twzrd-agent-intel
The trust + receipt layer that Solana agents call before and after paying over x402.
Before an autonomous agent pays a stranger's API, it should answer: what am I
buying, should I pay this seller, and what proof will I get? twzrd-agent-intel
answers that. Free resource join inventorizes callables; free preflight
returns allow/warn/block on the seller; a paid trust endpoint returns a portable,
offline-verifiable signed v6 receipt. Live on Solana mainnet.
Where this sits
TWZRD is a buyer-side reputation rail at the payment boundary. Services come from
bazaar / resource-join discovery first; settlement trust then enriches the
pay_to wallet when it appears in the Solana x402 corpus. The settlement graph is
not a service catalog. Corpus wallets are ecosystem observations — not TWZRD
customers, usage, or revenue. Read live counts from
/health and
get_top_intel_agents?limit=1
instead of frozen README snapshots.
Core product is the buyer gate: free ReadinessCard + wash refusal on the
merchant (payTo) before spend. Optional on npm twzrd-x402-gate@0.8.4
(canonical: installTwzrdAutoGate, also "mpp"):
resource-server createTwzrdSettleGuard / twzrdPayerScreen for merchant policy on
the payer before settle+serve (abuse/sanctions/bots) — not an equal mirror of the buyer
problem; wash resistance is mainly TWZRD scoring. Facilitator operators use
@wzrd_sol/plugin-trustgate/facilitator createOnBeforeSettleHook to screen the
merchant on brokered settle.
Try it now (no install, no signup)
Resource → seller → preflight (discover callables, then score the payee):
# 1) resource join SOT — listed | live_402 claims; settlement overlay separate
curl -s "https://intel.twzrd.xyz/v1/intel/resources?limit=5" | jq '.honesty, .stats, .resources[0].url'
# 2) multi-bazaar overlay (optional; Base/Polygon wash often unknown)
# curl -s "https://intel.twzrd.xyz/v1/intel/x402-directory?limit=20"
# 3) free merchant graph card (refuse if wash_flagged)
curl -s "https://intel.twzrd.xyz/v1/intel/merchant_card/X4o2D8op42a2jcNJJVZcDq3eYivh1oR9XiezPWCXosZ"
# 4) free preflight on the 402 payTo
curl -s -X POST https://intel.twzrd.xyz/v1/intel/preflight \
-H "Content-Type: application/json" \
-d '{
"seller_wallet": "X4o2D8op42a2jcNJJVZcDq3eYivh1oR9XiezPWCXosZ",
"agent_intent": "swap_quote"
}'
# decision = allow | warn | block → abort on block
# 5) optional paid teaser ($0.001) or full trust receipt ($0.05)
# GET https://intel.twzrd.xyz/v1/intel/quick/{pubkey}
# GET https://intel.twzrd.xyz/v1/intel/trust/{pubkey}
# 6) verify any returned V6 receipt (or the sample first)
curl -s https://intel.twzrd.xyz/v1/receipts/example > /tmp/ex.json
curl -s -X POST https://intel.twzrd.xyz/v1/receipts/verify \
-H 'content-type: application/json' -d @/tmp/ex.json
One-command operator dogfood (logs day0 deltas):
python packages/twzrd-agent-intel/scripts/seller_first_flywheel_dogfood.py
Returns a ReadinessCard with decision (allow / warn / block), trust_score,
can_spend, caveats, and the path to the paid deep-dive receipt.
Install
pip install -U twzrd-agent-intel==0.5.4
Wire into Claude Code / Claude Desktop / Cursor (60 seconds)
Add to your MCP config and restart. The remote server runs the live large cross-facilitator corpus — no env vars, no local server, no configuration:
{
"mcpServers": {
"twzrd-agent-intel": {
"type": "streamable-http",
"url": "https://intel.twzrd.xyz/mcp"
}
}
}
After restart your agent sees 24 tools (watch + seller signals + no-spend gate proof live). The most useful one to call first:
get_readiness_card_tool(seller_wallet="6EF8rrecthR5Dkzon8Nwu78hRvfCKubJ14M5uBEwF6P")
That's the full install. No database, no keypair, no USDC needed for the free tools.
24 MCP tools (resource/seller-first + watch + no-spend proof)
Free discovery + preflight (no payment, no auth):
- HTTP SOT (not a 25th MCP tool):
GET /v1/intel/resources— resource join inventory get_x402_directory— multi-bazaar overlay by payTo (prefer HTTP/resourcesfor join SOT)evaluate_x402_resource— fetch a resource's 402, extractpayTo, and run preflightget_readiness_card_tool— primary ReadinessCardlow_level_preflight— lower-level preflight + spend recommendationscore_wallet_for_intel— wallet intel scoreget_top_intel_agents— secondary payer-wallet leaderboard research only (not a service catalog)
Free corpus reputation + receipts:
get_provider_reputation— observed corpus reputation for a seller/merchant walletget_merchant_card— observed receive-side graph card (GET /v1/intel/merchant_card/{wallet})is_wash_fleet— circular-flow / wash-fleet check on a walletget_facilitator_footprint— which x402 facilitators a payer settled throughget_counterparties— capped teaser: top merchants a wallet pays (limit ≤ 25)score_wallets_batch— score up to 25 wallets in one callcompare_wallets— side-by-side intel for two walletsverify_receipt— offline v5/v6 receipt verification (recompute leaf + check Ed25519 sig)verify_root_inputs— recompute published leaves and server root inputs; this does not read an on-chain anchortwzrd_demo_gate— deterministic no-spend proof; block aborts before signer, attribution remains a candidate
Solana market intel (free over MCP; $0.03 USDC each over HTTP x402):
get_solana_market_visibility_mapget_solana_market_orderbook_depthget_solana_market_shapeget_solana_market_onchain_trades_summaryget_solana_market_status
Free re-call watches:
twzrd_watch_addtwzrd_watch_listtwzrd_watch_remove
The paid trust receipt (GET /v1/intel/trust/{pubkey}, 0.05 USDC over x402) is also
available over HTTP with a portable signed v6 receipt you can verify offline.
After the free gate — paying a third-party x402 seller: see
docs/PAYABLE_CLIENT_PATH.md (feePayer-sponsored
Solana: use @x402/svm, not AgentCash; pin stable feePayer — per-request
feePayer pools make retries a lottery; see 2s-io/sdk#3).
HTTP API
# Free resource join SOT
curl -s "https://intel.twzrd.xyz/v1/intel/resources?limit=20"
# Free preflight (no payment required)
curl -s -X POST https://intel.twzrd.xyz/v1/intel/preflight \
-H "Content-Type: application/json" \
-d '{"seller_wallet": "<pubkey>", "agent_intent": "swap_quote"}'
# Paid trust receipt (0.05 USDC over x402 — use an x402-capable client)
curl -s https://intel.twzrd.xyz/v1/intel/trust/<pubkey>
# → 402 with x402 v2 payment-required if payment-signature is absent
# Free sample receipt for offline verify testing
curl -s https://intel.twzrd.xyz/v1/receipts/example
- OpenAPI: https://intel.twzrd.xyz/openapi.json
- Agent card: https://intel.twzrd.xyz/.well-known/agent-card.json
- MCP card: https://intel.twzrd.xyz/.well-known/mcp-server.json
- Quick agent guide: https://intel.twzrd.xyz/llms.txt
- Full tool and payment reference: https://intel.twzrd.xyz/llms-full.txt
- Apex front door: https://twzrd.xyz/llms.txt (resources-first; full contract on intel)
Self-hosting
Note: Running
twzrd-agent-intel-mcplocally serves from a small static research registry by default — not the live large cross-facilitator corpus. Use the remote MCP above for real data. Self-host only if you need to run behind a firewall or extend the server.
# Local MCP stdio (static demo data by default)
twzrd-agent-intel-mcp
# Local HTTP server
uvicorn twzrd_agent_intel.server_http:app --port 8001
Point a self-hosted instance at live data with environment variables:
TWZRD_SIGNAL_SOURCE |
What it does |
|---|---|
artifacts |
Aggregate your own settlement volume from TWZRD_ARTIFACTS_DIR |
bazaar |
Pull the live cross-facilitator x402 catalog (CDP Bazaar) |
What makes the score trustworthy
- Wash/Sybil-resistant: circular-flow- and wash-resistant scoring over a large cross-facilitator corpus.
- Honest basis: trust components are renormalized (
score_version intel_renorm_v1) so arbitrary wallets aren't penalized for lacking TWZRD-native execution history. - Verifiable: every paid response carries a signed v6 receipt you can verify offline
at
/v1/receipts/verifywithout trusting our server.
More TWZRD surfaces
- Verify a live receipt offline in one command (TypeScript SDK, no wallet):
npx twzrd-receipt-verifier@1.2.2 receipt.json --pubkey <published-key>
- OpenClaw agents: install the skill that wraps this API into a pre-payment check:
clawhub install twzrd-trust
License
MIT
- Public leaderboard: https://twzrd.xyz/leaderboard
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file twzrd_agent_intel-0.5.5.tar.gz.
File metadata
- Download URL: twzrd_agent_intel-0.5.5.tar.gz
- Upload date:
- Size: 372.8 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.10.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
c567dabc9e1acad6335897a83d1e991c7197a0cf10b936e56b33c591a79ba0aa
|
|
| MD5 |
ed6736800f6572101c6b31738cee63ca
|
|
| BLAKE2b-256 |
e05ef82a630bea5db4f4843726364e02038061495ec2caf59534788f63a2c33b
|
File details
Details for the file twzrd_agent_intel-0.5.5-py3-none-any.whl.
File metadata
- Download URL: twzrd_agent_intel-0.5.5-py3-none-any.whl
- Upload date:
- Size: 404.5 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.10.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
01f1e0575458a669c0c85ae24691727aee26a77f7f1e5e5ca5ddb7a2029fdbcd
|
|
| MD5 |
41f7cdb0d375f06957cc4b9d0d1e29df
|
|
| BLAKE2b-256 |
ff83561142b2f070be956574985fbb8c3232c99d3632e33ed0a41dabe90dc988
|