- VALIDATED
- AUDITED
- HUMAN
- CONFIRMED
Short answer
How can I be sure my AI actually ran the tests and changed the code?
Your AI uses the verify_delivery tool to pull actual execution logs and diffs. You get hard evidence of what changed and whether the tests passed. You'll see the specific outcomes below.
The verification outcomes.
Where the proof lands.
The AI translates raw data into clear status updates.
VALIDATED
Verified test results
The AI pulls the actual output from your test runner. You see the pass or fail status directly.
AUDITED
Actual code diffs
The agent provides the specific lines changed in the repository. You see exactly what was added or removed.
HUMAN
Manual review needed
The AI flags discrepancies between the request and the actual output. You decide if the logic is correct.
CONFIRMED
Completion proof
The agent matches the task requirements against the verified logs. You get a clear signal that the work is done.
The workflow
What your AI does when the task arrives.
The AI stops talking and starts checking.
Analyze requirements
Your agent looks at the original prompt to understand what the code was supposed to do.
verify_deliveryFetch execution data
The AI pulls the raw logs and test outputs to see what actually happened during the run.
verify_deliveryCompare changes
The agent inspects the repository to find the specific diffs created by the code changes.
verify_deliveryGenerate proof
The AI compiles the logs and diffs into a report that proves the task is complete.
verify_delivery
Try it
Copy these to start.
Use these prompts to trigger the verification process.
Starting points
These are opening lines. Swap out the repo names or file paths for your specific project.
Delivery Integrity Prover Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_47XZJ7Jwbw4wbyoAk0N57vFtYyOnkyNmEkgNiNe8/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Delivery Integrity Prover capabilities are ready to use.
{
"mcpServers": {
"delivery-integrity-prover-mcp": {
"url": "https://edge.vinkius.com/vk_preview_47XZJ7Jwbw4wbyoAk0N57vFtYyOnkyNmEkgNiNe8/mcp"
}
}
}Verify the code changes in the current repo and show me the test results.
Check the output of the last test run in the main branch to confirm it passed.
Show me the exact diff for the changes you just made to the auth module.
Did the tests actually pass for the recent commit in the backend folder?
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Delivery Integrity Prover once, then ask.
One click connects this MCP to your client. Your credentials stay encrypted and the setup is handled on our end.
Connect Delivery Integrity Prover to your AIFAQ
How the verification behaves
- 01
Can the AI change my code during verification?
No. The verification process is read-only. The AI only inspects logs and diffs to prove what has already happened.
- 02
Does the AI delete any of my test files?
No. This MCP does not have tools to delete or modify your files.
- 03
How does it know the tests passed?
It uses the verify_delivery tool to read the actual exit codes and output from your test runner.
- 04
Can I use this with Cursor?
Yes, any MCP-compatible client like Cursor, Claude, or Windsurf works with this.
- 05
What if the test logs are huge?
The AI reads the relevant parts of the logs to find the specific failures or successes needed to answer your question.
More questions about Delivery Integrity Prover? The Connector page answers them. See everything the Delivery Integrity Prover Connector can do
Connect Delivery Integrity Prover to Claude, Cursor, ChatGPT & more
