client wants to know if AI tools touch their code, and if you know which ones. A vague answer tells them you have no list.
Three draft answers
Pick the level that is true for you today. Every statement marked [VERIFY]
must be true for your organisation before you send it. Delete a statement that is not true. Fill each
{{placeholder}}.
Not yet in place
Yes, some of our developers use AI coding tools. [VERIFY] We do not yet keep a complete list of the tools in use. [VERIFY] We will build the list by {{date}} and send it to you. [VERIFY]
Policy only
We keep a list of approved AI coding tools: {{tools}}. [VERIFY] Developers must use only tools on the list. [VERIFY] We build the list from what developers tell us. We do not check each machine. [VERIFY]
Enforced, with evidence
These AI coding tools are in use on your project: {{tools}}. [VERIFY] We compare this list with the tools found on developer machines every {{interval}}. [VERIFY] The last check was on {{date}}. [VERIFY] A tool that is not on the list is reviewed or removed. [VERIFY]
Evidence to attach for the strongest answer
- The tool register: tool, vendor, plan, owner, approval date.
- The newest scan or inventory output for the machines of the people on your project.
- A list of the people on the project, and which of their machines were scanned.
- Seat or admin screens of each tool, with the date.
How to move up one level
From not yet in place to policy only
Write the register. Ask every developer, and include contractors. Name an owner and set a review date.
From policy only to enforced, with evidence
Scan the machines with the tools you already have (device management, endpoint inventory, extension lists). Compare the scan with the register on a fixed schedule. Record each difference and what you did.
Where KYA fits
kya discover scans one workstation for AI tooling: command-line programs on PATH, IDE extensions, desktop apps, local model runtimes, and MCP servers named in config files. It writes a snapshot with a confidence level for each finding. It never runs the programs it finds and never connects to network ports.
It matches names from a built-in catalog of known tools. A tool that is not in the catalog is not found. It covers only the machines where someone runs it. It does not scan a network, and it does not find agents on machines that do not run it. --push sends the snapshot to a hosted console. Without --push the snapshot stays on the machine.
KYA does not tell you which tools are approved. The register and the approvals are yours.
Get all 13 answers in one document
Choose the level that is true for each group. You get a preview at once and the full pack by email.
These are drafts. Send only what is true for your organisation. This is not legal advice.