P2P construction contracts using Bitcoin, Taproot/MuSig2 and DLCs

Hi I'm exploring the design of an open-source, peer-to-peer protocol/application for construction contracts, particularly for environments where traditional escrow, arbitration and contract enforcement are expensive or unreliable.

The basic idea is to have no central server, custodian or marketplace. Clients would communicate directly over Tor/DHT and exchange the contractual information P2P: scope of work, milestones, deadlines, drawings/specifications, commercial terms, payment schedules, contractor guarantees, etc. The complete technical and commercial specification would be committed to through hashes.

Once the owner and contractor agree, the application creates a Bitcoin contract using Taproot/MuSig2. Funds would be compartmentalized per milestone rather than putting the entire project into a single pool. The contract would define the possible states and transitions in advance: milestone completion and partial payment, mutual agreement, disagreement, partial completion, abandonment, timeout, guarantee release/return, and potentially arbitration.

The normal path would be cooperative settlement using MuSig2. An arbitrator would not be a normal signer; instead, arbitration would be available through an alternative Taproot branch after predefined conditions or timeouts. The idea is that most disputes should ideally be resolved directly between the parties, with arbitration acting as a last resort.

The documents, communication and evidence would remain off-chain, while Bitcoin would primarily be used for settlement and enforcement of the economic consequences defined by the contract.

I'm also exploring an optional second layer using DLCs for foreign-exchange risk. For example, a contractor could hedge the exchange-rate exposure of a Bitcoin-denominated contract with a third party, using multiple independent oracles.

The goal is not to operate a marketplace, custody funds, charge fees, run an oracle, or maintain any centralized infrastructure. The intention would be to publish the protocol and a reference implementation in Rust under the MIT license, potentially with desktop and Android/F-Droid clients.

I'm not looking for implementation advice yet; I'm primarily looking for protocol-level criticism.

I'd particularly like to know:

Does the fundamental contract model make sense from a Bitcoin/cryptography perspective?

Are there obvious problems with the Taproot/MuSig2/state-machine approach?

Does the incentive and dispute model actually work? In particular, guarantees, partial completion, strategic defaults and the use of an arbitrator as a last resort.

this useful in practice? Or are there fundamental economic, legal, UX, privacy, or technical problems that make this approach impractical?