A2A toolkit

Tools for agent-to-agent traffic

What it is

Free tools for agent-to-agent (A2A) traffic: publish and check agent cards, agents.txt and agents.xml; classify agent traffic; sign quotes, acceptances, SLAs, task receipts, heartbeats, version and key-rotation statements, provenance stamps, referral and message receipts; seal dispute evidence; build usage invoices; read x402 price tags. Each tool is a POST endpoint; a GET on the same URL shows its input and an example. Index: https://www.rightshold.com/v1/a2a/toolkit.

How receipts work

Receipts are signed with Rightshold's issuer key (spec rs-a2a-net/0.1) and are not stored: the signature is the proof, so keep the receipt. Verify free at POST https://www.rightshold.com/v1/a2a/verify or offline with https://www.rightshold.com/.well-known/rightshold-keys.json. Where a tool takes an agent's own signature (handshakes, delegation chains, heartbeats, versions, key succession, message receipts), the agent signs the canonical JSON (sorted keys) of the statement with its Ed25519 key. These receipts are signed records, not ledger entries; for ledgered records use the agent controls and AI run receipts.

No AI in the answers

Every check, score and decision is plain code: the same input gives the same output. The listing quality score uses fixed published weights (rs-listing-quality/0.1: title 10, description 20, pricing 15, endpoint 15, examples 10, terms 10, contact 5, identity 10, tags 5). The traffic classifier is header rules; a User-Agent can be spoofed, so only a verified signature or credential identifies an agent.

Tools

• A2A agent card validator: POST https://www.rightshold.com/v1/a2a/agent-card/validate — Checks an A2A AgentCard (/.well-known/agent-card.json) for required fields, skill ids, media types and security hints; returns errors, warnings and a 0-100 score.
• A2A agent card generator: POST https://www.rightshold.com/v1/a2a/agent-card/generate — Builds a valid A2A AgentCard from a name, description, URL and skills, ready to publish at /.well-known/agent-card.json.
• A2A protocol task receipts: POST https://www.rightshold.com/v1/a2a/task-receipts — Validates an A2A Task object (id, contextId, status.state, artifacts) and signs a receipt of its state and artifact hashes, so client and server agents share one record of what happened.
• Agent-to-agent mutual-consent handshake: POST https://www.rightshold.com/v1/a2a/handshake — Two agents each sign {spec: rs-a2a-handshake/0.1, purpose, parties (sorted key ids), nonce}; when both signatures verify, Rightshold signs a consent receipt. An introduction only counts when both sides agreed.
• Signed delegation chain verifier (agent X acting for agent Y): POST https://www.rightshold.com/v1/a2a/delegation-chains/verify — Verifies a chain of Ed25519-signed grants from a root agent to the acting agent: each signature, expiry, link continuity, and that no link widens the scope above it. Optionally checks one action against the effective scope.
• Agent commerce quotes (signed offers): POST https://www.rightshold.com/v1/a2a/quotes — A seller agent records a quote: buyer, amount, currency, the SHA-256 of the terms and an expiry. Signed so the buyer can hold the seller to it.
• Agent commerce acceptances (quote → acceptance trail): POST https://www.rightshold.com/v1/a2a/acceptances — The buyer accepts an unexpired, signed quote; the acceptance receipt binds the quote hash, amount and terms, completing the offer-acceptance contract trail.
• Agent service-level (SLA) commitments: POST https://www.rightshold.com/v1/a2a/sla — A provider agent commits to uptime, p95 latency and error-rate targets over a window, with a service credit per breach. Signed.
• SLA breach records: POST https://www.rightshold.com/v1/a2a/sla/evaluate — Compares observed figures for a period against a signed SLA and signs the result: breached or not, which targets, and the service credit owed.
• agents.txt etiquette policy generator: POST https://www.rightshold.com/v1/agents-txt/generate — Writes an agents.txt file telling agents which paths they may use, the rate limit, the KYA level and whether HTTP message signatures are required.
• agents.txt policy checker: POST https://www.rightshold.com/v1/agents-txt/check — Given an agents.txt text and one request (agent, path, recent rate, KYA level, signed), says allowed or not with the exact reasons.
• robots.txt checker for AI agents: POST https://www.rightshold.com/v1/robots/agent-check — Parses a robots.txt (longest-match Allow/Disallow, wildcards, $) for one agent and path, and lists Content-Signal and Sitemap lines.
• Agent sitemap (/agents.xml) generator: POST https://www.rightshold.com/v1/agents-xml/generate — Builds an agents.xml listing a site's agent endpoints (MCP, A2A, OpenAPI, REST, webhook, x402, llms.txt, agent card) with auth hints, plus the Link header to advertise it.
• Agent sitemap (/agents.xml) validator: POST https://www.rightshold.com/v1/agents-xml/validate — Checks an agents.xml document: root element, endpoint types and absolute URLs.
• Agent traffic classifier for sites: POST https://www.rightshold.com/v1/traffic/classify — Classifies one request from its headers: human browser, script, crawler or AI agent; which family; and how strongly it is identified (claimed User-Agent, HTTP message signature, KYA credential).
• Webhook signature verifier for agent callbacks: POST https://www.rightshold.com/v1/webhooks/verify-signature — Checks a callback's signature: Stripe (t=,v1=), Standard Webhooks (svix), GitHub sha256=, plain HMAC-SHA256 hex or base64, or Ed25519. Constant-time comparison; the secret is used for this check only and never stored or logged.
• Agent uptime heartbeat attestations: POST https://www.rightshold.com/v1/a2a/heartbeats — An agent signs {agent, at, status, version} with its key; Rightshold checks the signature and clock skew and countersigns with its own time, proving the agent was alive and holding its key then.
• Agent version pinning and changelog attestations: POST https://www.rightshold.com/v1/a2a/versions — An agent signs a release statement {agent, version (semver), previous_version, changelog_sha256, config_sha256, breaking}; Rightshold countersigns so callers can pin to a version and see when behavior changed.
• Agent identity rotation (key succession records): POST https://www.rightshold.com/v1/a2a/key-succession — Both the old and the new Ed25519 key sign one succession statement; Rightshold countersigns it, so counterparties can follow an agent across key rotations.
• Data-provenance stamps for agent outputs: POST https://www.rightshold.com/v1/provenance/stamps — Stamps an agent output's SHA-256 with its producer, model, tool calls and sources (URI, hash, license). Output text is hashed and not stored.
• Provenance stamp verifier: POST https://www.rightshold.com/v1/provenance/verify — Verifies a provenance stamp's signature and, if you send the output text, that it matches the stamped hash.
• Agent referral and attribution receipts: POST https://www.rightshold.com/v1/a2a/referrals — Signs who sent traffic where (referrer, destination, campaign, revenue share, expiry) and returns an attribution token to pass along, for revenue-share claims between agents.
• Agent message delivered/read receipts: POST https://www.rightshold.com/v1/a2a/message-receipts — The recipient agent signs an acknowledgement {message_sha256, sender, recipient, status: delivered|read|rejected, at}; Rightshold verifies and countersigns it as proof of delivery.
• Agent-to-agent dispute evidence bundles: POST https://www.rightshold.com/v1/a2a/evidence-bundles — Seals a list of evidence hashes for a dispute under one Merkle root with a proof per item, so each item can be disclosed later on its own without revealing the rest.
• Usage-based invoice generator for agent services: POST https://www.rightshold.com/v1/invoices/usage — Builds a signed invoice from metered usage lines (metric, quantity, free units, unit price) in exact decimals, with tax and totals rounded for the currency (USD, USDC to 6 places).
• Agent marketplace listing quality score: POST https://www.rightshold.com/v1/listings/quality-score — Scores a marketplace listing 0-100 with a published fixed-weight formula (rs-listing-quality/0.1) and tells you exactly what to fix.
• x402 price-tag reader for third-party endpoints: POST https://www.rightshold.com/v1/x402/price-tag — Reads an endpoint's HTTP 402 x402 payment requirements (body or base64 header) into plain price tags: network, asset, pay-to, amount (USDC shown in dollars) and the cheapest option. Checks required fields.
• Free verifier for A2A toolkit receipts: POST https://www.rightshold.com/v1/a2a/verify — Verifies any receipt issued by the A2A toolkit (quotes, acceptances, SLAs, task receipts, heartbeats, stamps, invoices...) and whether it has expired.

Price

Free. No account needed.