integration/v0.14.0-rc1

internal/agentcrd/agent_contract_integration_test.go

@@ -40,8 +40,8 @@ import (
 //	    (agentcrd.HostNoBundledSkillsMarkerPath), so Hermes' installer/sync
 //	    skips seeding its ~80 bundled skills;
 //	(2) the rendered hermes-config ConfigMap in the agent's namespace carries
-//	    the capped knobs: lifetime_seconds: 90, max_turns: 30,
-//	    reasoning_effort: low, and disabled_toolsets {memory, web};
+//	    the capped knobs: lifetime_seconds: 180, max_turns: 30,
+//	    reasoning_effort: low, and disabled_toolsets {memory, web, code_execution};
 //	(3) a BEHAVIORAL signal that bundled skills were actually skipped — see
 //	    assertBundledSkillsSkippedInPod for why we assert pod filesystem state
 //	    rather than grep a log line.
@@ -233,7 +233,7 @@ func getHermesConfigYAML(t *testing.T, cfg *config.Config, ns string) string {
 func assertHermesConfigCaps(t *testing.T, cfgYAML string) {
 	t.Helper()
 	for _, want := range []string{
-		"lifetime_seconds: 90",
+		"lifetime_seconds: 180",
 		"max_turns: 30",
 		"reasoning_effort: low",
 		"disabled_toolsets:",

internal/buyprompts/buyprompts.go

@@ -158,7 +158,10 @@ func modelOr(in Input, placeholder string) string {
 }
 
 func agentBlock(in Input) Block {
-	modelFlag := modelOr(in, "<model-id>")
+	// Deliberately no model anywhere: an Obol Agent runs its own pinned
+	// model, skills, and memory — the buyer never picks one, pay-agent takes
+	// no --model, and the agent ignores the chat-completions `model` field,
+	// so the wire example omits it too.
 	return Block{
 		CallShape: CallShape{
 			Method:    "POST",
@@ -171,8 +174,8 @@ func agentBlock(in Input) Block {
 				"Use the buy-x402 skill's `pay-agent` command to buy one round of work from this "+
 					"Obol Agent — it has its own skills, tools, and memory, not just a model. Edit the "+
 					"message, then run:\n\n"+
-					"pay-agent %s --model %q --message %q",
-				in.URL, modelFlag, task(in),
+					"pay-agent %s --message %q",
+				in.URL, task(in),
 			),
 			PromptGenericLLM: fmt.Sprintf(
 				"Help me call the Obol Agent at %s — it's an autonomous agent (tools + skills + memory), "+
@@ -187,7 +190,7 @@ func agentBlock(in Input) Block {
 				in.URL, task(in),
 			),
 		},
-		Example: ChatExample(in.URL, in.Model, in.TaskExample),
+		Example: ChatExample(in.URL, "", in.TaskExample),
 	}
 }
 

internal/embed/infrastructure/cloudflared/values.yaml

@@ -5,7 +5,7 @@ transport:
 
 image:
   repository: cloudflare/cloudflared
-  tag: "2026.7.1@sha256:188bb03589a32affed3cf4d0590565ffe67b78866e6b5582574afab2b705bafe"
+  tag: "2026.7.2@sha256:4f6655284ab3d252b7f28fedb19fe6c8fc82ee5b1295c20ac74d475e5398a52d"
 
 metrics:
   address: "0.0.0.0:2000"

internal/embed/infrastructure/helmfile.yaml

@@ -17,7 +17,7 @@ repositories:
 # Single source of truth: change this to switch networks
 values:
   - network: mainnet
-  - gatewayApiVersion: v1.6.0
+  - gatewayApiVersion: v1.6.1
   # Default the cloudflared release to enabled. `obol stack up` overrides via
   # `--state-values-set cloudflared.enabled=false` when it detects a running
   # quick tunnel that should be preserved across the sync.

internal/embed/networks/ethereum/helmfile.yaml.gotmpl

@@ -91,20 +91,34 @@ releases:
           image:
             {{- if eq .Values.executionClient "reth" }}
             # renovate: datasource=github-releases depName=paradigmxyz/reth
-            tag: v2.3.0
+            tag: v2.4.1
             {{- else if eq .Values.executionClient "geth" }}
             # renovate: datasource=github-releases depName=ethereum/go-ethereum
             tag: v1.17.4
             {{- else if eq .Values.executionClient "nethermind" }}
             # renovate: datasource=github-releases depName=NethermindEth/nethermind
-            tag: "1.39.0"
+            tag: "1.39.1"
             {{- else if eq .Values.executionClient "besu" }}
             # renovate: datasource=github-releases depName=hyperledger/besu
             tag: "26.7.0"
             {{- else if eq .Values.executionClient "erigon" }}
             # renovate: datasource=github-releases depName=erigontech/erigon
-            tag: v3.5.1
+            tag: v3.5.2
             {{- end }}
+          {{- if eq .Values.executionClient "reth" }}
+          # reth keeps p2p (30303) closed while running post-crash consistency
+          # recovery, which can exceed any bounded probe window on a large
+          # datadir. The chart's default tcp liveness probe then SIGKILLs it
+          # mid-recovery, forcing a longer recovery on the next start — a
+          # permanent crash loop. A genuinely dead process exits the container
+          # on its own, so make the probe effectively never fire.
+          livenessProbe:
+            tcpSocket:
+              port: p2p-tcp
+            initialDelaySeconds: 60
+            periodSeconds: 120
+            failureThreshold: 10000
+          {{- end }}
           persistence:
             enabled: true
             size: {{ if (index .Values "executionStorageSize" | default "") }}{{ index .Values "executionStorageSize" }}{{ else }}{{ if eq .Values.network "mainnet" }}{{ if eq .Values.mode "archive" }}4500Gi{{ else }}500Gi{{ end }}{{ else }}{{ if eq .Values.mode "archive" }}300Gi{{ else }}100Gi{{ end }}{{ end }}{{ end }}
@@ -120,7 +134,7 @@ releases:
             tag: v8.2.0
             {{- else if eq .Values.consensusClient "prysm" }}
             # renovate: datasource=github-releases depName=OffchainLabs/prysm
-            tag: v7.1.6
+            tag: v7.1.7
             {{- else if eq .Values.consensusClient "teku" }}
             # renovate: datasource=github-releases depName=Consensys/teku
             tag: "26.7.1"

internal/embed/skills/buy-x402/SKILL.md

@@ -21,7 +21,7 @@ python3 ${OBOL_SKILLS_DIR:-/data/.openclaw/skills}/buy-x402/scripts/buy.py go <u
 `go` never creates persistent state; for a pre-authorized inference pool it points you at `buy`. The expert flows underneath, picked by usage shape:
 
 - **`pay <url>`** — single-shot. Probe the URL, sign **one** payment authorization, attach `X-PAYMENT`, send the request, return the response. Stateless. Use for `type:http` services and any one-off purchase. Max loss = price of one request. Settlement normally lands only after the request succeeds — but a facilitator can submit the settle tx on-chain and *then* fail the request. When that happens the failure report prints `⚠️ SETTLEMENT MAY HAVE COMPLETED ON-CHAIN` with the tx hash: verify with `balance --chain <X>` before retrying (mechanism: docs/observability.md, "Verify settlement against the chain"). Applies to `pay-agent` too.
-- **`pay-agent <url> --model <id>`** — single-shot paid **streaming** agent call. Same payment shape as `pay` (one auth, X-PAYMENT, max-loss = price), but POSTs to `<url>/v1/chat/completions` with `stream: true` and forwards every SSE event verbatim to stdout as it arrives. Use this for `type:agent` ServiceOffers when the calling agent wants to consume the response *itself* (memory, tool-call traces, partial results) instead of routing it through LiteLLM as a paid alias. Default HTTP read timeout is **1 hour** — agent calls can legitimately run for many minutes; override with `--timeout <seconds>`.
+- **`pay-agent <url>`** — single-shot paid **streaming** agent call. Same payment shape as `pay` (one auth, X-PAYMENT, max-loss = price), but POSTs to `<url>/v1/chat/completions` with `stream: true` and forwards every SSE event verbatim to stdout as it arrives. No `--model`: a `type:agent` offer runs its own model (the request `model` field is ignored), so you only send a prompt. Use this for `type:agent` ServiceOffers when the calling agent wants to consume the response *itself* (memory, tool-call traces, partial results) instead of routing it through LiteLLM as a paid alias. Default HTTP read timeout is **1 hour** — agent calls can legitimately run for many minutes; override with `--timeout <seconds>`.
 - **`buy <name>`** — pre-authorize a budget. Sign **N** authorizations up front (the buyer pays nothing yet), declare them in a `PurchaseRequest` CR, let the `x402-buyer` sidecar redeem them transparently as the agent calls the model through LiteLLM at `paid/<remote-model>`. Use for long-running paid inference. Max loss = N × price (only as vouchers are spent); runtime path holds zero signer access.
 - **`buy <name> --model <id> --set-default`** — same as `buy` above, then adopt `paid/<remote-model>` as the agent's **own primary model**, in-pod, by itself: an atomic `hermes config set model.default` that Hermes re-reads per request (effective next chat turn, **no restart**, no host-side `obol model prefer`/`obol model sync`). Refuses if the model isn't selectable in LiteLLM. Pair with `--auto-refill` so the primary model doesn't brick when the pre-authorized budget runs out.
 
@@ -170,7 +170,7 @@ python3 ${OBOL_SKILLS_DIR:-/data/.openclaw/skills}/buy-x402/scripts/buy.py pay h
 # One-shot paid STREAMING agent call (SSE events flushed to stdout as they arrive)
 python3 ${OBOL_SKILLS_DIR:-/data/.openclaw/skills}/buy-x402/scripts/buy.py pay-agent \
     https://seller.example.com/services/demo-quant \
-    --model qwen3.5:9b --message 'summarize the latest research on staking'
+    --message 'summarize the latest research on staking'
 
 # Pay-agent with a full OpenAI-compatible body (stream:true is forced on)
 python3 ${OBOL_SKILLS_DIR:-/data/.openclaw/skills}/buy-x402/scripts/buy.py pay-agent \
@@ -224,7 +224,7 @@ python3 ${OBOL_SKILLS_DIR:-/data/.openclaw/skills}/buy-x402/scripts/buy.py maint
 | `go <url> [--message <text>] [--data <json>] [--method GET\|POST]` | Probe, detect offer type (agent / chat inference / http), dispatch to the right flow |
 | `probe <url> [--model <id>] [--type http\|inference\|agent] [--method GET\|POST]` | Send request without payment, parse 402 response for pricing |
 | `pay <url> [--type http\|inference] [--method GET\|POST] [--data <body>]` | Single-shot paid request: sign 1 auth, attach X-PAYMENT, send |
-| `pay-agent <url> --model <id> [--message <text> \| --data <json>] [--timeout <s>]` | Single-shot paid streaming agent call: SSE events flush to stdout as they arrive (default timeout 1h) |
+| `pay-agent <url> [--message <text> \| --data <json>] [--timeout <s>]` | Single-shot paid streaming agent call (no `--model` — the agent runs its own): SSE events flush to stdout as they arrive (default timeout 1h) |
 | `buy <name> --endpoint <url> --model <id> [--budget <amount>] [--count N]` | Pre-sign auths, create/update `PurchaseRequest`, expose `paid/<model>`. `--budget` takes atomic units or token units (`'1.5 USDC'`) |
 | `buy <name> --endpoint <url> --model <id> --set-default [--auto-refill]` | As above, then set `paid/<model>` as the agent's own primary model in-pod (no restart, no host CLI) |
 | `process <name> \| --all` | Reconcile `autoRefill` policies against live `x402-buyer` status |

internal/embed/skills/buy-x402/scripts/buy.py

@@ -2647,7 +2647,7 @@ def cmd_pay(url, method="GET", data=None, kind="http", network=None, timeout=Non
         sys.exit(1)
 
 
-def cmd_pay_agent(url, messages=None, model_id=None, network=None, timeout=None, body=None, token=None, payment_option=None):
+def cmd_pay_agent(url, messages=None, network=None, timeout=None, body=None, token=None, payment_option=None):
     """Single-shot paid streaming agent call: probe -> sign one auth -> SSE-stream.
 
     Sibling of `cmd_pay` for `type=agent` ServiceOffers. Differences from
@@ -2667,9 +2667,10 @@ def cmd_pay_agent(url, messages=None, model_id=None, network=None, timeout=None,
         alias.
 
     `body` is an optional JSON-encoded request body. When omitted, `messages`
-    + `model_id` are required and a `{model, messages, stream:true}` body is
-    synthesized. When provided, the body is parsed and `"stream": true` is
-    forced onto whatever the caller passed.
+    is required and a `{messages, stream:true}` body is synthesized — NO `model`
+    field: a type=agent offer runs its own model and ignores any `model` sent.
+    When provided, the body is parsed and `"stream": true` is forced onto
+    whatever the caller passed.
     """
     if timeout is None or float(timeout) <= 0:
         timeout = 3600.0
@@ -2689,27 +2690,24 @@ def cmd_pay_agent(url, messages=None, model_id=None, network=None, timeout=None,
         # Force streaming on. cmd_pay handles non-streaming; cmd_pay_agent
         # exists precisely to stream.
         parsed_body["stream"] = True
-        if model_id and not parsed_body.get("model"):
-            parsed_body["model"] = model_id
     else:
         if not messages:
             print(
                 "Error: --message (or --data <json>) is required for `pay-agent`.\n"
-                "Example: pay-agent <url> --model qwen3.5:9b --message 'summarize the docs'",
+                "Example: pay-agent <url> --message 'summarize the docs'",
                 file=sys.stderr,
             )
             sys.exit(1)
-        if not model_id:
-            print("Error: --model is required when using --message.", file=sys.stderr)
-            sys.exit(1)
+        # type=agent ServiceOffers run their own model — there is nothing to
+        # select and the agent ignores any `model` field — so pay-agent sends
+        # only the prompt.
         parsed_body = {
-            "model": model_id,
             "messages": [{"role": "user", "content": messages}],
             "stream": True,
         }
 
     print(f"Probing {url} ...")
-    pricing = _probe_endpoint(url, model_id=model_id or "test", kind="inference")
+    pricing = _probe_endpoint(url, model_id="probe", kind="inference")
     if not pricing:
         print("Failed to get x402 pricing.", file=sys.stderr)
         sys.exit(1)
@@ -3095,7 +3093,9 @@ def usage():
     print("  pay <url> [--type http|inference] [--method GET|POST] [--data '<body>'] [--timeout <seconds>]")
     print("       [--token <SYMBOL>] [--network <name>] [--payment-option <N>]")
     print("                                               Single-shot paid request (sign 1 auth, attach X-PAYMENT)")
-    print("  pay-agent <url> --model <id> [--message '<text>' | --data '<json>'] [--timeout <seconds>]")
+    print("                                               Multi-currency offers: pick which asset/price to pay with")
+    print("                                               --token/--network/--payment-option (probe to see options)")
+    print("  pay-agent <url> [--message '<text>' | --data '<json>'] [--timeout <seconds>]")
     print("       [--token <SYMBOL>] [--network <name>] [--payment-option <N>]")
     print("                                               Single-shot paid streaming agent call (POST /v1/chat/completions,")
     print("                                               stream: true). Each SSE event flushes to stdout. Default timeout 1h.")
@@ -3190,7 +3190,7 @@ def usage():
         positional, opts = parse_flags(rest)
         if not positional:
             print(
-                "Usage: pay-agent <url> --model <id> [--message '<text>' | --data '<json>'] "
+                "Usage: pay-agent <url> [--message '<text>' | --data '<json>'] "
                 "[--network <name>] [--timeout <seconds>]",
                 file=sys.stderr,
             )
@@ -3207,7 +3207,6 @@ def usage():
         cmd_pay_agent(
             positional[0],
             messages=opts.get("message"),
-            model_id=opts.get("model"),
             network=opts.get("network"),
             timeout=timeout,
             body=opts.get("data"),

internal/monetizeapi/types.go

@@ -304,6 +304,7 @@ type ServiceOfferPayment struct {
 	// x402 payment scheme.
 	// +kubebuilder:default="exact"
 	// +kubebuilder:validation:Enum=exact
+	// TODO(auth-capture): regenerate CRD manifests + wire per-offer unlock config
 	Scheme string `json:"scheme,omitempty"`
 	// Chain identifier for payments (human-friendly). Reconciler resolves
 	// to CAIP-2 format (e.g., "base-sepolia" → "eip155:84532").

internal/serviceoffercontroller/assets/README.md

@@ -0,0 +1,55 @@
+# Agent chat widget assets
+
+`chat.html` — the self-contained agent chat page served at `/chat` on every
+agent-type offer's dedicated origin. Hand-maintained; no build step. It
+derives the agent name from the hostname and price/model/network/asset from
+the live 402 challenge, so the same file works for every agent offer on any
+stack and network (Base mainnet `eip155:8453` and Base Sepolia
+`eip155:84532` are supported).
+
+## Known limitations
+
+- **Session key is derived from a wallet signature** (`keccak256(personal_sign(<EIP-4361 message>))`),
+  so the wallet must produce a *deterministic* signature for the same message.
+  Standard RFC-6979 EOAs (MetaMask, Rabby, Ledger, Trezor) do; the `getCode`
+  guard rejects contract wallets (non-reproducible ERC-1271) while allowing
+  EIP-7702-delegated EOAs. **Residual gap:** MPC / threshold-ECDSA wallets are
+  code-less EOAs that pass the guard but sign non-deterministically — funding a
+  session from one strands the balance on the next visit. There is no on-chain
+  signal to detect this without a second signature popup, which this
+  one-signature flow deliberately avoids. Use a standard EOA.
+- **Per-turn spend is capped at the price shown when the page loaded** (and the
+  session balance). A turn whose 402 amount exceeds the displayed price is
+  refused; a legitimate price change is picked up on reload. Max loss per
+  session is bounded by what you fund into the session wallet.
+- **The signature itself is key material.** Anything that can read it (a
+  malicious extension, a hooked `window.ethereum`) controls the session funds —
+  keep session balances small.
+
+`chat-vendor.js` — generated single-file ESM bundle of the widget's
+dependencies. Do not edit by hand. Rebuild:
+
+```sh
+npm init -y && npm i viem@2.21.25 @x402/fetch@2.18.0 @x402/evm@2.18.0
+cat > vendor-entry.mjs <<'EOF'
+export { createWalletClient, createPublicClient, custom, http, erc20Abi,
+         formatUnits, parseUnits, keccak256 } from "viem";
+export { privateKeyToAccount } from "viem/accounts";
+export { base, baseSepolia } from "viem/chains";
+export { wrapFetchWithPayment, x402Client } from "@x402/fetch";
+export { ExactEvmScheme, toClientEvmSigner } from "@x402/evm";
+EOF
+npx esbuild vendor-entry.mjs --bundle --format=esm --minify --target=es2022 \
+  --outfile=chat-vendor.js
+```
+
+When the bundle is rebuilt, bump the `?v=` cache-buster on the
+`chat-vendor.js` import in `chat.html` to the new sha256's first 8 hex
+chars — intermediaries (e.g. Cloudflare) cache `.js` aggressively.
+
+sha256 of the committed bundle:
+`895fd923aa84d7cf80e2b1df299068aa38dba7307a9a380526c0b5426489724d`
+
+The pinned versions are the exact pair validated end-to-end against the
+x402-verifier with real on-chain settlements (X-PAYMENT v1 and
+PAYMENT-SIGNATURE v2 flows both accepted since #690).