agentleFS
0xSoftBoi0xSoftBoi/suwappubot/.claude/skills/ethglobal-goal/SKILL.md

ethglobal-goal

Skill4 starsChanged 3 months agoApache-2.0View on GitHub

Standing goal: build the ETHGlobal Tokyo 2026 'Agent Swap Passport' hackathon entry (docs/plans/ethglobal-tokyo2026-agent-passport.md) — World ID/AgentKit + Intercepta + ENSv2 — until every phase is implemented, verified against real sandbox/testnet endpoints, and demoed end to end. Usage: /ethglobal-goal [phase|item]

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If an item is blocked on credentials/access this session doesn't have, mark it

    line 22
  2. Distinctive:

    Never fake a live result: If a sandbox/testnet call can't actually be made this session (missing key, no network egress, etc.), the item …

    line 103

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/ethglobal-goal — Ship the Agent Swap Passport hackathon entry

Standing goal: every phase of docs/plans/ethglobal-tokyo2026-agent-passport.md implemented and live-verified against real sandbox/testnet endpoints — not just code that parses. This targets three ETHGlobal Tokyo 2026 sponsor tracks (World IDKit+AgentKit, Intercepta, ENSv2) by extending Suwappu's existing agent link/claim flow (api-ts/src/routes/agent.ts), not building identity from scratch.

How to use

  • /ethglobal-goal — pick the next unchecked item in phase order, re-verify its file:line anchor against current code (things drift), implement it, run the phase's live verification (not just bun run check — an external API call actually succeeding), check it off.
  • /ethglobal-goal <phase|item> — jump to a specific phase (e.g. /ethglobal-goal 1 or /ethglobal-goal 2.2).
  • "Done" for this goal means: tested live against the real World/Intercepta/ENS sandbox+testnet endpoints, not mocked. A unit test with a stubbed HTTP client does not close an item — only a real round-trip response (or a documented, unresolvable external blocker) does.
  • If an item is blocked on credentials/access this session doesn't have, mark it BLOCKED ON USER with the exact secret/access needed, and move to the next unblocked item instead of spinning. Never fabricate a credential, contract address, or API response to make an item look done — that produces a demo that fails live in front of judges, which is worse than an honest blocked item.
  • Goal is DONE when every box is checked or explicitly accepted as blocked-and-reported to the user; then archive this skill (delete the directory in a final commit noting completion/handoff).

Phase 0 — Credentials & unresolved API mechanics (must clear before Phase 1-3 can go live)

BLOCKED ON USER — needs secrets/access this session does not have:

  • 0.1 BLOCKED (needs World Developer Portal access) — Create a World App ID + action ID at developer.world.org for IDKit; obtain sandbox verification credentials for AgentKit.
  • 0.2 BLOCKED (needs a byte-level source read) — Read ./x402/DOCS.md in github.com/worldcoin/agentkit (or Must-be-Ash/world-x402-agentkit-example source directly) to confirm the exact agentkit header payload schema. Prior research only confirmed the header name and round-trip shape.
  • 0.3 BLOCKED (needs Intercepta sandbox key) — Obtain an Intercepta/Web3Antivirus X-API-KEY and the event's demo clean/risky addresses (reportedly distributed via event Discord/booth, not self-serve).
  • 0.4 BLOCKED (needs a funded Sepolia wallet + ENSv2 registrar access) — Register/acquire an ENSv2 parent name on Sepolia (e.g. a suwappu.eth-style test namespace) for subname issuance, and confirm the actual ENSv2 subname-creation call from docs.ens.domains/ensv2/registry + /ensv2/permissions directly (NOT the ENSv1 NameWrapper.setSubnodeOwner pattern — confirmed wrong API for v2).
  • 0.5 Confirm current agents.ownerUserId / POST /v1/agent/link/code flow still matches api-ts/src/routes/agent.ts:4173-4236 before extending it (re-verify line anchors — this plan is a point-in-time read).

Phase 1 — World: strengthen the existing link flow (both $5k bounties)

Files: api-ts/src/routes/agent.ts, new api-ts/src/lib/worldId.ts, new api-ts/src/middleware/worldIdAuth.ts

  • 1.1 lib/worldId.ts: client for POST https://developer.world.org/api/v4/verify/{app_id} (IDKit proof verification). Live-verify: a real proof from the World App simulator/sandbox round-trips to a 200. This alone satisfies the IDKit bounty (server-side verify).
  • 1.2 Extend POST /v1/agent/link/code (agent.ts:4176) to optionally accept a World ID proof; verify via 1.1 before minting the code; store proof hash/nullifier/timestamp in agents.metadata.
  • 1.3 Add agentkit-header verification (per the schema confirmed in 0.2) as a gate ahead of meteredPayment in middleware/x402Payment.ts — protected action = a real metered swap/agent call.
  • 1.4 Denied/expired/cancelled path: mirror the existing already_linked 409 pattern (agent.ts:4188) for world_id_verification_failed, audited via writeAuditLog like the existing agent.link_code_rejected event.
  • 1.5 Live end-to-end test: mint a link code with a valid World ID proof (succeeds), then with an invalid/expired one (rejected, audited) — both against the real sandbox endpoint.

Phase 2 — Intercepta: screen before charging ($2k bounty)

Files: new api-ts/src/lib/intercepta.ts, api-ts/src/middleware/x402Payment.ts

  • 2.1 lib/intercepta.ts: client for the quick-scan endpoint, X-API-KEY auth. Live-verify: real call against the event's demo clean address returns a low toxicScore.
  • 2.2 Wire into meteredPayment before consumePayment/verifyX402Payment finalizes a charge; reject with traits[] reasoning above threshold.
  • 2.3 Persist last scan result to agents.metadata (compounding risk history, not a one-shot check).
  • 2.4 Live end-to-end test: one call against the demo clean address (approved, charge proceeds) and one against the demo risky address (blocked, reasoning shown) — the bounty's exact required demo shape.

Phase 3 — ENS: durable identity anchor ($6k bounty)

Files: new api-ts/src/lib/ensSubname.ts, extend agents.metadata

  • 3.1 Mint an ENSv2 subname under the Phase-0.4 parent name once an agent is claimed AND World ID-verified. Live-verify: subname resolves on Sepolia via a public ENS resolver/explorer.
  • 3.2 Store subname in agents.metadata.ensName; surface ens_name in the existing agent-status response alongside owner_linked (agent.ts:565).
  • 3.3 Live end-to-end demo: resolve <agent>.suwappu.eth (or chosen namespace) and show it returns the World ID-verified owner claim + current Intercepta risk state — the composed "one identity, three sponsors" story.

Phase 4 — Submission packaging

  • 4.1 One README with a per-track section pointing judges at exact files/lines (each sponsor judges independently).
  • 4.2 Intercepta feedback (3-5 lines on their API) in the README per bounty requirement — re-check exact wording at submission time.
  • 4.3 Confirm git history is normal commits (not squashed/single-commit) — several bounties (1inch, Uniswap) required this; carry the same hygiene even though those tracks are out of scope, since ETHGlobal judges may check general repo quality.
  • 4.4 Final live run-through of the full flow (claim → World ID verify → ENS mint → metered call gated by Intercepta) on a clean environment, recorded/screenshotted as submission evidence.

Rules

  • Never fake a live result. If a sandbox/testnet call can't actually be made this session (missing key, no network egress, etc.), the item stays unchecked and BLOCKED — do not write code that only "looks" verified from a mocked response and report it as tested live.
  • Anything touching real metered payments, agent auth, or the claim flow is MONEY-PATH → money-path-reviewer before it lands anywhere outside a hackathon-only branch.
  • bun run check incrementally for TS; this is necessary but NOT sufficient — live verification per phase is the actual completion bar for this goal.
  • Route work per CLAUDE.md conductor table: api-ts-dev for all of Phases 1-3 (all api-ts/), main loop handles Phase 4 packaging + synthesis directly.
  • Re-verify every file:line anchor before editing — this backlog is a point-in-time snapshot from 2026-09-26.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.