AI connectors · MCP

Ask Claude, ChatGPT or Copilot about your defects.

Connect IssuesId to the AI app your team already uses and ask about projects, defects and how-tos in plain words. Answers come back with the reference, the location, the status and a link to the record. It sees what you see and changes nothing.

What you can ask

The questions you’d otherwise ring the site office for

The AI app picks the right tool for the question, and can chain them: look up “Apex”, narrow to your project, then count by level. Every answer carries links back into IssuesId, so the register is still where you act on it.

Find defects
“Find the cracked tile at the lift lobby on level 2.”
Up to five matching defects with the reference, description, location, status and a link to each, plus how many matched in total.
Count or list defects
“How many open defects are issued to Apex Plumbing?”
A count, or the first rows as a table, with a link to the full filtered list in IssuesId.
Count defects by group
“Break down Riverside by status, then by level.”
One row per status, Issued To company, classification or level, each with its count and a link to that group.
List projects
“Which projects am I on?”
Your active projects, so the app can narrow every other question to the right one.
Look up a name
“Apex” · “level 3” · “Sam from the builder”
Turns a project, location, company, person or classification into the exact record, and asks you to pick when two fit.
Search the help
“How do I send a trade a link to their list?”
The matching help articles with links, the steps, and whether your role can do it.
Who it acts as

Your login, your fences, reads first

The record is evidence, so the connector reads it, and the only thing it can add is a note you allowed, marked as written by the app. It is the same boundary roles and dashboards already draw, reached from a different window. More on how records are kept on the security page.

It acts as you
Every question runs with your role, your permissions and your project fences, through the same checks as your own login. A manager on two projects gets answers from those two.
It reads, unless you say otherwise
Out of the box it only reads. Tick “Also let it add notes” when you connect and it can add a note for you, shown as “via Claude”, internal unless you ask it to share. It cannot change a status, reassign a defect, or edit or delete anything.
Checked on every call
Each request re-checks that your account is active, you are still in the company and connecting is still on. Remove someone and their apps stop on the next question.
Logged without the data
Each call leaves an audit row: which app, which tool, which fields it filtered on and whether it worked. The values and the answers are not kept.
Yours to disconnect
Your Profile lists every app you have allowed, when it was connected and when it was last used. Disconnect ends its access straight away.
Nothing reusable stored
Sign-in codes and tokens are stored only as one-way hashes, refresh tokens work once, and replaying one cuts that app off.
How to connect

Five minutes, once per person

  1. 01
    Check you can connect
    Company Admins, Managers and Users can connect, and it is on for every company. If the card on your Profile says connecting is not turned on, your company asked us to turn it off: your company admin can ask us to turn it back on.
  2. 02
    Copy the address
    In IssuesId, open your Profile. The Connected AI apps card has the address and a copy button. It is the same for everyone:https://api.issuesid.com/mcp
  3. 03
    Add it to your AI app
    Each app calls it something different: a custom connector, an app, an MCP tool. The steps for each are below.
  4. 04
    Sign in and allow
    The app sends you to IssuesId. Sign in, and the consent screen names the app, the account and company it will act as, and where it sends you back to. Tick “Also let it add notes” only if you want it to write notes for you. Only allow an app you started connecting yourself.
  5. 05
    Ask
    Start with “Which projects am I on?” to check it is connected, then ask the question you came with.
Claude
Free (one custom connector), Pro, Max, Team and Enterprise
  1. Open Customize → Connectors, choose + Add, then Add custom connector.
  2. Name it IssuesId, paste the address and choose Continue.
  3. If Claude asks how to register, choose Register automatically. Then Add.
  4. Choose Connect, sign in to IssuesId and choose Allow.
On Team and Enterprise an Owner adds it once, under Organization settings → Connectors → Add → Custom → Web. Everyone else then finds it in Customize → Connectors and chooses Connect.
Anthropic’s guide ↗
ChatGPT
Plans with developer mode; on Business and Enterprise your workspace admin decides
  1. In Settings, turn on Developer mode.
  2. Create a new app or connector. Name it IssuesId, paste the address and leave authentication on OAuth.
  3. Sign in to IssuesId when it opens and choose Allow.
  4. In a chat, pick IssuesId from the developer mode tools in the composer, then ask.
OpenAI moves these menus more often than the others, so if a label here does not match what you see, their help article has the current path.
OpenAI’s guide ↗
Microsoft 365 Copilot
Through an agent built in Copilot Studio
  1. In your agent, open Tools → Add a tool → New tool → Model Context Protocol.
  2. Server name IssuesId, a one-line description of what it holds, and the address as the Server URL.
  3. Authentication OAuth 2.0, type Dynamic discovery. Choose Create, then Next.
  4. Choose Create a new connection, sign in to IssuesId, choose Allow, then Add to agent.
Publish the agent to Microsoft 365 Copilot or Teams as you would any other. Each person who uses it signs in to IssuesId as themselves the first time.
Microsoft’s guide ↗
Frequently asked

AI connector questions

What is MCP?
The Model Context Protocol is the open standard Claude, ChatGPT and Microsoft Copilot use to reach tools and data outside the chat. IssuesId runs an MCP server, so any of them can be pointed at it with one address and a sign-in, without a plug-in or an API key.
Can the AI app change anything in IssuesId?
Only if you allow it, and only notes. Out of the box the connection reads: it can find, count, group and list defects, list your projects and report types, look up names and search the help. If you tick “Also let it add notes” when you connect, it can also add a note to a defect for you, internal unless you ask it to share, and marked in the defect’s activity as written via that app. It cannot change a status, reassign a defect, or edit or delete anything.
What can it see?
Exactly what you can see when you log in. It runs as you, with your role, your permissions and your project fences, through the same checks as your browser. Answers include the defect reference, description, location, status and a link that opens the record in IssuesId.
Who can connect an app?
Company Admins, Managers and Users, in every company. A company that would rather its people did not connect AI apps can ask us to turn it off.
Where does the data go?
The results of each question go to the AI app that asked, under your agreement with that provider, the same as anything else you paste into it. On our side, each call is logged by app, tool and outcome, without the values or the answers, and sign-in tokens are kept only as one-way hashes.
How do I take access away?
Each person can disconnect any app from the Connected AI apps card on their Profile, and it stops straight away. Deactivate someone or remove them from the company and every app they connected stops on its next question. If the company switch is turned off, every app in the company stops.
Already on IssuesId?
Open your Profile, copy the address and connect. Not yet? Bring the question you would ask it to a demo.
IssuesId