Openfunderse

Openfunderse is an agent-driven fund protocol for Monad: data claims are attested, intents are validated, and only approved intents can execute onchain.

Consensus Rebalancing Model (Risk-Projected Aggregation)

Let whitelist assets be $\mathcal A={1,\dots,n}$ and participants be $\mathcal P$.

Flow (v0 safe patch):

1. Participants submit target weights $c_{i,t}\in\Delta^n$.
2. Stake-weighted aggregate is projected into a feasible risk set.
3. Strategy executes toward projected target under venue constraints and execution cost.
4. Participant prediction score $g_{i,t}$ is settled on scoring oracle return.
5. Positive NAV alpha mints extra shares; next weights are induced only by share balances.

Fund state:

$$V_t=p_t^\top h_t,\qquad s_t=\frac{p_t\odot h_t}{V_t}\in\Delta^n,\qquad \sum_{i\in\mathcal P} w_{i,t}=1,\;w_{i,t}\ge0.$$

Aggregation and projection:

$$\bar s_t=\sum_{i\in\mathcal P}w_{i,t}c_{i,t}, \qquad s_t^\star=\Pi_{\mathcal R_t}(\bar s_t) =\arg\min_{x\in\mathcal R_t}\|x-\bar s_t\|_2^2.$$ $$\mathcal R_t= \left\{ x\in\Delta^n: \|x-s_t\|_1\le\tau_t,\; \forall j\in\mathcal A,\;x_j\le u_j \right\}.$$

Execution:

$$z_t^\star=\arg\min_{z} \left(\|s_{t+1}(z)-s_t^\star\|_2^2+\lambda\cdot \mathrm{Cost}(z)\right),$$

with per-leg feasibility (example):

$$y_\ell\ge x_\ell\pi_\ell,\qquad x_\ell\in[0,X_\ell],\;y_\ell\in[0,Y_\ell].$$

Oracle model (execution vs scoring):

$$p_t^{\mathrm{exec}}:=P_{\mathrm{twap}}(t;W_{\mathrm{exec}}),\qquad p_t^{\mathrm{score}}:=P_{\mathrm{score}}(t),\qquad r_t^H=\frac{p_{t+H}^{\mathrm{score}}-p_t^{\mathrm{score}}}{p_t^{\mathrm{score}}},$$

where $P_{\mathrm{twap}}$ is execution-safe oracle and $P_{\mathrm{score}}$ is performance-scoring oracle.

Participant prediction score:

$$g_{i,t}=c_{i,t}^\top r_t^H-\mathrm{bench}_t.$$

NAV alpha and mint budget:

$$\alpha_t^{\mathrm{NAV}} =\frac{V_{t+H}-V_t}{V_t}-\mathrm{bench}_t,\qquad M_t=\mu[\alpha_t^{\mathrm{NAV}}]_+N_t.$$

Mint allocation (Sybil-resistant via stake weighting):

$$\phi(g)=[g]_+, \qquad \Delta N_{i,t} =M_t\frac{w_{i,t}\phi(g_{i,t})}{\sum_{k\in\mathcal P}w_{k,t}\phi(g_{k,t})+\varepsilon}.$$

Share-weight identity:

$$N_{i,t+1}=N_{i,t}+\Delta N_{i,t},\qquad N_{t+1}=\sum_{k\in\mathcal P}N_{k,t+1},\qquad w_{i,t+1}=\frac{N_{i,t+1}}{N_{t+1}}.$$

Parameters:

$$\tau_t,\{u_j\}_{j\in\mathcal A},\lambda,\mu,\varepsilon,W_{\mathrm{exec}},\mathrm{bench}_t.$$

*Inspired by stake-weighted subjective-consensus literature (incl. Yuma-style clipping), adapted to portfolio allocation with explicit risk projection and execution constraints. *Operational note: projection $\Pi_{\mathcal R_t}$, score computation, and oracle assembly are offchain (relayer/strategy), while settlement constraints remain onchain-enforced.

MVP Scope

• Included: participant co-investment claims (targetWeights) -> epoch aggregate -> strategy intent propose -> intent attestation/bundle -> ready-for-onchain execution payload.
• Included: index mapping rule targetWeights[i] <-> riskPolicy.allowlistTokens[i]; strategy uses aggregate view to choose rebalance direction.
• Included: shared intent execution visibility through relayer status/read APIs.
• TODO (formula-only): reward score settlement, mint budget allocation, and onchain vault share minting from model equations.
• Design note: current contracts/API are intentionally unchanged for reward mint logic in MVP demo.

What matters first

• packages/contracts: Foundry contracts (ClawFundFactory, IntentBook, ClawCore, ClawVault4626, NadFun adapter)
• packages/relayer: API/aggregation layer for fund events and attestations
• packages/agents: participant/strategy agent runtime
• packages/sdk: shared hashing + EIP-712 utilities

Mainnet Contracts (Verified)

• ClawFundFactory: 0xf16a12fCeC5FD8eb67eEd57f9659AE730734AA74
• MockSnapshotBook: 0x06676F1eE9480085c01BEdb348C60167EBeE0Cc9
• IntentBook (Implementation): 0x417dDEdbECb746e4474B587dF056a1EB42957e80
• IntentBook (Proxy): 0x23403970eD4891b8f85a414B9B00C239D364D16A
• ClawVault4626 (Implementation): 0x2bcB5a88942D474797baB5628c4c04A07E9c1597
• ClawVault4626 (Proxy): 0xd1Cd80B55cD1c77116f2b817356f157C116dDAca
• ClawCore (Implementation): 0x251377f58D2F8e1A296c783Dab54A16f70795417
• ClawCore (Proxy): 0x6e81c81912f7714F2BbeD0a369D00Bd218439988
• NadfunExecutionAdapter (Implementation): 0xc4B1F14B85D4DF9C1cf2d5BCc55a114A2860d553
• NadfunExecutionAdapter (Proxy): 0x418F5A1d728b3e23B6B01A04a4FEEa7894f9b2B2

Package roles

Package	Primary role	Owns	Docs
packages/contracts	Onchain execution and fund governance	UUPS contracts, deployment scripts, Foundry tests	packages/contracts/README.md
packages/relayer	Offchain API and orchestration	v1 API, attest/intent aggregation, execution jobs, storage integration	packages/relayer/README.md
packages/agents	Runtime bots for participant/strategy	Reddit MVP participant flow, bot runtime exports	packages/agents/README.md
packages/sdk	Canonical protocol utilities	Hashing, EIP-712 verification, weighted-threshold helpers	packages/sdk/README.md
packages/openfunderse	Codex skill-pack installer/distribution	openfunderse install CLI, pack manifests/prompts/skills	packages/openfunderse/README.md

Quick start

nvm use
npm install

cd packages/contracts
cp .env.example .env.local
forge build
forge test

Main workflows

• Admin fund bootstrap (factory): ./packages/contracts/scripts/deploy-fund-factory.sh
• Contract deployment: ./packages/contracts/scripts/deploy-clawcore-stack.sh
• Intent preflight check: ./packages/contracts/scripts/validate-intent-call.sh
• Intent dry-run check: ./packages/contracts/scripts/dry-run-intent-call.sh
• End-to-end NadFun flow (optional): ./packages/contracts/scripts/run-nadfun-e2e.sh

Detailed script usage and required envs:

• packages/contracts/scripts/README.md

Documentation index

Use the docs index as the single navigation entry:

• docs/README.md

High-signal docs:

• Manual intent validation: docs/protocol/intent-manual-validation.md
• Protocol hashing/EIP-712: docs/protocol/hashing-eip712-v1.md
• Contracts package guide: packages/contracts/README.md