verify-after-mcp

Verifies GitHub and Linear MCP actions by generating read-only receipts and verdicts.

Updated Jun 20, 2026
One-click install
npx skills add https://github.com/Noetfield-Systems/SourceA --skill verify-after-mcp-noetfield-systems
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: verify-after-mcp
Source: https://github.com/Noetfield-Systems/SourceA/tree/main/cursor-plugin/sourcea-forge-governance/skills/verify-after-mcp
Command: npx skills add https://github.com/Noetfield-Systems/SourceA --skill verify-after-mcp-noetfield-systems

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? After an AI agent performs actions through GitHub or Linear MCP tools, there is no built-in proof that the action actually happened or complied with policy. This Skill closes that gap by running a verification pass that produces an auditable receipt and an honest PASS, FAIL, or MOCK_ONLY verdict. ## Core Features & Use Cases - Post-action verification: After any GitHub MCP call (PR, issue, comment) or Linear MCP call (issue create/update), invoke verify_run with a receipt or witness object to confirm the action. - Read-only receipt emission: Use emit_receipt_readonly to schema-validate draft receipts without writing anything to local disk. - Honest labeling: Distinguish MOCK_ONLY (local stdio tier, no cloud federation) from PASS and FAIL so claims about shipped work stay accurate. - Use Case: After creating a pull request via GitHub MCP, run the verify chain and reply to the founder with the verdict plus a one-line proof summary before claiming the work is complete. ## Quick Start After the GitHub MCP creates a pull request, run the sourcea-verify verify_run tool with the receipt and report the verdict.

Frequently Asked Questions about verify-after-mcp

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify a GitHub MCP action after creating a pull request?

Build a witness object containing the tool name, an args summary, and a timestamp, then call verify_run with that receipt or an existing receipt id. The tool returns a PASS, FAIL, or MOCK_ONLY verdict plus receipt JSON you can quote in your reply.

How to audit Linear MCP issue updates for compliance?

Run verify_run after any Linear issue create or update call, passing the receipt id or an inline receipt. The verdict confirms whether the action's receipt matches schema and policy before you report completion.

What does a MOCK_ONLY verdict mean in receipt verification?

MOCK_ONLY means the run executed on the local stdio tier and the cloud receipt bucket is not federated yet. You should state this honestly and avoid claiming cloud-backed proof for that action.

Does emit_receipt_readonly write receipts to local disk?

No, emit_receipt_readonly performs schema validation only and never writes to local disk. The Skill's rules also prohibit writes to ~/.sina or founder secrets through MCP tools.

When should I run verification before claiming work is shipped?

Run emit_receipt_readonly on the draft receipt before declaring a ship complete, and run the full verify chain whenever the user asks for proof, a receipt, or an audit. Chat messages are not the source of truth; the receipt JSON is.