Mirrored from filpgf.io — ProPGF Batch 3 (Karma program 1479, application 6a347406f5b633d50adb3bc7, status: pending). Contact details redacted; canonical application lives on filpgf.io. 1.1 Project Name Curio 1.2 Project Github https://github.com/filecoin-project/curio 1.4 Team Lead/Point of Contact Jennifer W, FilOz 1.5 Category [ "Core Infrastructure" ] 1.6 Open Source Status Fully Open Source 2.1 Project Summary Curio is the go-to infrastructure that enables Filecoin storage providers to maintain power and reliably execute paid storage deals. Maintained by the FilOz team, this proposal funds the continued development and maintenance of Curio for a 6-month period (H2 2026), covering four critical workstreams: 1. Network upgrade support, keeping every SP running Curio compatible with mainnet upgrades; 2. Ongoing maintenance and hardening — bug fixes, security and bug-bounty response, and SP community support (including PoRep pipelines); 3. Curio architecture and feature work that lets SPs run a single, lightweight runtime to provide paid services in FOC and beyond; 4. Next-generation PoRep research with the FilOz research team. *Curio is credibly neutral core infrastructure. As a client implementation under FilOz, engineering priorities are driven solely by what is best for the Filecoin network — not by any commercial interest. This is the same model used by credibly neutral client teams in Filecoin (e.g., lotus, Forest) and other L1 ecosystems (e.g., Geth, Lighthouse, Prysm in Ethereum). FilOz has allocated ~3 engineers focused on SP product and Curio (they may also support other FilOz initiatives); other FilOz members contribute to Curio as well, so for budget planning we use ~3 FTE. This grant covers the engineering cost of Curio-specific development, essential infrastructure (test servers, CI), and a small portion of organizational cost.* 2.2 Who does this work support? [ "Pods", "Onramps", "Storage Providers", "Application Builders", "Network Infrastructure" ] 2.3 Total Funding Requested (USD) $320,000 USD & 67,200 FIL 2.4 Milestones & Budget [ { "title": "\"make curio-pdp\" — Lightweight, PDP-First Build of Curio", "description": "Produce a simplified Curio PDP binary.", "dueDate": "2026-08-30", "fundingRequested": "$80,000 USD & 16,800 FIL", "completionCriteria": "- decouples PoRep sealing overhead to create a lightweight binary optimized for Proof of Data Possession (PDP) workloads, while preserving both PDP and PoRep as runtime options so operators can scale into full SP capability without switching software; \n- enables object-store database as yugabyte alternative;\n- enables simple deployment of a light Curio (see https://golantern.io/ as a reference);\n- hardening and scalability: enabling higher onchain throughput and high availability(zero downtime upgrades)\n- be able to communicate w/ fil.one client;" }, { "title": "NV29 Development", "description": "Support NV29 with compatible Curio releases, including engineering toward NV29 candidate features.", "dueDate": "2026-10-31", "fundingRequested": "$80,000 USD & 16,800 FIL", "completionCriteria": "Based on current knowledge, NV29 candidates may include:\n- FIL+ reform\n- F3 reboot\n- TBD\n\nDue Date may extend to Nov 2026 (depends on the NV29 roadmap)" }, { "title": "next-gen PoRep Research", "description": "Work with FilOz protocol designers to investigate a more storage-product-friendly PoRep.", "dueDate": "2026-11-15", "fundingRequested": "$80,000 USD & 16,800 FIL", "completionCriteria": "Curio to deliver the relevant research and design work." }, { "title": "Lotus-miner Deprecation Plan", "description": "Produce a lotus-miner deprecation and Curio migration plan.", "dueDate": "2026-12-31", "fundingRequested": "$80,000 USD & 16,800 FIL", "completionCriteria": "Curio delivers plan targeting end-of-life support for lotus-miner in 2027." } ] 3.1 Impact pathway <u>**<u>Objectives: </u>**</u> **Objective 1:** Drive Paid Onchain Deals: By giving existing and new SPs a PDP-first binary that is easy to onboard and operate, and be able to onboard FOC/fil.one deal onto a single maintained runtime, this work streamlines SP operations — allowing service providers to have a consistent stack and focus on delivering service rather than running a different piece of software for each Filecoin program. SP operational excellence is required to attract paying demand. **Objective 2:** Strengthen Network Profitability & Cryptoeconomics: By keeping Curio stable, performant, and compatible with network upgrades, this work directly preserves the QAP that secures the network and sustains SP profitability. The lightweight PDP-first build lets SPs offer provable storage services with minimal operational overhead, lowering SP OpEx and reducing FIL outflow while creating new revenue streams beyond block rewards. The next-gen PoRep research may let SPs onboard power for consensus while running a viable storage service on the same hardware. A clear path to deprecate lotus-miner and Boost and migrate their users to Curio could save the network up to ~$150,000 in maintaining an outdated software stack. **Objective 3:** Scale Paid Onchain Flagship Client Adoption: The Filecoin Onchain Cloud (FOC) is a flagship GTM channel and the foundation for Fil.One. FOC depends on Curio's PDP proving infrastructure for proof of data possession, its data pipeline for storage execution, and its retrieval infrastructure for content serving. The PDP-first build support are being designed in direct coordination with the FOC & Fil.One pod to consolidate their onchain storage data path onto a single maintained runtime. Scaling FOC adoption is gated on Curio's reliability and feature completeness. **<u>Impact Pathway:</u>** **Output →** Stable Curio releases with a PDP-first build, fil.one Client API support, production PDP proving, network upgrade support (including NV29 PoRep candidates), and continuous maintenance. **Outcome →** Service providers can onboard onto a painless, PDP-first runtime with minimal operational overhead, while retaining a clear path to full PoRep SP capability. SPs gain new revenue opportunities through provable storage services and lower sealing costs. **Impact →** Paid onchain data volume increases as the friction to onboard and provide verifiable Filecoin storage services is reduced. Network QAP is preserved through reliable upgrade support and a clear lotus-miner→Curio migration path. The FOC flagship client pipeline accelerates, driving enterprise adoption and FIL in paid deals. 3.2 Verification metrics | Metric | Data Source | How Measured | Target (End of Grant) | |--------|------------|--------------|----------------------| | Network upgrade compatibility | GitHub releases | 100% of mainnet upgrades supported with timely Curio releases | 100% upgrade coverage | | SP adoption (Curio users) | FOC SP insights & ProbeLab insights | Number of SPs providing FOC service; number of consensus nodes running Curio vs. legacy implementations | Maintain or grow current Curio SP base | | PDP data storage | On-chain PDP data, FOC metrics | Volume of data onboarded / payment of PDP-based pipeline | [depend on pod integration, e.g, fil.one ] | | PDP proving reliability | On-chain PDP proofs, monitoring dashboards | PDP proof submission success rate on mainnet | ≥95% proof success rate | | Bug resolution time | GitHub issues/PRs | Time from critical bug report to merged fix | Critical bugs resolved within 72 hours | 3.3 References #### Reference 1 Name: FOC Pod Relationship: Curio's PDP and deal pipeline are direct dependencies for FOC's onchain storage products. #### Reference 2 Name: LDO Pod Relationship: Curio's PoRep pipeline are direct dependencies for LDO onchain storage products. #### Reference 3 Name: FWSS SPs (TippyFlits, Milad, Mongo & etc.) Relationship: Curio software enables SPs to provide data storage service #### Reference 4 Name: Magik from Aurora Relationship: Curio software is one of the main components of their PoRep-based pipeline. 4.1 Monthly Operating Burn [ "$10-$100K (small team)" ] 4.2 What % of total team monthly burn depends on this grant? ~100% of the Curio-dedicated engineering cost. 4.3 If this grant is not awarded, what happens? Without this grant, FilOz loses the funding to sustain its ~3 dedicated Curio engineers. They would need to be reallocated to other FilOz priorities or let go, and Curio development would drop to best-effort maintenance with no dedicated capacity. Concretely: (1) network upgrade releases would be delayed or unavailable — SPs running Curio risk being unable to participate in protocol transitions, threatening network QAP; (2) PDP software development and new feature support for FOC would halt, stalling SP operations for Filecoin onchain storage services; (3) PDP proving improvements stop, directly blocking SPs' ability to offer verifiable onchain storage at scale and risking the loss of paying customers; (4) bug fixes and security patches would be severely delayed, increasing the risk of SP outages, missed block production, data loss, and security exposure. 4.4 Core Team The Curio engineering team was originally at Curio Storage Inc. and has since been absorbed into FilOz for long-term organizational stability and tighter alignment with Filecoin core development. FilOz allocates roughly 3 FTE focused on Curio development and maintenance (some of whom also support other FilOz initiatives), with project leadership provided by FilOz management. **Jennifer Wang & Molly Mackinlay** — Co-founders FilOz. Oversees Curio's roadmap and SP strategy with other network stakeholders; **@zenground0** — Senior Engineer, FilOz. Leading Curio sub-team at FilOz. core contributor of FIlecoin protocol actors, lotus & FOC since 2021 **@snadrus (Andrew)** — Senior Engineer, FilOz. core curio developer, previously leads Lotus-miner team at Protocol Labs. **@LexLuthr** — Engineer, FilOz. core curio developer, previously core contributor of boost at Protocol Labs 4.5 Has your team received a ProPGF grant or funding from PLFIF before? [ "Yes" ] 5.1 Key risks & dependencies **Protocol change velocity.** Filecoin's upgrade cadence may accelerate beyond planned capacity. - *Likelihood:* Medium | *Impact:* Medium - *Mitigation:* Team has successfully supported all upgrades in Batch 2. Monthly milestone structure allows reallocation of engineering time to urgent upgrade work. **FOC timeline dependency.** "Make curio-pdp" and fil.one integration design depend on finalized FOC/fil.one requirements & roadmap. - *Likelihood:* Low | *Impact:* Medium - *Mitigation:* Active coordination with FOC pod & Fil.one pod. **Key person risk.** Three-engineer team with deep specialized knowledge. - *Likelihood:* Low | *Impact:* High - *Mitigation:* All work happens in public GitHub with comprehensive PR documentation. AI agent context documentation added to repo. FilOz organizational structure provides continuity and bus factors than most other network teams. **Upstream dependency changes.** Changes in filecoin-ffi, proofs, or actor code may require unexpected adaptation work. - *Likelihood:* Medium | *Impact:* Low - *Mitigation:* Team maintains close coordination with core protocol teams within FilOz and the broader Filecoin ecosystem, and has deep familiarity with the full stack. **Security vulnerabilities.** Zero-day vulnerabilities in Curio or its dependencies could require emergency response. - *Likelihood:* Low | *Impact:* High - *Mitigation:* Continuous security monitoring, rapid response capability (72-hour critical bug SLA), and responsible disclosure process in place. Anything else you want to share that we didn't ask? **Infra-Specific Signals (per Batch 3 evaluation framework):** - **Ecosystem criticality:** Curio is the only actively developed, forward-looking SP implementation and supports SP strategy developments. - **Replacement difficulty:** If Curio lost dedicated engineering, there is no drop-in replacement. Lotus Miner is in maintenance-only mode. Rebuilding or replacing Curio's capabilities — including but not limited to PoRep optimization and FOC service support — from scratch would take many months and add significant SP operational overhead, costing the network millions of dollars. **Organizational stability.** Operating within FilOz gives Curio continuity a standalone startup could not sustain: FilOz handles operational overhead (HR, legal, finance) so engineers focus on Curio, and the project is no longer dependent on a single company's financial and operational health. **Credible neutrality through structure.** As a client implementation team within FilOz, Curio's engineering priorities are set by network needs — not one entity's interests. The team does not take paid consulting or integration work. This structural neutrality is essential for a client implementation that all SPs must trust to handle their sectors, proofs, and deals without bias. Contributing to Core Infrastructure? Curio is the primary actively-developed Filecoin storage provider (SP) implementation, maintained by the FilOz team. It is the ONLY software that allows SPs to produce blocks & participate in consensus; provide PoRep backed service; provide PDP backed service all in one place. As Lotus Miner has transitioned to maintenance-only mode, Curio is the forward-looking implementation that the SP ecosystem is migrating to. Every SP running Curio depends on this team for network upgrade compatibility, bug fixes, security patches, and new protocol features. The service built within Filecoin Onchain Cloud (FOC) is increasingly relying on Curio's data pipeline and service/onchain infrastructure to function as the supply side. Filecoin service providers are also looking for a single piece of software they can deploy in the data center to provide services across the Filecoin platform. Objective 1 Direct Objective 2 Direct Objective 3 Direct Open Source Context Curio is fully open source under the Filecoin Project GitHub organization (MIT + Apache 2.0 dual license). All development happens in the open with publicly reviewable PRs, signed releases, and community-accessible issue tracking.