Specor for Developers
Your AI IDE reads the spec before it writes code.
Specor exposes your product’s features, rules, flows, and acceptance criteria to Cursor, Claude, and Copilot over the open MCP standard — so the code your AI writes actually matches what the product is supposed to do.
Sound familiar?
You’re guessing at intent
The ticket says “add wishlist.” The rules, edge cases, and acceptance criteria live somewhere you can’t reach from your editor.
Docs are stale — or missing
By the time there’s a doc, the code has moved on. Onboarding to an unfamiliar area means archaeology.
Your AI tools work blind
Copilot and Cursor are brilliant, but they don’t know your product’s rules — so they invent plausible-but-wrong behaviour.
Specor becomes the live product brain your tools read from — before a single line is written.
Model Context Protocol
Live product context, right in your AI IDE
Point Cursor, Claude Desktop, or any MCP client at Specor and your assistant gains 20 read-only tools — get_product_context, get_rules, get_flow_steps, get_acceptance_criteria, and more.
Ask “implement guest checkout” and it already knows the rules, the flow, and the ACs. No copy-pasting specs into a prompt.
~/.cursor/mcp.json
{
"mcpServers": {
"specor": {
"url": "https://mcp.specor.ai/mcp",
"headers": {
"Authorization": "Bearer <token>",
"X-Workspace-ID": "<workspace>"
}
}
}
}Brownfield? Start from your code
Bootstrap a spec from your existing repo
Don’t have a spec yet? Connect a GitHub repository and Specor analyses your codebase to draft the product graph — features, capabilities, and flows — that you can refine instead of writing from scratch.
Your first source of truth is generated from the truth that already exists: the code.
Analysing repo…
src/checkout/ →
＋ Capability “Checkout”
＋ Feature “Guest checkout”
＋ Flow “Cart → Pay → Confirm”
12 features · 34 rules drafted for review
Code ↔ spec drift
Every PR checked against the definition
Specor watches your pull requests and flags when code drifts from the spec — a shipped behaviour that no feature describes, or a rule the diff appears to violate. It re-analyses automatically as you push new commits.
Drift is caught in review, not in production.
⚠︎ Spec drift detected
Rule “Verify email before first login” isn’t reflected by this change.
A day with Specor
- 1
Pick up a ticket. Ask Cursor to implement it — it pulls the feature’s rules and ACs from Specor over MCP.
- 2
The generated code respects the real edge cases instead of inventing them. You review, not rewrite.
- 3
Open a PR. Specor confirms the change matches the spec — or flags exactly where it drifted.
- 4
Onboarding to a new area next sprint? Your AI already has the whole product graph to explain it.
Work with other teams?
Specor is one shared source of truth — everyone reads the same product.
Point your AI tools at the source of truth
Start free — no credit card. Viewers are always free, so your whole team can read along.