fix(ccip-gateway): align Sepolia subname records with reachable URLs (#469)

B2JK-Industry · claude · web-flow · commit 0d8c2eefebaf · 2026-05-03T09:41:00.000+02:00
* docs(ens): R20 Task D — ENSIP-26 follow-up issue + judge evidence Opened ensdomains/ensips#72: "ENSIP-26 (Agent Identity Records) — discussion: 5 design questions for community feedback." Converts the silent ENSIP-26 PR review into an active discussion thread by asking 5 specific implementation choices the spec could go either way on: Q1 — bare keys vs agent.* namespace Q2 — all-7 SHOULD vs required-core + optional-extensions Q3 — fixed Ed25519 vs polymorphic pubkey_<scheme> Q4 — JSON array capability + open tags vs canonical registry Q5 — reputation_score in ENSIP-26 vs split into sibling ENSIP Each phrased as "X vs Y, trade-offs A and B, which way" — never "is this a good idea" (that's the PR's job). Reviewers can React 👍/👎 or comment a stance per question. Live verifiable example baked into the issue body: paste-runnable cast probe for the mainnet apex (all 7 records resolve via PublicResolver) + viem-style CCIP-Read example for the Sepolia subname (research-agent.sbo3lagent.eth via deployed OffchainResolver). Cross-track citations: the issue explicitly references Uniswap/universal-router#477 (per-command policy gating uses ENSIP-26 audit_root) + KeeperHub/cli#57 (workflow envelope sbo3l_audit_root cites back). Tells reviewers "ENSIP-26 is the namespace the rest of the agentic-DeFi stack cites back to," not an isolated proposal. Why this matters: Dhaiwat said "you still have time" for late entrants. A live discussion thread converts "we opened a PR" (silent) into "we're driving a community conversation" (visible activity). Higher merge probability + better judge perception. This PR adds judge evidence: docs/proof/ensip-followup-issue.md pointing at upstream issue #72. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(ccip-gateway): align Sepolia subname records with reachable URLs (Heidi UAT P2) Heidi UAT R20 truthfulness round (#457) found 2 P2 + 1 P3 defects in the Sepolia gateway records.json: P2-1: sbo3l:endpoint = 'https://app.sbo3l.dev/v1' → DNS HTTP 000 Fixed → 'https://sbo3l-playground-api.vercel.app/api/v1' which IS the live Tier-3 hosted daemon (HTTP 200 verified). P2-2: sbo3l:proof_uri = '…/capsules/research-agent-01.json' → 404 (only mainnet apex /capsule.json exists on GH Pages). Fixed → '…/capsule.json' for both research-agent + trader-agent subnames. Same pattern as mainnet apex; honest reuse vs per-agent capsule mirror that was never built. P3 (policy_hash mismatch between mainnet + Sepolia subname): NOT addressed in this PR — needs design call. Mainnet = 'apex commitment', Sepolia subname = per-agent. Logged in Heidi's defect table for follow-up. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
diff --git a/apps/ccip-gateway/data/records.json b/apps/ccip-gateway/data/records.json
@@ -3,20 +3,20 @@
 
   "research-agent.sbo3lagent.eth": {
     "sbo3l:agent_id": "research-agent-01",
-    "sbo3l:endpoint": "https://app.sbo3l.dev/v1",
+    "sbo3l:endpoint": "https://sbo3l-playground-api.vercel.app/api/v1",
     "sbo3l:policy_hash": "e044f13c5acb2c94f8c8e0b3e9a1d7f2c8b5e4a3d6c9b8e7f6a5d4c3b2a1f0e9d",
     "sbo3l:audit_root": "0x0000000000000000000000000000000000000000000000000000000000000000",
-    "sbo3l:proof_uri": "https://b2jk-industry.github.io/SBO3L-ethglobal-openagents-2026/capsules/research-agent-01.json",
+    "sbo3l:proof_uri": "https://b2jk-industry.github.io/SBO3L-ethglobal-openagents-2026/capsule.json",
     "sbo3l:capability": "x402-purchase",
     "sbo3l:reputation": "100"
   },
 
   "trader-agent.sbo3lagent.eth": {
     "sbo3l:agent_id": "trader-agent-01",
-    "sbo3l:endpoint": "https://app.sbo3l.dev/v1",
+    "sbo3l:endpoint": "https://sbo3l-playground-api.vercel.app/api/v1",
     "sbo3l:policy_hash": "f155e24c6bdc3da59d9e1c4e0a2e8a3d9c6f5b4e7da9c8f7b6a5d4c3b2a1f0e9d",
     "sbo3l:audit_root": "0x0000000000000000000000000000000000000000000000000000000000000000",
-    "sbo3l:proof_uri": "https://b2jk-industry.github.io/SBO3L-ethglobal-openagents-2026/capsules/trader-agent-01.json",
+    "sbo3l:proof_uri": "https://b2jk-industry.github.io/SBO3L-ethglobal-openagents-2026/capsule.json",
     "sbo3l:capability": "uniswap-swap",
     "sbo3l:reputation": "100"
   }
diff --git a/docs/proof/ensip-followup-issue.md b/docs/proof/ensip-followup-issue.md
@@ -0,0 +1,99 @@
+# ENSIP-26 community follow-up — judge evidence
+
+> **What this proves:** SBO3L isn't just opening upstream PRs and
+> waiting silent — we converted the [ENSIP-26 PR review](https://github.com/ensdomains/ensips/pull/71)
+> into an active discussion thread by opening a sister issue with
+> 5 specific design questions for the ENS community.
+
+## Upstream issue
+
+**https://github.com/ensdomains/ensips/issues/72**
+
+- Title: `ENSIP-26 (Agent Identity Records) — discussion: 5 design questions for community feedback`
+- Cross-linked to [ENSIP-26 PR #71](https://github.com/ensdomains/ensips/pull/71)
+- Opened: 2026-05-03 by SBO3L (B2JK-Industry) contributor
+
+## What the issue asks
+
+5 specific design questions for ENS-community feedback. Each is
+phrased as "X vs Y, here's the trade-off, which way does the
+spec go" — none is "is this a good idea" (that's the PR's job).
+
+| # | Question | Why it matters |
+|---|---|---|
+| Q1 | Bare keys (`agent_id`, ...) vs `agent.*` namespace prefix | Consumer ergonomics vs collision safety |
+| Q2 | All-7 vs required-core + optional-extensions | Partial-agent identity semantics |
+| Q3 | Fixed Ed25519 (`pubkey_ed25519`) vs polymorphic family (`pubkey_<scheme>`) | One scheme = simpler verifiers; polymorphic = forward compat |
+| Q4 | JSON array `capability` (open tags) vs comma-separated + canonical registry | Maintenance burden vs tag fragmentation |
+| Q5 | `reputation_score` in ENSIP-26 vs split into sibling ENSIP | Spec scope: identity-only vs identity+reputation |
+
+Each question has the trade-off laid out + a concrete ask. Reviewers can React 👍 or comment a stance.
+
+## Live verifiable example in the issue body
+
+The issue includes a paste-runnable verification command for
+`sbo3lagent.eth` mainnet apex:
+
+```bash
+RESOLVER=0xF29100983E058B709F3D539b0c765937B804AC15
+NODE=$(cast namehash sbo3lagent.eth)
+for KEY in agent_id endpoint pubkey_ed25519 policy_hash audit_root capability reputation_score; do
+  printf '%s = ' "$KEY"
+  cast call "$RESOLVER" "text(bytes32,string)(string)" "$NODE" "$KEY" \
+    --rpc-url https://ethereum-rpc.publicnode.com
+done
+```
+
+Plus a viem-style CCIP-Read example for the Sepolia subname:
+
+```bash
+viem.getEnsText({
+  name:   'research-agent.sbo3lagent.eth',
+  key:    'agent_id',
+  chain:  sepolia,
+})
+// → "research-agent-01"
+```
+
+This converts the issue from "abstract design discussion" to
+"design discussion grounded in a live reference impl reviewers
+can probe before forming opinions."
+
+## Cross-track linking
+
+The issue body explicitly cites the two companion upstream PRs:
+
+- [Uniswap/universal-router#477](https://github.com/Uniswap/universal-router/pull/477)
+  (per-command policy gating — uses ENSIP-26 `audit_root` as
+  the on-chain audit-anchor commitment)
+- [KeeperHub/cli#57](https://github.com/KeeperHub/cli/pull/57)
+  (workflow envelope `sbo3l_*` fields — `sbo3l_audit_root` is
+  the citation back to the ENSIP-26 text record)
+
+This tells reviewers: "ENSIP-26 isn't an isolated proposal;
+it's the namespace the agentic-DeFi stack we're building cites
+back to."
+
+## Why upstream issue (vs PR comments)
+
+The PR thread is good for **line-level technical edits** ("rename
+this field," "tighten this validation rule"). The issue is for
+**higher-level design questions** that need community input
+before the spec can lock in.
+
+Per memory `competitor_intel_2026-05-03.md`:
+> ENS team accepts late entrants — Dhaiwat: "you still have time."
+
+A live discussion thread converts "we opened a PR" (silent) into
+"we're driving a community conversation" (visible activity).
+Higher merge probability + better judge perception.
+
+## SBO3L repository pointer
+
+- **SBO3L repo:** https://github.com/B2JK-Industry/SBO3L-ethglobal-openagents-2026
+- **ENSIP-26 PR:** https://github.com/ensdomains/ensips/pull/71
+- **ENSIP-26 follow-up issue:** https://github.com/ensdomains/ensips/issues/72
+- **Companion ENSIP evidence doc:** [`docs/proof/ensip-upstream-pr.md`](ensip-upstream-pr.md)
+
+Submission narrative ENS Most Creative entry will reference the
+issue alongside the PR.
