1. Q1 at a Glance
2025 carryover projects closed this quarter:
Blackbox completed with an excellent evaluation, delivering CKB’s first systematic hardware POS usability study. HashThis was terminated after its core on-chain functionality failed to reach a usable state despite multiple rounds of technical support and remediation (details in Section 3). Fiber Link exceeded Spark’s scope during the review process, graduated to the Community Fund DAO, passed the vote, and is now in milestone execution.
2026 new applications (as of mid-March):
| Project | Direction | Status | Grant |
|---|---|---|---|
| fiber-checkout | Fiber payment React component | Approved | $1,000 |
| ckb-probe | CKB node eBPF diagnostics | Approved | $1,000 |
| Nervos Brain | AI developer onboarding engine | Approved | $2,000 |
| CKB Dev Doctor | CLI environment diagnostics | Pending | — |
| CKB-UGAP | Game achievement protocol (JoyID + Spore) | Pending | — |
| DAO Live Widget | Discourse governance widget | Pending | — |
| Scryve | Expert testing of existing product | Rejected | — |
7 new applications in Q1, compared to 19 across the entire 8 months of 2025. Application rate has roughly tripled.
Q1 outcome summary across both carryover and new projects: 1 completed (Blackbox), 1 terminated (HashThis), 1 graduated to DAO (Fiber Link), 3 approved and in progress, 3 pending revision, 1 rejected.
2. Ecosystem Insights: Three Directions Emerging Without Top-Down Guidance
No one told these builders what to build. They arrived independently, from different countries, with different backgrounds. Yet their proposals cluster around three clear directions.
Direction 1: Fiber’s missing access layer.
fiber-checkout builds a React component so any web developer can add a “Pay with Fiber” button. Fiber Link builds a Discourse tipping plugin using Fiber for off-chain settlement. Blackbox builds a physical POS terminal supporting Fiber payments. Three projects, three entry points (frontend component, Web2 plugin, hardware device), one signal: Fiber’s protocol layer is gradually ready, but the tooling that lets ordinary developers and end users actually use it does not yet exist.
Direction 2: Developer experience remains the primary bottleneck.
ckb-probe provides system-level node diagnostics that CKB currently lacks entirely. CKB Dev Doctor automates environment setup validation for new builders. Nervos Brain uses RAG-based AI to answer CKB-specific development questions that general models get wrong. All three projects target the same friction: getting started on CKB is still hard, and developers are building their own onramps.
Direction 3: Application-layer experiments on Web5 and CKB’s unique architecture.
CKB-UGAP proposes a cross-game achievement protocol using JoyID account abstraction and Spore DOB, turning game achievements into portable digital identity rather than tradable financial assets. DAO Live Widget embeds governance data directly into Nervos Talk. These projects are early attempts to answer “what can you build on CKB that you can’t build elsewhere.”
What this means: Builders are drawing the contours of an ecosystem strategy from the bottom up. The directions they’re converging on (Fiber access tooling, developer experience, Web5 applications) are not random. They reflect where the real gaps are. Whether this bottom-up signal gets incorporated into top-down planning is outside Spark’s scope, but the signal itself is worth documenting.
3. Quality Control: Filtering, Correcting, and Terminating
A grant program that only reports successes is not being honest. Spark’s value lies equally in what it funds, what it rejects, and what it shuts down when delivery falls short.
HashThis, Terminated. The project proposed a lightweight proof-of-existence dApp anchoring file hashes to CKB. The committee supported the concept and provided sustained technical assistance across three review cycles: first flagging an outdated SDK and guiding migration to CCC; then identifying that on-chain timestamps were user-editable (destroying the proof’s value) and that the fee model was unsustainable; finally, after the developer attempted fixes, independently verifying that the core transaction construction still failed consistently on testnet. The committee also observed that the developer relied heavily on AI-generated code without sufficient conceptual grasp of the product — as committee member Hanssen noted, AI accelerates work you already understand, but cannot substitute for understanding you haven’t built. After exhausting remediation options, funding was terminated and remaining funds returned.
Scryve, Rejected. Proposed structured expert testing of an already-functioning product. Spark funds prototypes and early validation, not QA audits.
CKB-UGAP, Pending major revision. Title said “Feasibility Study,” deliverables described a full MVP. Requested $1,500 against a $1,000 pure-tech ceiling. Heavy promotional language. The committee asked the applicant to reconcile the scope contradiction, provide reproducible verification, and ground the proposal in substance rather than rhetoric.
SoMo, Penalized (2025). Technical MVP delivered, but promised seed user testing was not conducted. The committee paid only the technical portion and deducted 10% of total funding. Precedent established: technical delivery alone is insufficient; commitments must be kept in full.
These cases collectively define what Spark considers fundable: a genuine understanding of the problem, a realistic scope, honest communication, and the ability to deliver working software — not just proposals and code that look plausible on the surface. In a landscape where AI has dramatically lowered the cost of producing both proposals and code, this bar matters more than ever.
4. The Spark → DAO Pipeline
Two projects have now graduated from Spark to the Community Fund DAO:
WarSpore · Saga completed seed user testing through Spark and submitted a Community DAO proposal for its next phase.
Fiber Link was reviewed by Spark, but the team determined (and the committee agreed) that the project’s scope and budget fit the DAO better. It passed the DAO vote and is now in milestone execution.
This pipeline is working as intended. Spark validates the idea and the team at low cost; the DAO funds the scale-up. The two programs are complementary, not competing.
5. Operational Notes
Platform migration from Discord to Nervos Talk is underway. All 2026 applications are now processed on Talk with the Spark-Program tag.
Payment options now include 100% USDI (previously 50/50 mixed and 100% CKB).
“How to Verify” requirement is now standard for all applications. Applicants must describe a reproducible, low-cost way for the committee and community to verify deliverables.
Funding tier policy update: Projects with complexity significantly exceeding their category ceiling may cite the program’s overall $2,000 cap with detailed justification. Nervos Brain was the first project approved under this policy.
Spark Program Committee
March 2026
星火计划 2026 Q1：Builder 们在告诉我们什么
1. Q1 概览
2025 结转项目本季度收尾：
Blackbox 以优秀评级完成结项，交付了 CKB 生态首个系统化硬件 POS 可用性研究。HashThis 在经过多轮技术支持和修复尝试后，因核心链上功能始终未能达到可用状态，资助已被终止（详见第 3 节）。Fiber Link 在评审过程中因 scope 超出 Spark 范围，毕业至 Community Fund DAO，通过投票后现正在执行 milestone。
2026 新申请（截至 3 月中旬）：
| 项目 | 方向 | 状态 | 资助金额 |
|---|---|---|---|
| fiber-checkout | Fiber 支付 React 组件 | 已批准 | $1,000 |
| ckb-probe | CKB 节点 eBPF 诊断工具 | 已批准 | $1,000 |
| Nervos Brain | AI 开发者 onboarding 引擎 | 已批准 | $2,000 |
| CKB Dev Doctor | CLI 环境诊断工具 | Pending | — |
| CKB-UGAP | 游戏成就协议（JoyID + Spore） | Pending | — |
| DAO Live Widget | Discourse 治理挂件 | Pending | — |
| Scryve | 现有产品的专家测试 | 已拒绝 | — |
Q1 收到 7 份新申请，而 2025 全年 8 个月共 19 份。申请频率大约翻了三倍。
Q1 综合结果（含结转项目）： 1 个完成（Blackbox）、1 个终止（HashThis）、1 个毕业至 DAO（Fiber Link）、3 个已批准进行中、3 个等待修改、1 个拒绝。
2. 生态洞察：三个自发聚集的方向
没有人告诉这些 builder 应该做什么。他们来自不同国家、不同背景，独立提交申请。但他们的提案聚集在三个清晰的方向上。
方向一：Fiber 缺少接入层。
fiber-checkout 做 React 组件让 web 开发者一键集成 Fiber 支付。Fiber Link 做 Discourse 打赏插件，用 Fiber 进行链下结算。Blackbox 做支持 Fiber 支付的物理 POS 终端。三个项目、三个入口（前端组件、Web2 插件、硬件设备），指向同一个信号：Fiber 的协议层已经逐步就绪，但让普通开发者和终端用户能用起来的工具层尚有欠缺。
方向二：开发者体验仍然是首要瓶颈。
ckb-probe 提供 CKB 目前完全缺失的系统级节点诊断能力。CKB Dev Doctor 自动化新开发者的环境配置验证。Nervos Brain 用 RAG AI 回答通用模型答不对的 CKB 开发问题。三个项目瞄准同一个摩擦：在 CKB 上起步仍然很难，开发者在自己造入口。
方向三：Web5 和 CKB 独特架构上的应用层实验。
CKB-UGAP 提出用 JoyID 账户抽象和 Spore DOB 做跨游戏成就协议，将游戏成就变成可携带的数字身份而非可交易的金融资产。DAO Live Widget 将治理数据直接嵌入 Nervos Talk。这些项目是对"CKB 能做什么别的链做不了的事"这个问题的早期回答。
这意味着什么： Builder 们正在自下而上地勾勒出一幅生态战略的轮廓。他们聚集的方向（Fiber 接入工具、开发者体验、Web5 应用）不是随机的，反映的是真实的缺口在哪里。这些自下而上的信号是否被纳入自上而下的规划，不在 Spark 的权限范围内，但信号本身值得记录。
3. 质量管控：筛选、纠偏与终止
一个只汇报成功的资助项目是不诚实的。Spark 的价值同样体现在它资助了什么、拒绝了什么，以及在交付不达标时关停了什么。
HashThis, 已终止。 该项目提出构建一个轻量级存在性证明 dApp，将文件哈希锚定到 CKB。委员会认可这个方向，并在三轮评审中提供了持续的技术支持：第一轮指出 SDK 过时并引导迁移至 CCC；第二轮发现链上时间戳由用户端填写（从根本上破坏了证明的价值），且费用模型不可持续；第三轮在开发者尝试修复后，委员会独立验证核心交易构造在测试网上仍持续失败。委员会同时观察到开发者严重依赖 AI 生成代码，对产品缺乏足够的概念性理解。正如委员会成员 Hanssen 指出：AI 加速的是你已经理解的工作，它不能替代你尚未建立的理解。在穷尽修复路径后，资助被终止，剩余资金退回。
Scryve, 已拒绝。 申请对一个已运行产品进行结构化专家测试。Spark 资助的是原型开发和早期验证，不是 QA 审计。
CKB-UGAP, 等待重大修改。 标题写"可行性研究"，交付物描述完整 MVP。申请 $1,500 超出纯技术项目 $1,000 上限。大量宣传性语言。委员会要求申请人统一 scope、提供可复现验证方案、以实质内容而非修辞锚定项目定位。
SoMo, 已扣款（2025 年）。 技术 MVP 交付合格，但未执行承诺的种子用户测试。委员会仅支付技术部分资金并扣除总资金 10%。先例已建立：仅有技术交付不够，承诺必须完整兑现。
这些案例共同定义了 Spark 认为值得资助的标准：对问题的真实理解、合理的范围设定、诚实的沟通，以及交付可工作的软件的能力。不是仅仅看起来可信的提案和代码。在 AI 大幅降低了产出提案和代码成本的今天，这条线比以往任何时候都重要。
4. Spark → DAO 毕业管道
两个项目已从 Spark 毕业至 Community Fund DAO：
WarSpore · Saga 通过 Spark 完成种子用户测试后，提交了 Community DAO 提案进入下一阶段。
Fiber Link 经 Spark 评审后，团队判断（委员会同意）项目的 scope 和预算更适合 DAO。该项目通过 DAO 投票，现正在执行 milestone。
这条管道按设计运转：Spark 以低成本验证想法和团队，DAO 资助规模化。两个项目互补而非竞争。
5. 运营更新
平台迁移： 从 Discord 迁移至 Nervos Talk 已启动，2026 年所有申请在 Talk 上以 Spark-Program 标签处理。
支付选项： 现支持 100% USDI（此前仅 50/50 混合支付或100% CKB ）。
验证要求： “How to Verify” 现为所有申请的标准要求。申请人须说明委员会和社区如何以低成本方式复现和验收交付物。
资金层级政策更新： 复杂度显著超出其类别上限的项目可引用项目整体 $2,000 上限，需提供详细论证。Nervos Brain 是该政策下首个获批项目。
星火计划委员会
2026 年 3 月