Connect Claude to Stori

Stori exposes your workspace to AI assistants through the Model Context Protocol (MCP) — the open standard used by claude.ai, Claude Desktop, ChatGPT and a growing set of AI tools. Once connected, you can ask questions in plain language and act on the answers:

• “Which of my jobs has the most interviewed candidates?”
• “On the Product Designer role, who has actually rebuilt a design system? Quote what they said.”
• “Compare Kevin Lin and Clare Woods on the evidence we hold.”
• “Add a note on Kevin Lin and move him to the final stage.”
• “What did the assistant change today?”

Every answer is grounded in your candidates' actual interview transcripts. When Claude quotes a candidate, that quote is real, cited, and carries a video timestamp so you can verify it in one click. Ask about what a candidate did, not which competency they show: a Stori interview records what happened in the candidate's own account, which is why the answer can be quoted.

1. Open claude.ai → Settings → Connectors (or Claude Desktop → Settings → Connectors) and choose Add custom connector.
2. Paste this URL: https://www.onestori.com/api/mcp
3. Claude will send you to Stori's sign-in. Sign in with your workspace (employer) account — the same one you use for the Stori dashboard.
4. Review the consent screen and click Allow. You're connected — start a new chat and ask away.

A connected assistant can read:

• Your jobs and their candidate counts, and any role in full
• A job's pool, or a person found by name or email across every role
• Your candidates' interview transcripts, with quotes and timestamps
• A candidate's report, their Stori record, and the interview transcript verbatim
• Candidates side by side, and a pool ordered by strength of evidence
• What interviews are booked, who can sit on a panel, and where offers stand
• The pipeline across roles, and your credit balances

And, within the permissions your workspace has switched on, it can act:

• Add notes, move candidates between stages, and reject a candidate with a written reason
• Invite a candidate to interview and schedule the interview loop
• Draft an offer against the role's salary band and send it round the approval chain
• Source candidates from outside your pool, reveal a sourced person's contact details, and draft or send outreach

Every change is written to your audit log against the person whose connection made it, so “what did the assistant change today?” has an exact answer.

It structurally cannot:

• See any other company's data — access is locked to your workspace at the moment you click Allow
• Send an offer document to a candidate — that stays a deliberate act in the app
• Read compensation from an offer list — figures appear only when you ask about one offer case
• Change workspace settings or create API keys — the admin capability is not available to sign-in connections
• Act without a signed-in member of your team having approved the connection

Under the hood: OAuth 2.1 with PKCE, short-lived rotating tokens stored only as hashes, and per-workspace isolation enforced on every request.

Workspace admins set a ceiling on what any connected app may do in Settings → Integrations → AI Apps. It sits above whatever a member approved when they connected: a tool whose capability is switched off is refused, and the assistant is told which switch to ask an admin about.

A new workspace starts with reading and record changes switched on. External sourcing, revealing contact details, drafting outreach, sending outreach and interview invitations are each off until an admin turns them on, because each one reaches a person outside the workspace or spends credits.

Every active connection is listed in Settings → Integrations → AI Apps: which app, who connected it, when it was last used, and what it was approved for. Any member can disconnect their own connection; workspace admins can disconnect anyone's. Disconnecting takes effect immediately — the app's tokens are revoked and its next request is refused.