Proof Layer
Attach cryptographically signed, hash-chained receipts to agent tool calls so you can prove what an AI did in production or during review.
Overview
Proof Layer is an MCP server for the Ship phase that issues signed, hash-chained provenance receipts for AI agent actions through the Proof Layer API.
What is this MCP server?
- Emits signed, hash-chained provenance receipts for AI agent actions via Proof Layer API
- Stdio MCP package @proof-layer/mcp (v1.0.1) for Claude Code and compatible hosts
- Configurable API URL and optional telemetry off via PROOF_LAYER_TELEMETRY
- Dashboard API keys at prooflayer.world999labs.com for fleet-wide receipt policy
- Package @proof-layer/mcp at registry version 1.0.1
- Default API base https://prooflayer.world999labs.com
- Stdio transport; PROOF_LAYER_API_KEY required
What problem does it solve?
When agents change data or call external APIs, you cannot easily show a third party—or your future self—exactly what happened in order.
Who is it for?
Indie builders launching agent-backed SaaS who need a hosted audit trail without building blockchain or custom logging infrastructure.
Skip if: Teams that only need local debug logs in dev, or products with no requirement to prove agent behavior to anyone outside the repo.
What do I get? / Deliverables
After install, agent sessions can leave verifiable receipt chains you can reference during incident review, customer disputes, or security sign-off.
- Signed provenance receipts linked to agent actions in Proof Layer
- Hash-chained audit sequence retrievable via the hosted API
- Agent-configured stdio MCP entry in your catalog
Recommended MCP Servers
Journey fit
How it compares
Hosted provenance MCP integration, not an in-editor agent skill or generic observability dashboard.
Common Questions / FAQ
Who is Proof Layer MCP for?
Solo and small-team builders running Claude Code or similar agents in production who want signed receipts instead of unstructured chat history.
When should I use Proof Layer MCP?
Use it during Ship when you are adding security and accountability around agent tool use before or right after launch.
How do I add Proof Layer MCP to my agent?
Install @proof-layer/mcp, set PROOF_LAYER_API_KEY (and optional PROOF_LAYER_API_URL), register the stdio server in Claude Code or Cursor, then run agents as usual.