⚠️ Editorial note: The open source ecosystem in China operates under a distinct institutional framework — characterized by state-led initiatives, intranet-like boundaries, and top-down governance. Readers should be aware that this context differs from the community-driven open source model common in other regions. The term “open source” as used in Chinese media may refer to practices that diverge from the conventional definition.

China Open Source Daily — 2026-09-14

🏛️ Institutional Design Anchor — MIIT’s “AI + Software” Special Action Implementation Plan (人工智能+软件专项行动实施方案): The Seventh Surface of State-Directed Open-Source Institutional Formalization

1. Sina Tech / IT之家 (September 11, 2026) — “工信部印发’AI + 软件’专项行动实施方案：支持软件企业与智能手机、智能家居、智能网联汽车等终端厂商联合创新”

2. Sina Finance (September 11, 2026) — “工信部印发’人工智能+软件’专项行动实施方案，到2028年覆盖2万家规模以上软件企业”

3. Tencent News (September 11, 2026) — “工信部部署’AI+软件’，五年后这些新业态要成为产业新增长极”

4. EET China (September 11, 2026) — “工信部印发《‘人工智能+软件’方案》：2028年覆盖2万家规上软件企业”

5. IT Home (September 11, 2026) — “我国力争到 2030 年关键软件全面实现智能化升级”

MIIT issued on September 11, 2026 — the same week as this series’ September 11 briefing documented the Chen Kaihua Qiushi party-theoretical-flagship placement and the Qinghe Zhishashe independent-institutional-economics diagnostic — a national industrial-policy document: the “AI + Software” Special Action Implementation Plan (人工智能+软件专项行动实施方案). The document is a state-directed industrial-policy instrument on the software-and-information-services industrial surface, and its open-source clauses — the only clauses this briefing analyzes — make five institutionally new moves on the same unsolved object that this series’ prior twenty-four briefings have been tracking.

The two documented numerical targets of the plan, verbatim from the Sina Tech / IT之家 report:

“By 2028, the software and information-technology-services industry’s intelligence-transformation will see significant improvement… incubate 5 or more quality open-source projects (5个以上优质开源项目).”

“By 2030, achieve a new leap in the fusion-development of AI with the software and information-technology-services industry… build multiple open-source communities and industrial clusters with international influence (建成若干具有国际影响力的开源社区和产业集群).”

This is the first documented instance of a national industrial-policy document attaching quantitative open-source-project targets (5+ quality open-source projects by 2028, multiple internationally influential open-source communities by 2030) to the industrial-policy horizon. The September 6 briefing’s AI SME Startup Support Plan (31号) documented open-source-community designation and open-source-contributor recognition; the September 11 plan documented the same open-source surface at the state-directed-open-source-project-selection / quantitative-open-source-project-target / international-open-source-community-target layer — the first time a Chinese national industrial-policy document attaches quantitative open-source targets to the industrial-policy horizon.

The five institutionally new moves documented in the plan’s open-source clauses, extracted verbatim from the Sina Tech / IT之家 report and cross-referenced with the EET China and Tencent News reports:

Move 1 — The Agent Software Application Store (智能体软件应用商店) and Skills Resource Library (Skills 资源库): the first state-directed institutional apparatus for agent-software distribution.

The plan reads: “推动建设智能体软件应用商店和技能包（Skills）资源库，规范平台上架审核与运营管理，促进智能体、技能包广泛应用” (Advance the construction of Agent Software Application Stores and Skill-Pack (Skills) Resource Libraries, standardize platform listing review and operational management, and promote the widespread application of agents and skill-packs).

The same paragraph reads: “培育智能体开源专区，构建多元协同、技能共享的开放生态” (Cultivate an agent open-source zone, constructing a plural-collaboration, skill-sharing open ecosystem).

This is institutionally the sharpest move in the plan’s open-source clauses: MIIT is directing the construction of a state-directed Agent Software Application Store and Skills Resource Library as institutional apparatus for agent-software distribution and skill-pack distribution. The same distribution-surface move this series’ September 3 briefing documented at the Android-app-store layer (the CAC Qinglang Phase 2 announcement naming Huawei, Xiaomi, OPPO, and vivo app stores as AI-app-admission subjects) is now being operationalized at the agent-software layer — a structurally parallel institutional-form move at a different distribution surface. From an institutional economics standpoint, this state-directed-agent-app-store / android-app-store-admission pairing is a structural finding: the state-directed-app-store institutional-form is being operationalized at two distribution surfaces simultaneously — the traditional Android-app-distribution surface (September 3 CAC briefing) and the agent-software-distribution surface (September 11 plan) — with the two surfaces operating in parallel on the same institutional-form object.

Move 2 — Independent open-source license application and compliance guidance (自主开源许可协议应用与合规指引): the first administrative-document use of Mulan-family-license promotion as a national industrial-policy instrument.

The plan reads: “依托全国性开源基金会与人工智能开源社区，引导企业开放智能编程、软件产品智能化、智能体等成果，支持重点开源项目在社区首发，设立智能操作系统等领域开源专区，加强自主开源许可协议应用与合规指引” (Relying on national open-source foundations and AI open-source communities, guide enterprises to open-source their results in smart programming, software-product-intelligence, and agents, support key open-source projects in community-first release, establish open-source zones in fields such as intelligent operating systems, and strengthen independent open-source license application and compliance guidance).

The clause “加强自主开源许可协议应用与合规指引” (strengthen independent open-source license application and compliance guidance) is institutionally the sharpest legal-and-licensing move in the plan: it is the first administrative-document use of Mulan-family-license promotion as a national industrial-policy instrument. From an institutional economics standpoint, this independent-license-compliance / national-industrial-policy pairing is a structural finding: the Mulan-family-license infrastructure (MulanPSL, MulanPSL-2.0, MulanPOM, Mulan-DSL — documented in this series’ August 30 briefing on the Mulan Foundation and in the September 9 briefing on the openKylin × Astron-Agent challenge’s MulanPSL-2.0 licensing) is now being formally placed inside the national industrial-policy document as a compliance-guidance instrument — the Mulan-family-license apparatus is no longer a community-level or foundation-level licensing instrument but is a state-directed national-industrial-policy instrument.

Move 3 — Quality open-source project selection-and-cultivation mechanism with community-first-release (社区首发): the first administrative-document institutionalization of a state-directed open-source-project selection-and-cultivation apparatus paired with a community-first-release operationalization.

The plan reads: “建立优质开源项目遴选和培育机制，鼓励探索多元化商业模式，推动优质项目在制造、金融等领域深度应用” (Establish a quality-open-source-project selection-and-cultivation mechanism, encourage exploration of diverse business models, and promote deep application of quality projects in manufacturing, finance, and other fields).

The clause “支持重点开源项目在社区首发” (support key open-source projects in community-first release) — embedded in Move 2’s paragraph — is institutionally the sharpest structural finding of this cycle on the community-first-release surface: the plan pairs a state-directed project-selection-and-cultivation mechanism with a community-first-release operationalization — the state selects quality open-source projects and requires them to be first-released at a community rather than at an enterprise. From an institutional economics standpoint, this state-directed-project-selection / community-first-release pairing is a structural finding: the plan is not just creating a state-directed open-source-project selection mechanism but is pairing it with a community-first-release operationalization — the same “community-first-release” operationalization this series’ August 19 briefing documented in the context of the AtomGit national-level open-source-community designation, now being operationalized at the state-directed-project-selection surface.

Move 4 — Open-source contributions incorporated into research-evaluation and talent-certification systems (科研评价与人才认定): the sharpest administrative absorption of open-source community norms at the higher-education and research-institute layer.

The plan reads: “完善开源贡献激励，支持高校、科研院所和企业将开源贡献纳入科研评价与人才认定体系，探索将开源贡献作为挖掘软件领域优质中小企业重要参考指标” (Improve open-source contribution incentives, support universities, research institutes, and enterprises in incorporating open-source contributions into research-evaluation and talent-certification systems, and explore using open-source contributions as an important reference indicator for identifying quality SMEs in the software field).

This is institutionally the sharpest move in the plan’s open-source clauses: it extends the September 6 briefing’s star-count-as-administrative-metric move (which placed GitHub-style star counts alongside invention patents and highly-cited papers as one of four national-level evaluation indicators) into a broader research-evaluation-incorporation move at the higher-education and research-institute layer. The September 6 briefing documented the star-count-as-administrative-metric move as the first administrative-document use of the GitHub-style star metric as an official institutional evaluation indicator, positioned alongside invention patents and highly-cited papers (both administrative-metric indicators). The September 11 plan extends that move into a broader research-evaluation-incorporation move — universities, research institutes, and enterprises are all directed to incorporate open-source contributions into their internal research-evaluation and talent-certification systems, and open-source contributions are additionally being explored as a reference indicator for identifying quality SMEs in the software field. From an institutional economics standpoint, this star-count-administrative-metric / research-evaluation-incorporation pairing is a structural finding: the administrative absorption of open-source community norms has moved from the star-count-as-administrative-metric layer (September 6 briefing) to the research-evaluation-incorporation layer (September 11 plan) — the same open-source-contribution object is being absorbed at two administrative-metric layers simultaneously.

Move 5 — Quantitative open-source-project targets (5+ quality open-source projects by 2028, multiple internationally influential open-source communities by 2030): the first national industrial-policy document attaching quantitative open-source targets to the industrial-policy horizon.

The plan’s 2028 target reads: “开放协同的开源社区建设取得积极进展，孵化 5 个以上优质开源项目” (open-collaborative open-source community construction achieves positive progress, incubating 5 or more quality open-source projects). The plan’s 2030 target reads: “建成若干具有国际影响力的开源社区和产业集群” (build multiple open-source communities and industrial clusters with international influence).

The pairing of these two targets is institutionally the sharpest structural finding of this cycle on the quantitative-open-source-target surface: the plan is the first national industrial-policy document attaching quantitative open-source-project targets (5+ quality open-source projects by 2028) and qualitative open-source-community targets (multiple internationally influential open-source communities by 2030) to the industrial-policy horizon. The 2028 target (5+ quality open-source projects) is being paired with the plan’s Move 3 (state-directed open-source-project selection-and-cultivation mechanism with community-first-release) — the quantitative target is being operationalized through the state-directed project-selection-and-cultivation apparatus. The 2030 target (multiple internationally influential open-source communities) is being paired with this series’ prior briefings’ documentation of the Western-foundation-governance-seat surface (PyTorch Foundation, September 10 briefing; AAIF, September 7 briefing) and the party-theoretical-flagship surface (Chen Kaihua Qiushi, September 11 briefing) — the internationally-influential-open-source-community target is being operationalized through the same Western-foundation-governance-seat and party-theoretical-flagship surfaces that this series’ prior briefings have been documenting.

Institutional significance: The AI + Software Special Action Implementation Plan is a structurally new placement on the same unsolved object that this series’ prior twenty-four briefings have been documenting — the institutional identity of Chinese open source across institutional borders — and it is the sharpest documented formalization of open-source as a state-directed industrial-policy instrument at the national-industrial-policy surface.

From an institutional economics perspective, the AI + Software Special Action Implementation Plan matters on five axes, and each of the five axes pairs institutionally with a prior briefing’s documented surface:

First, the state-directed Agent Software Application Store + Skills Resource Library pairing pairs institutionally with this series’ September 3 briefing’s Android-app-store-admission documentation (CAC Qinglang Phase 2, naming Huawei, Xiaomi, OPPO, vivo app stores as AI-app-admission subjects). The state-directed-app-store institutional-form is being operationalized at two distribution surfaces simultaneously — the traditional Android-app-distribution surface and the agent-software-distribution surface — with the same institutional-form object (state-directed app-store) at the two surfaces.

Second, the independent open-source license application and compliance guidance (自主开源许可协议应用与合规指引) move pairs institutionally with this series’ August 30 briefing’s Mulan Foundation documentation and September 9 briefing’s openKylin × Astron-Agent challenge MulanPSL-2.0 licensing. The Mulan-family-license apparatus is no longer a community-level or foundation-level licensing instrument but is a state-directed national-industrial-policy instrument.

Third, the state-directed open-source-project selection-and-cultivation mechanism with community-first-release (社区首发) pairing pairs institutionally with this series’ August 19 briefing’s AtomGit national-level open-source-community designation documentation. The community-first-release operationalization is now being paired with a state-directed project-selection-and-cultivation mechanism.

Fourth, the research-evaluation-incorporation move (open-source contributions into research-evaluation and talent-certification systems) extends the September 6 briefing’s star-count-as-administrative-metric move (GitHub-style star counts as one of four national-level evaluation indicators alongside invention patents and highly-cited papers). The administrative absorption of open-source community norms has moved from the star-count-as-administrative-metric layer to the research-evaluation-incorporation layer.

Fifth, the quantitative open-source-project targets (5+ quality open-source projects by 2028, multiple internationally influential open-source communities by 2030) pair institutionally with this series’ September 10 briefing’s PyTorch Foundation Western-foundation-governance-seat documentation (Alibaba Cloud and Cambricon as Platinum members with Governing Board + Technical Advisory Council seats, Ant Group as Gold member) and September 11 briefing’s Chen Kaihua Qiushi party-theoretical-flagship documentation (proposal for a national-level AI-research open-source community). The internationally-influential-open-source-community target is being operationalized through the same Western-foundation-governance-seat and party-theoretical-flagship surfaces.

Sources:

🔍 Commentary

One institutional-design anchor (MIIT’s “AI + Software” Special Action Implementation Plan) on the same unsolved institutional object that this series’ prior twenty-four briefings have been documenting: the institutional identity of Chinese open source across institutional borders.

This cycle’s briefing documents a structurally new placement on the same unsolved institutional object that the prior twenty-four briefings have been tracking — the institutional identity of Chinese open source across institutional borders — and the placement is a specific state-directed-industrial-policy-instrument move at the national-industrial-policy surface.

The AI + Software Special Action Implementation Plan makes five institutionally new moves on the same unsolved institutional object as this series’ prior briefings — the state-directed Agent Software Application Store + Skills Resource Library pairing (a distribution-surface move), the independent open-source license application and compliance guidance move (a licensing-and-legal move), the state-directed open-source-project selection-and-cultivation mechanism with community-first-release pairing (a project-selection-and-cultivation move), the research-evaluation-incorporation move (a research-evaluation-and-talent-certification move), and the quantitative open-source-project targets pairing (a quantitative-target move).

One structural pattern across the five moves: each move pairs institutionally with a prior briefing’s documented surface, and the five moves together document that the state-directed-industrial-policy surface has crossed into the seventh surface of state-directed open-source institutional formalization.

  • The state-directed Agent Software Application Store + Skills Resource Library pairing pairs institutionally with the September 3 briefing’s Android-app-store-admission documentation (CAC Qinglang Phase 2) — the same state-directed-app-store institutional-form operationalized at two distribution surfaces.
  • The independent open-source license application and compliance guidance move pairs institutionally with the August 30 briefing’s Mulan Foundation documentation and September 9 briefing’s openKylin × Astron-Agent challenge MulanPSL-2.0 licensing — the Mulan-family-license apparatus placed inside the national industrial-policy document as a compliance-guidance instrument.
  • The state-directed open-source-project selection-and-cultivation mechanism with community-first-release pairing pairs institutionally with the August 19 briefing’s AtomGit national-level open-source-community designation documentation.
  • The research-evaluation-incorporation move extends the September 6 briefing’s star-count-as-administrative-metric move at the national-industrial-policy layer.
  • The quantitative open-source-project targets pairing pairs institutionally with the September 10 briefing’s PyTorch Foundation Western-foundation-governance-seat documentation and September 11 briefing’s Chen Kaihua Qiushi party-theoretical-flagship documentation.

One structural risk across the five moves: the same state-directed-industrial-policy apparatus is being operationalized at multiple institutional surfaces simultaneously, and the same “community” surface is being absorbed into multiple state-directed mechanisms simultaneously.

The state-directed Agent Software Application Store, Skills Resource Library, and Agent Open-Source Zone are being operationalized at the agent-software-distribution surface; the state-directed open-source-project selection-and-cultivation mechanism with community-first-release is being operationalized at the project-selection-and-cultivation surface; the independent open-source license compliance guidance is being operationalized at the licensing-and-legal surface; the research-evaluation-incorporation is being operationalized at the higher-education and research-institute surface. The same “community” surface — the operational community where open-source contribution actually happens — is being absorbed into multiple state-directed mechanisms simultaneously (the app-store, the selection-and-cultivation, the license-compliance, the research-evaluation-incorporation, the quantitative-target surfaces).

The institutional-economics question — which this cycle’s briefing documents but does not answer — is whether the Chinese institutional architecture can hold the seven institutional surfaces simultaneously without one converting the other. If the state-directed Agent Software Application Store absorbs the traditional Android-app-distribution surface (CAC Qinglang Phase 2) into a state-directed-agent-software-distribution surface, the state-directed-app-store institutional-form converts into a state-directed-distribution-monopoly object rather than a state-directed-distribution-instrument object; if the independent open-source license compliance guidance (Mulan-family) converts the Mulan Foundation from a community-level licensing foundation into a national-industrial-policy compliance-guidance apparatus, the Mulan Foundation’s community-level institutional-form converts into a state-directed-licensing-instrument object; if the research-evaluation-incorporation move converts universities and research institutes’ research-evaluation systems into open-source-contribution-capture systems rather than open-source-contribution-recognition systems, the research-evaluation-incorporation move becomes an open-source-contribution-capture apparatus rather than an open-source-contribution-incentive apparatus.

One perspective, not a verdict.

The AI + Software Special Action Implementation Plan is best read as an observation of institutional movement in progress, not as a verdict on institutional direction. The plan does not guarantee that the state-directed Agent Software Application Store will be operationalized as a state-directed-distribution-instrument rather than a state-directed-distribution-monopoly; the plan does not guarantee that the independent open-source license compliance guidance will maintain Mulan’s community-level institutional-form rather than convert into a state-directed-licensing-instrument apparatus; the plan does not guarantee that the research-evaluation-incorporation move will remain an open-source-contribution-incentive apparatus rather than convert into an open-source-contribution-capture apparatus. What this cycle’s briefing documents is that the state-directed-industrial-policy surface has crossed into the seventh surface of state-directed open-source institutional formalization — from a six-surface-simultaneous-formalization pattern (September 6 briefing’s documentation of the AI SME Startup Support Plan 31号 as the sixth surface) to a seven-surface-simultaneous-formalization pattern — and that the seventh surface is the one that pairs state-directed-app-store / independent-license-compliance / community-first-release / research-evaluation-incorporation / quantitative-project-target moves into a single national-industrial-policy document, with each of the five moves pairing institutionally with a prior briefing’s documented surface.

Editorial note on perspective: This briefing presents one institutional-economics reading of Chinese open-source developments, not a verdict. The “institution” in this story — the AI + Software Special Action Implementation Plan — is treated as an object of observation, not a target of critique. The Great Divergence 2.0 framework (FLOSS vs. State-Chartered Codebase vs. Intranet Shared Source vs. Cyber-Estate) is a lens, not a universal answer. One perspective, not a verdict.