Skip to content

Forepay Instawards — build plan, verified facts, and what is out of scope #30

Open

Description

@Lexirieru

The map for the Forepay Instawards build. Scope comes from instawards/Forepay Instawards SOW.pdf. 29 issues, four milestones, four owners.

Milestones

Milestone Window Gate
M1 — Feasibility Oct 9–16 One real AdSense revenue proof verified on testnet
M2 — Advance contract Oct 17–23 A valid proof sizes and disburses USDC on testnet
M3 — Relay and front end Oct 24–30 A creator completes the flow in a browser
M4 — Demo and negative cases Oct 31 – Nov 8 Full pipeline + three negative cases, video ≤3 min, evidence pack

Owners

Who Issues What
@Lexirieru #2 #3 #6 #8–#15 #18 #23–#27 Advance contract, digest canonicalisation, evidence
@Lin1er #1 #4 #5 #16 #17 #18 AdSense provider, proof relay, backend, CI
@AncungAulia #19 #20 #21 #22 Creator front end
@Nabil-Aufa #7 #28 #29 Footage, demo video, Chapter Lead

Four things I checked that are not in the SOW

Verified against live sources on 2026-10-09, not taken from the proposal text.

1. The Reclaim verifier is real and live ✅ — testnet CA3EMXR6JOOTNP44T3OAJFMMMGKRRETDJKBLZP2RU3SIY4SDFAH54DU5, mainnet CD4M2KHW3ESOV3RUT7KCTC6BX37PIL2Z3BEK47IA74KIMFIFUI3JJDMO. Testnet instance created 2025-12-21. The SOW's central feasibility claim holds.

2. verify_proof does far less than the SOW sentence implies ⚠️ — this resizes Deliverable 1.

verify_proof(message_digest: BytesN<32>, signature: BytesN<64>, recovery_id: u32)

It recovers a secp256k1 signer and checks it is a witness in the current epoch. It does not parse a claim, build the digest, or extract the revenue figure or bound address. We must canonicalise the claim into exactly the bytes the witness signed. That is the sharpest edge in the project, it is unnamed in the SOW, and it is #3 — scheduled before the gate on purpose.

3. The example repo the SOW names is a 404 ⚠️ — zkfetch-stellar-example does not exist. Use reclaimprotocol/reclaim-stellar-example (pushed 2026-09-11). See #4.

4. Testnet is protocol 29, not 27 ⚠️ — pin soroban-sdk = "29". See #1.

Also confirmed: USDC testnet SAC CBIELTK6YBZJU5UP2WWQEUCYKLPU6AUNZ2BQ4WWFEIE3USCIHMXQDAMA = USDC-GBBD47IF6LWK7P7MDEVSCWR7DPUWV3NY3DTQEVFL4NAT4AQH3ZLLFLA5. A SAC transfer to a G… address fails without a USDC trustline — covered in #12 and #19.

Explicitly out of scope — SOW section 4.1

No issues exist for these, so nobody builds them by accident:

Excluded Why
Real underwriting, credit risk, default or recovery The durable business, not a 30-day deliverable
Licensing, KYC, AML, mainnet funds Consumer lending is regulated; testnet only
Blend integration Permissioned-pool dependency and audit surface; funding backend comes later
Revenue sources beyond AdSense Each provider is its own build
Third-party audit, production hosting, attestor SLAs Budget is development and testing only

The scaffolded landing-page/ is not an Instawards deliverable and has no issue. Say if you want one and it will be added labelled out-of-sow.

Open questions for @Lexirieru

  1. The 30-day clock. The SOW suggested 2026-08-14; acceptance is now. Milestones are dated 2026-10-09 → 2026-11-08. Confirm, or give the real window and every milestone shifts.
  2. @Nabil-Aufa's email — the SOW lists nabilaufa2005@gmail.com; Sterun and Sterish use sharkzneverending@gmail.com. Same person?
  3. Nabil's load is light until Week 4, because the SOW gives him no landing page or design deliverable. He has footage from Week 1 (Capture footage as each milestone lands #7) so the video is not scrambled at the end, the video (Demo video, 3 minutes or less #28), and Chapter Lead coordination (Final demo with the Chapter Lead, and the submission #29). No scope was invented to fill the column — say if you want more.

The one that matters most

#5 is the make-or-break feasibility gate, due 16 October. Everything downstream assumes it. If it does not pass, @Nabil-Aufa tells the Chapter Lead in Week 1 — not Week 4.

Activity

  1. self-assigned this
    on Oct 9, 2026
  2. added
    gateBlocks everything downstream if it fails
    on Oct 9, 2026
  3. pinned this issue on Oct 9, 2026
  4. Nabil-Aufa commented on Oct 10, 2026

    @Nabil-Aufa
    Collaborator

    Answers to the open questions

    1. The 30-day clock: the sprint started 2026-10-02, when the funding landed, and the deadline is 2026-10-30. The plan's 2026-10-09 → 2026-11-08 window is 9 days too late, so every milestone shifts earlier:
    2. Email: yes, both addresses are mine. Same person.
  5. AncungAulia commented on Oct 15, 2026

    @AncungAulia
    Collaborator

    Ownership change, agreed between Ancung and Nabil (2026-10-15).

    Who Now owns
    @Nabil-Aufa #19 #20 #21 #22, the creator front end (M3), in addition to #7 #28 #29
    @AncungAulia #54, the landing page (out-of-sow)

    The "Owner" line and assignee on #19 to #22 are updated. Nothing else about those issues changed: same branches, same scope, same M3 window (24 to 30 October).

    Handoff for Nabil: every screen and state for #19 to #21 is designed in docs/design/forepay.pen, merged in #52. Read docs/design/brief.md first.

    Worth watching: #22 deploy lands at the end of M3, right before #28 (video) and #29 (Chapter Lead) in M4.

    🤖 Generated with Claude Code

  6. Lexirieru commented on Oct 15, 2026

    @Lexirieru
    CollaboratorAuthor

    Role swap recorded, and one consequence worth a decision

    @AncungAulia and @Nabil-Aufa have swapped: Nabil takes the creator front end, Ancung takes the landing page (#54, correctly labelled out-of-sow with its own guardrail that it must not pull time from M3 or M4).

    The swap itself is clean. The load it leaves is not, and it is a PM call rather than mine.

    Where the work sits now

    Who Open What
    @Nabil-Aufa 7 #19 #20 #21 #22 (the whole creator front end), #7 footage, #28 video, #29 Chapter Lead
    @Lin1er 4 #4 #5 #16 #17
    @Lexirieru 2 #27, #30
    @AncungAulia 1 #54 — not an Instawards deliverable

    Why that is a risk rather than just uneven

    Deliverable 2's evidence is "live demo link, screenshots" — that is #22, and it depends on #19 → #20 → #21. All four are now one person's.

    Deliverable 3's video (#28) and the Chapter Lead demo (#29) are the same person's.

    So one person is the single point of failure for two of the three deliverables, with M3 due 30 October and M3 currently 3/8.

    Meanwhile the person who just designed those exact screens — #52, with every state and failure named — has moved off the critical path to a page that is explicitly not a deliverable.

    What I would do

    Split the front end rather than hand over all four. @AncungAulia wrote the design brief for #19–#21 and knows the states better than anyone; keeping one or two of them with her costs nothing in ramp-up and removes the single point of failure. #54 stays hers and stays last.

    Not reassigning anything myself — @Lexirieru, your team and your call. Flagging it now rather than on 29 October.

    Unaffected by this

    M4's contract side is done: #23 #24 #25 #26 closed, with the harness at 7/7 GREEN and real USDC moving. #27 is the evidence pack and is mine. The front end is the only thing between us and Deliverable 2.

  7. Nabil-Aufa commented on Oct 16, 2026

    @Nabil-Aufa
    Collaborator

    @Lexirieru, the creator front end status (#19, #20, #21), and what only you can unblock.

    Ready for review, stacked: merge in this order

    1. #19 Connect a wallet, and check the USDC trustline before anything else #56 (Front end: connect a wallet, and check the USDC trustline #19), connect and USDC trustline. Closes Front end: connect a wallet, and check the USDC trustline #19. Tested with a real Freighter wallet on testnet.
    2. #20 Prove revenue, see the offer with its derivation, accept #57 (Front end: prove revenue, see the offer, accept it #20), prove, offer with its derivation, accept. Refs Front end: prove revenue, see the offer, accept it #20. It does not close the issue yet, see below.
    3. #21 Repayment and history, read from the ledger #58 (Front end: view repayment #21), repayment and history. Refs Front end: view repayment #21. It does not close the issue yet, see below.

    Each PR passed a strict review against the pen file and the brief (Playwright at 1440 and 390, plus real Chrome). Every commit typechecks on its own, and CI is green. All of this front-end work is still testnet-unproven past onboarding: #20 and #21 run on a Reclaim mock, API fixtures and a stubbed RPC in e2e.

    What blocks testnet, on your side

    1. set_claim_template on CCQFL2…I2HS. Until it is set, every proof is refused with 10 TemplateNotSet, so 2b, 2c and 2d cannot happen with a real wallet. It needs the first real AdSense proof (FEASIBILITY GATE: one real AdSense revenue proof verified on testnet #5).
    2. USDC liquidity. The contract holds 5 USDC, so the largest payable advance is 5 USDC and the 100 USDC cap is unreachable. A top-up needs a human at faucet.circle.com (captcha). Can you, or should I?
    3. A getter for the cap would let the front end stop hard-coding 100 USDC in the offer's derivation (Front end: prove revenue, see the offer, accept it #20). Optional, your call.

    Open items

  8. Lin1er commented on Oct 20, 2026

    @Lin1er
    Collaborator

    @Lexirieru @Nabil-Aufa backend status (@Lin1er), 2026-10-20.

    Ready for review, in this order

    1. #16 /advances and /transactions, from Horizon and the contract's own record #66 (Backend API the front end needs #16): /advances and /transactions, verified live. This is the last open item in M2, which is due 23 Oct.
    2. #4 Provider side of #6: the contract's context message, no date matches #74 (Custom Reclaim AdSense provider: trailing revenue out, credentials redacted #4, provider side of Freeze the interface everyone builds against #6): it's independent of the others, so it can go in any time.
    3. #17 Proof relay: POST /proofs, bound to the borrower inside the signed claim #73 (Proof relay: carry the proof and bind it to a Stellar address #17 relay): POST /proofs and GET /proofs/:id, stacked on #16 /advances and /transactions, from Horizon and the contract's own record #66.
    4. #17 Front end relay client behind proveRevenue #75 (Proof relay: carry the proof and bind it to a Stellar address #17 front end client): stacked on Nabil's #20 Prove revenue, see the offer with its derivation, accept #57, and it only touches the proveRevenue seam.

    Under the working agreement, if verified work hasn't had an ACC within two days, I merge it and say so in the PR. For #66 that point is about 21–22 Oct.

    Blocked

    Blocker Blocks Who
    Contract side of the date/context encoding (#6) #61, any real proof on chain @Lexirieru
    No READY AdSense account (#5) #4, #5, #17, #61, the demo everyone, see #5
    Google OAuth Web client id for the sign-in popup #75 running for real whoever owns the Google Cloud project, see #5
    Somewhere to host the relay (~330 MB of circuit files, ~10 min of CPU per proof; not serverless) #75, #22 I'm handling this
  9. Lin1er commented on Oct 20, 2026

    @Lin1er
    Collaborator

    @Lexirieru @Nabil-Aufa @AncungAulia I opened three issues for the blockers, so each one has an owner:

  10. Lin1er commented on Oct 25, 2026

    @Lin1er
    Collaborator

    For the risk list, from #93 (@Lexirieru's note there):

    Per-creator exposure is 2× the cap per window once AdMob lands (#94). The AdMob nullifier is tagged with its source, so one pub- id can take an AdSense advance and an AdMob advance for the same window. That is intended: website and in-app ad revenue are different money. It does mean the limit per creator per window is 2 × 100 USDC, not 100 USDC. Nothing changes on testnet. It becomes a real limit question past this grant. Post-SOW, out-of-sow.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

gateBlocks everything downstream if it fails

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions