> 页面示范：仅用于说明 Task Skill 2.0 如何管理软件开发长任务。
> 此示范不参与生成规则，不是领域模板；具体架构/模块/代码由后续 Coding LLM 基于真实项目决定。

# RelVectors Task Skill

> Task Skill 2.0 public process projection.
> This file is the persistent process memory and executable plan for one active Subject.
> I1→I2→A3→A4→O5→O6 are all present before execution; planned future nodes are not evidence that those events have happened.
> Continue the same task from Current Process / Next Process and update this Skill only from meaningful new evidence or process change.

## 0. Operating Contract

- Root process: `I → A → O` (`I1+I2 → A3+A4 → O5+O6`).
- One Subject → one Process → one continuously growing Task Skill.
- Planned / inferred / simulated content is not verified reality.
- This generator does not use software-domain templates. Specific architecture, modules, code and technology choices are the downstream Coding LLM's responsibility after reading real project context.
- Coding Dictionary atoms only normalize coding-process semantics; they do not replace software-domain knowledge or RelVectors process authority.
- Host / Tool execution is required for verified external reality evidence.
- `KEEP / CHANGE / REWIND / EXTEND` must be decided relative to the active Work, Evidence, Criterion and current process—not from an event name alone.

## 1. Identity

```json
{
  "task_id": "web-a58064d9d4a0202bc033a1bf",
  "schema": "relvectors.task-skill/2.0",
  "revision": 1,
  "cycle": 0
}
```

## 2. Subject

```json
{
  "subject_id": "subject:978b4031798c13b30393393d",
  "version": 1,
  "title": "开发咖啡店软件",
  "goal": "开发咖啡店软件\n\n用户希望这个软件做到：\n销售咖啡\n\n用户认为达到以下结果即可完成：\n用户可以在线支付付钱买咖啡",
  "origin": "USER",
  "scope": {
    "include": [
      "开发咖啡店软件",
      "销售咖啡",
      "用户可以在线支付付钱买咖啡"
    ],
    "exclude": [
      "未由用户、真实项目、工具或可验证资料支持的具体实现事实"
    ]
  },
  "status": "ACTIVE"
}
```

## 3. SELF

```json
{
  "self_id": "self:web-a58064d9d4a0202bc033a1bf",
  "definition": "用户获得可用的软件结果，并达到：用户可以在线支付付钱买咖啡",
  "success_criteria_refs": [
    "CRI-user-result@1"
  ]
}
```

## 4. Time

```json
{
  "mode": "LIVE"
}
```

## 5. I → A → O Process

### I1

```json
{
  "plan_status": "DEFINED",
  "process_status": "REALIZED",
  "meaning": "用户提出软件任务与起因：开发咖啡店软件"
}
```

### I2

```json
{
  "plan_status": "PARTIAL",
  "process_status": "REALIZED",
  "meaning": "已记录用户提供的功能/完成条件；实际仓库、运行环境和未明确产品事实按 Gap/分支管理。"
}
```

### A3

```json
{
  "plan_status": "DEFINED",
  "process_status": "REALIZED",
  "meaning": "已建立模板无关的软件开发长任务管理 A3；第二版 Coding Dictionary 仅归一化过程对象/操作，不预设具体软件架构。"
}
```

### A4

```json
{
  "plan_status": "DEFINED",
  "process_status": "NOT_STARTED",
  "meaning": "由 Coding LLM/Agent 先基于真实上下文形成具体开发计划，再分步执行；真实工具结果必须绑定回具体 Work/Criterion。"
}
```

### O5

```json
{
  "plan_status": "DEFINED",
  "process_status": "NOT_STARTED",
  "meaning": "交付物只有在当前 Criterion 获得对应验证证据后才可判定完成。"
}
```

### O6

```json
{
  "plan_status": "DEFINED",
  "process_status": "NOT_STARTED",
  "meaning": "评价 O5 是否与 SELF/用户目标一致；失败时定位受影响位置并 CHANGE/REWIND/EXTEND。"
}
```

## 6. Current Process

```json
{
  "position": "A4",
  "function": {
    "from_state": "A3 已建立软件开发长任务管理骨架，但具体技术方案尚未由执行 LLM 基于真实项目形成。",
    "to_state": "先读取真实仓库/环境，再由 Coding LLM 制定具体开发计划并持续用 Host/Tool Evidence 更新同一个 Task Skill。"
  },
  "route": "CONTINUE",
  "focus_ref": "WORK-context@1",
  "computability": {
    "decision": "NO_ASK",
    "status": "COMPUTABLE",
    "blocking_gap_refs": [],
    "question": null
  },
  "because": "当前可以先读取真实环境；具体模块、架构、语言、框架和实现细节由执行 LLM 基于真实 Evidence 决定。"
}
```

## 7. A3 Software Development Long-Task Plan

```json
{
  "task_kind": "SOFTWARE_DEVELOPMENT",
  "dictionary_ref": {
    "name": "relvectors-coding-core-atoms",
    "version": "relvectors-coding-v0.2.0-dev.2/65-atoms",
    "content_hash": "fed49da128782633e79da064d4a5c71876c9266a1209f3728207acf021e08da9"
  },
  "domain_terms": [
    {
      "term_id": "TERM-ProjectContext",
      "name": "ProjectContext",
      "object_atoms": [
        "context",
        "file",
        "path",
        "dependency",
        "config"
      ],
      "operation_atoms": [
        "read",
        "search",
        "parse",
        "resolve",
        "validate"
      ]
    },
    {
      "term_id": "TERM-ImplementationPlan",
      "name": "ImplementationPlan",
      "object_atoms": [
        "task",
        "graph",
        "dependency",
        "context"
      ],
      "operation_atoms": [
        "derive",
        "build",
        "link",
        "select",
        "validate"
      ]
    },
    {
      "term_id": "TERM-ChangeSet",
      "name": "ChangeSet",
      "object_atoms": [
        "content",
        "file",
        "state",
        "result"
      ],
      "operation_atoms": [
        "execute",
        "mutate",
        "write",
        "commit"
      ]
    },
    {
      "term_id": "TERM-VerificationEvidence",
      "name": "VerificationEvidence",
      "object_atoms": [
        "result",
        "event",
        "error",
        "condition"
      ],
      "operation_atoms": [
        "execute",
        "watch",
        "validate",
        "verify"
      ]
    },
    {
      "term_id": "TERM-Recovery",
      "name": "Recovery",
      "object_atoms": [
        "error",
        "state",
        "dependency",
        "result"
      ],
      "operation_atoms": [
        "resolve",
        "invalidate",
        "retry",
        "restore"
      ]
    },
    {
      "term_id": "TERM-Deliverable",
      "name": "Deliverable",
      "object_atoms": [
        "output",
        "content",
        "result"
      ],
      "operation_atoms": [
        "serialize",
        "emit",
        "verify"
      ]
    }
  ],
  "work_units": [
    {
      "work_id": "WORK-context",
      "version": 1,
      "title": "读取真实项目、运行环境、已有文件、依赖、工具与约束",
      "domain_term_refs": [
        "TERM-ProjectContext"
      ],
      "object_atoms": [
        "context",
        "file",
        "path",
        "dependency",
        "config"
      ],
      "operation_atoms": [
        "read",
        "search",
        "parse",
        "resolve",
        "validate"
      ],
      "expected_evidence": [],
      "status": "READY"
    },
    {
      "work_id": "WORK-plan",
      "version": 1,
      "title": "由 Coding LLM 根据用户目标与真实上下文制定当前具体开发计划",
      "domain_term_refs": [
        "TERM-ImplementationPlan"
      ],
      "object_atoms": [
        "task",
        "graph",
        "dependency",
        "context"
      ],
      "operation_atoms": [
        "derive",
        "build",
        "link",
        "select",
        "validate"
      ],
      "expected_evidence": [],
      "status": "PLANNED"
    },
    {
      "work_id": "WORK-implement",
      "version": 1,
      "title": "按当前计划分步开发；每个有意义的实现变化都保留 Evidence 并更新同一 Task Skill",
      "domain_term_refs": [
        "TERM-ChangeSet"
      ],
      "object_atoms": [
        "content",
        "file",
        "state",
        "result"
      ],
      "operation_atoms": [
        "execute",
        "mutate",
        "write",
        "commit"
      ],
      "expected_evidence": [],
      "status": "PLANNED"
    },
    {
      "work_id": "WORK-verify",
      "version": 1,
      "title": "使用测试、构建、运行或其他真实工具证据验证当前结果与活动 Criterion",
      "domain_term_refs": [
        "TERM-VerificationEvidence"
      ],
      "object_atoms": [
        "result",
        "event",
        "error",
        "condition"
      ],
      "operation_atoms": [
        "execute",
        "watch",
        "validate",
        "verify"
      ],
      "expected_evidence": [],
      "status": "PLANNED"
    },
    {
      "work_id": "WORK-recover",
      "version": 1,
      "title": "当真实 Evidence 否定当前路线时定位受影响范围，并支持 CHANGE / REWIND / EXTEND 后继续",
      "domain_term_refs": [
        "TERM-Recovery"
      ],
      "object_atoms": [
        "error",
        "state",
        "dependency",
        "result"
      ],
      "operation_atoms": [
        "resolve",
        "invalidate",
        "retry",
        "restore"
      ],
      "expected_evidence": [],
      "status": "PLANNED"
    },
    {
      "work_id": "WORK-deliver",
      "version": 1,
      "title": "整理可交付结果、验证证据、剩余 Gap 与继续/结束状态",
      "domain_term_refs": [
        "TERM-Deliverable"
      ],
      "object_atoms": [
        "output",
        "content",
        "result"
      ],
      "operation_atoms": [
        "serialize",
        "emit",
        "verify"
      ],
      "expected_evidence": [],
      "status": "PLANNED"
    }
  ],
  "relations": [
    {
      "relation_id": "REL-1",
      "type": "DEPENDS_ON",
      "source_ref": "WORK-plan@1",
      "target_ref": "WORK-context@1"
    },
    {
      "relation_id": "REL-2",
      "type": "DEPENDS_ON",
      "source_ref": "WORK-implement@1",
      "target_ref": "WORK-plan@1"
    },
    {
      "relation_id": "REL-3",
      "type": "DEPENDS_ON",
      "source_ref": "WORK-verify@1",
      "target_ref": "WORK-implement@1"
    },
    {
      "relation_id": "REL-4",
      "type": "DEPENDS_ON",
      "source_ref": "WORK-deliver@1",
      "target_ref": "WORK-verify@1"
    },
    {
      "relation_id": "REL-5",
      "type": "VERIFIES",
      "source_ref": "WORK-verify@1",
      "target_ref": "CRI-user-result@1"
    },
    {
      "relation_id": "REL-6",
      "type": "VERIFIES",
      "source_ref": "WORK-verify@1",
      "target_ref": "CRI-specific-proof@1"
    },
    {
      "relation_id": "REL-7",
      "type": "VERIFIES",
      "source_ref": "WORK-deliver@1",
      "target_ref": "CRI-no-blocking-gap@1"
    },
    {
      "relation_id": "REL-8",
      "type": "CONSUMES",
      "source_ref": "WORK-recover@1",
      "target_ref": "TERM-VerificationEvidence"
    }
  ],
  "criteria": [
    {
      "criterion_id": "CRI-user-result",
      "version": 1,
      "role": "COMPLETION",
      "text": "达到用户声明的完成结果：用户可以在线支付付钱买咖啡",
      "verifier": {
        "id": "user-result",
        "type": "HUMAN",
        "expect": {
          "user_result_satisfied": true
        }
      }
    },
    {
      "criterion_id": "CRI-specific-proof",
      "version": 1,
      "role": "INVARIANT",
      "text": "执行 LLM 必须把用户完成条件细化为与实际实现对应的可验证检查，并用当前周期 Evidence 证明后才能进入 O5 完成。",
      "verifier": {
        "id": "relvectors-evidence-contract",
        "type": "EXTERNAL",
        "expect": {
          "criterion_specific_proof": true
        }
      }
    },
    {
      "criterion_id": "CRI-no-blocking-gap",
      "version": 1,
      "role": "INVARIANT",
      "text": "最终交付时不存在会阻止用户目标成立的 OPEN blocking Gap。",
      "verifier": {
        "id": "relvectors-gap-contract",
        "type": "EXTERNAL",
        "expect": {
          "open_blocking_gaps": 0
        }
      }
    }
  ],
  "branches": []
}
```

## 8. A4 Execution

```json
{
  "status": "NOT_STARTED",
  "current_action_ref": null,
  "last_committed_evidence_ref": null,
  "recovery": {
    "state": "CLEAN",
    "host_journal_ref": null,
    "reason": null
  }
}
```

## 9. O5 Completion

```json
{
  "delivery_refs": [],
  "criteria_state": [
    {
      "criterion_ref": "CRI-user-result@1",
      "status": "UNPROVEN",
      "proof_refs": []
    },
    {
      "criterion_ref": "CRI-specific-proof@1",
      "status": "UNPROVEN",
      "proof_refs": []
    },
    {
      "criterion_ref": "CRI-no-blocking-gap@1",
      "status": "UNPROVEN",
      "proof_refs": []
    }
  ]
}
```

## 10. O6 Evaluation / Re-entry

```json
{
  "status": "NOT_STARTED",
  "alignment": "UNRESOLVED",
  "effect_on_self": "",
  "global_effect": "",
  "failed_refs": [],
  "reentry_position": null
}
```

## 11. Process Gaps

```json
[
  {
    "gap_id": "GAP-context",
    "position": "I2",
    "missing": "实际仓库、运行环境、已有代码/文件、依赖和可用工具状态",
    "why_needed": "具体技术方案必须由执行 LLM 基于真实环境决定，生成器不能预设。",
    "blocking": false,
    "user_resolvable": false,
    "next": "RETRIEVE",
    "status": "OPEN",
    "evidence_refs": []
  }
]
```

## 12. Decisions

```json
[
  {
    "decision_id": "DEC-no-domain-template",
    "kind": "PLAN",
    "position": "A3",
    "statement": "生成器不预设软件领域架构、模块、语言、框架或具体实现；这些由执行 Coding LLM 在读取真实上下文后决定并写回同一 Task Skill。",
    "basis_refs": [
      "EV-user-1"
    ],
    "status": "ACTIVE"
  }
]
```

## 13. Changes

```json
[]
```

## 14. Evidence

```json
[
  {
    "evidence_id": "EV-user-1",
    "origin": "USER_REPORTED",
    "actuality": "REPORTED",
    "position": "I1",
    "subject_version": 1,
    "work_ref": null,
    "criterion_refs": [],
    "reality_verified": false,
    "summary": "开发咖啡店软件\n\n用户希望这个软件做到：\n销售咖啡\n\n用户认为达到以下结果即可完成：\n用户可以在线支付付钱买咖啡",
    "host_evidence_ref": null
  }
]
```

## 15. Meaningful History

```json
[
  {
    "seq": 1,
    "cycle": 0,
    "position": "I1",
    "kind": "INPUT",
    "ref": "EV-user-1"
  }
]
```

## 16. Frozen Past

```json
[]
```

## 17. Derived Frontiers

```json
{
  "reality_frontier": {
    "position": "I1",
    "because": "当前生成阶段只有用户报告的需求属于已发生输入；尚无真实 Host/Tool 实施结果。"
  },
  "knowledge_frontier": {
    "position": "O6",
    "actuality": "PLANNED",
    "because": "完整方案可规划到 O6，但知识/计划不等于现实已经发生。"
  },
  "planning_coverage": {
    "covered": [
      "I1",
      "A3",
      "A4",
      "O5",
      "O6"
    ],
    "partial": [
      "I2"
    ],
    "missing": []
  },
  "execution_frontier": {
    "position": "A4",
    "focus_ref": "WORK-context@1",
    "computability": {
      "decision": "NO_ASK",
      "status": "COMPUTABLE",
      "blocking_gap_refs": [],
      "question": null
    }
  }
}
```

## 18. Next Process

```json
{
  "position": "A4",
  "route": "CONTINUE",
  "focus_ref": "WORK-context@1",
  "because": "当前可以先读取真实环境；具体模块、架构、语言、框架和实现细节由执行 LLM 基于真实 Evidence 决定。"
}
```

## 19. Agent Resume

```json
{
  "task_id": "web-a58064d9d4a0202bc033a1bf",
  "revision": 1,
  "cycle": 0,
  "current_position": "A4",
  "focus_ref": "WORK-context@1",
  "recovery_state": "CLEAN"
}
```

Treat this Task Skill as the authoritative process memory for the active Subject. Preserve valid Evidence, Decisions, Changes, Frozen Past and unresolved Gaps. Continue from `Current Process` / `Next Process`. Do not turn planned A4/O5/O6 into completed history. When real repository/environment information becomes available, let the Coding LLM create or revise the concrete implementation plan inside the same Task Skill. Preserve Subject, Evidence, Decisions, Changes and history; do not restart from chat or treat generator examples as templates.

---

Core binding: `relvectors.a3-skill/1.1` · `1.1.0-dev.3` · `skill-e7ebf60dbaba8d3af44b6e42`

Coding semantics: second-version RelVectors Coding Dictionary. Website examples are demonstrations only and are not generation templates.