mpps.io
Tamper-proof receipts for AI agent work.
Hash the artifact or action manifest. Get an HSM-signed receipt. Verify it later.
No API key. Hash-only. Stored for 10 years.
Agents produce work faster than organizations can audit it.
An agent writes a patch, generates a report, builds a dataset, calls an API, or publishes a package. A week later someone asks what was produced, which inputs were attached, and whether the record changed.
Logs are local. Chat transcripts are noisy. CI artifacts expire. mpps.io gives the workflow a small external evidence anchor: a signed receipt for the hash of what happened.
A receipt layer, not another orchestration stack.
mpps.io complements agent traces, build logs, package provenance, and content credentials. It does not replace them. It gives those systems a compact, independently verifiable receipt.
Create a structured agent action receipt.
The API is live. No registration, no token, no SDK required.
curl -X POST https://api.mpps.io/v1/receipts \ -H "Content-Type: application/json" \ -d '{ "action": "agent.task.complete", "subject": "final-report.md", "artifact_hashes": [ { "label": "final-report.md", "sha256": "sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" } ], "context": { "runner": "codex", "repo": "gdlg-ai/example", "commit": "abc123" } }'
{
"uuid": "mpps_att_8e2f4a1b3c5d4e6f",
"receipt_type": "agent_action",
"manifest_hash": "sha256:...",
"signature": "<base64>",
"verify_url": "https://api.mpps.io/v1/verify/mpps_att_..."
}
Receipts for work that should not rely on memory.
Attach a receipt to the final artifact of a code change, analysis, report, image, video, or dataset.
Anchor build outputs, package manifests, deployment bundles, and verification material.
Record what was sent or received after API calls, data transfers, paid workflows, or handoffs.
Precise proof, deliberately narrow claims.
agent_id is useful correlation from network source data, not strong identity.
mpps.io does not prove content quality, truth, legality, authorship, or delivery success. It proves that a hash or bounded manifest was submitted, signed, timestamped, and retained.
Small enough to add after the work is done.
- Agent frameworks: add a receipt after tool calls, file writes, or final responses.
- CI/CD: receipt build outputs and deployment manifests without changing the build system.
- Data workflows: receipt the exact dataset or response hash that downstream systems consumed.
- Skills and plugins: publish a receipt UUID beside the package users install.
- Payment or procurement workflows: link the paid event as context, but keep mpps.io independent of any payment protocol.
Three receipt modes.
- 10 structured receipts per hour
- 10 raw hash receipts per hour
- 10 certified metadata receipts per day
- HSM-signed receipts
- 10-year immutable storage
- Stripe-backed 402 payment challenge
- Rich metadata support
- Human-readable certificate page
- Parent receipt chaining
- No dependency for normal receipt use
Use HTTP directly or install the agent skill.
- Service discovery: api.mpps.io/llms.txt
- Open source: github.com/gdlg-ai/mpps.io (MIT)
- OpenClaw:
npx clawhub install mpps-attestation - skills.sh:
npx skills add gdlg-ai/mpps.io - All installation options