autosnippet 3.2.18 → 3.2.21

package/skills/autosnippet-intent/SKILL.md

@@ -1,55 +0,0 @@
----
-name: autosnippet-intent
-description: A light-weight router skill. Decide which AutoSnippet capability should be used (recipes/search/create/candidates/guard/structure/concepts). Use this to pick the right skill, then delegate to that skill for detailed steps.
----
-
-# AutoSnippet — Intent Router
-
-Use this skill when the user's intent is unclear or overlaps multiple capabilities. Your job is to select **one** primary capability, then hand off to the specific skill.
-
-## Decision map (pick one primary path)
-
-| User intent / situation | Primary skill | Notes |
-|---|---|---|
-| "冷启动 / 初始化知识库 / 首次接入 / bootstrap" | **autosnippet-coldstart** | Full 9-dimension cold-start. |
-| "全项目分析 / 提取架构模式 / 生成知识库" | **autosnippet-coldstart** | Same — full project analysis. |
-| "有没有现成写法 / 查一下规范 / 给我标准代码" | **autosnippet-recipes** | Use Recipe context first. |
-| "把这段加入知识库 / 提交到知识库" | **autosnippet-create** | Submit via Dashboard or draft. |
-| "生成候选 / 批量扫描 / 扫 Target" | **autosnippet-candidates** | Generate rich candidates. |
-| "帮我审计 / Lint / 规范检查" | **autosnippet-guard** | Use Recipe as audit standard. |
-| "依赖关系 / 目标结构 / targets / spmmap" | **autosnippet-structure** | Targets + dep graph. |
-| "知识库/Recipe/Trigger 是什么" | **autosnippet-concepts** | Explain concepts. |
-| "补全候选字段 / enrich / 深度分析" | **autosnippet-analysis** | Semantic enrichment. |
-
-## Rules
-
-1. **Pick one primary path**. Avoid sending multiple flows unless the user explicitly asks.
-2. **Do not force a flow**. Recommend softly (“可以…”), and only when relevant.
-3. After choosing, **use the chosen skill** for the detailed steps.
-
-## Deprecated skills (removed — do not reference)
-
-- `autosnippet-when` → replaced by this skill (autosnippet-intent)
-- `autosnippet-search` → merged into autosnippet-recipes
-- `autosnippet-batch-scan` / `autosnippet-recipe-candidates` → merged into autosnippet-candidates
-- `autosnippet-dep-graph` → merged into autosnippet-structure
-
-## MCP tools (reference only — 20 个整合工具)
-
-- System: `autosnippet_health`, `autosnippet_capabilities`
-- Search: `autosnippet_search` (mode: auto / context / keyword / semantic)
-- Knowledge: `autosnippet_knowledge` (operation: list / get / insights / confirm_usage)
-- Graph: `autosnippet_graph` (operation: query / impact / path / stats)
-- Structure: `autosnippet_structure` (operation: targets / files / metadata)
-- Submit: `autosnippet_submit_knowledge`, `autosnippet_submit_knowledge_batch`
-- Document: `autosnippet_save_document`
-- Guard: `autosnippet_guard` (code 单文件 / files[] 批量)
-- Bootstrap: `autosnippet_bootstrap` (no params — Mission Briefing) + `autosnippet_dimension_complete`
-- Wiki: `autosnippet_wiki_plan` (topic planning) + `autosnippet_wiki_finalize` (meta.json + dedup)
-- Task: `autosnippet_task` (operation: prime (session entry) / create / ready / claim / close / fail / defer / progress / stats / decompose / record_decision / revise_decision / unpin_decision / list_decisions)
-- Skills: `autosnippet_skill` (operation: list / load / create / update / delete / suggest)
-- Admin: `autosnippet_enrich_candidates`, `autosnippet_knowledge_lifecycle`, `autosnippet_validate_candidate`, `autosnippet_check_duplicate`
-
-This skill is a router only; it does not perform actions itself.
-
-```

package/skills/autosnippet-lifecycle/SKILL.md

@@ -1,100 +0,0 @@
----
-name: autosnippet-lifecycle
-description: Understand the Recipe lifecycle (draft -> active -> deprecated) and the Agent's role boundaries. Agent can submit candidates, validate, confirm usage, but CANNOT create/modify/publish/delete Recipes directly.
----
-
-# AutoSnippet Recipe Lifecycle
-
-This skill documents the Recipe lifecycle and clarifies what Agent can and cannot do.
-
----
-
-## Lifecycle Stages
-
-```
-                  Human approves
-  Candidate -----+-----> Draft Recipe -----> Active Recipe -----> Deprecated
-  (Agent submits) |      (human creates)     (human publishes)    (human deprecates)
-                  |
-                  +-----> Rejected
-                         (human rejects)
-```
-
-### Stage Details
-
-| Stage | Who | How |
-|-------|-----|-----|
-| **Candidate** | Agent submits via MCP | `submit_candidate` / `submit_candidates` / `submit_draft_recipes` |
-| **Review** | Human via Dashboard | Dashboard Candidates page -> approve/reject |
-| **Draft Recipe** | Human via Dashboard | Approved candidate becomes draft Recipe |
-| **Active Recipe** | Human via Dashboard/API | `PATCH /api/v1/recipes/:id/publish` |
-| **Updated** | Human via Dashboard/API | `PATCH /api/v1/recipes/:id` |
-| **Deprecated** | Human via Dashboard/API | `PATCH /api/v1/recipes/:id/deprecate` |
-| **Deleted** | Human via Dashboard/API | `DELETE /api/v1/recipes/:id` |
-
----
-
-## Agent Role: What You CAN Do
-
-### 1. Submit Candidates
-- `autosnippet_submit_knowledge` - single structured candidate (built-in auto-validate + dedup)
-- `autosnippet_submit_knowledge_batch` - batch structured candidates
-
-### 2. Validate and Enhance Candidates
-- Submit via `autosnippet_submit_knowledge` which auto-validates and checks duplicates
-- Add structured content: rationale, steps, codeChanges, knowledgeType, complexity, tags, constraints, headers
-
-### 3. Search and Query Recipes
-- `autosnippet_knowledge(operation=list)` - list with filters (kind/language/category/knowledgeType/status/complexity)
-- `autosnippet_knowledge(operation=get, id)` - get full Recipe details
-- `autosnippet_search` - unified search (mode=auto/context/keyword/semantic)
-- `autosnippet_graph(operation=query/impact)` - knowledge graph
-
-### 4. Record Usage Telemetry
-- `autosnippet_knowledge(operation=confirm_usage)` - record when user adopts/applies a Recipe
-  - `usageType: "adoption"` - user adopted the Recipe
-  - `usageType: "application"` - user applied the Recipe in code
-
----
-
-## Agent Role: What You CANNOT Do
-
-| Forbidden Action | Why | Human Alternative |
-|-----------------|-----|-------------------|
-| Create Recipe directly | Agent produces candidates, not Recipes | Dashboard: approve candidate |
-| Modify Recipe content | Recipe content is human-controlled | Dashboard: edit Recipe |
-| Publish Recipe (draft -> active) | Publishing is a human review decision | Dashboard: publish button |
-| Deprecate Recipe | Deprecation is a human lifecycle decision | Dashboard: deprecate button |
-| Delete Recipe | Deletion is irreversible, human-only | Dashboard: delete button |
-| Update quality scores | Quality assessment is human-controlled | Dashboard: quality panel |
-| Write to `AutoSnippet/recipes/` directly | Bypass candidate review process | Submit as candidate first |
-
----
-
-## Typical Agent Workflows
-
-### Workflow 1: User asks to add a code pattern
-1. Analyze the code pattern
-2. Generate structured candidate with rationale, steps, etc.
-3. `autosnippet_submit_knowledge` - submit to candidate pool (auto-validates + dedup check)
-4. Tell user: "Candidate submitted. Review in Dashboard Candidates page."
-
-### Workflow 2: User asks about an existing Recipe
-1. `autosnippet_search` or `autosnippet_knowledge(operation=list)` to find it
-2. `autosnippet_knowledge(operation=get, id)` to get full details
-3. Present the Recipe content to user
-4. If user adopts it: `autosnippet_knowledge(operation=confirm_usage)` with usageType=adoption
-
-### Workflow 3: User asks to update a Recipe
-1. Explain: "I cannot modify Recipes directly. I can submit an improved version as a new candidate."
-2. `autosnippet_knowledge(operation=get, id)` to read current content
-3. Generate improved candidate based on user's feedback
-4. `autosnippet_submit_knowledge` - submit improved version
-5. Tell user: "Improved version submitted as candidate. Update the Recipe in Dashboard."
-
----
-
-## Related Skills
-
-- **autosnippet-recipes**: Project context, Recipe lookup, MCP tools reference
-- **autosnippet-candidates**: Candidate generation workflow (single file / batch scan)

package/templates/copilot-instructions.md

@@ -1,66 +0,0 @@
-<!-- autosnippet:begin -->
-<!-- MANDATORY: On EVERY human message, call autosnippet_task({ operation: "prime" }) FIRST. -->
-
-# {{projectName}} — AutoSnippet 项目指引
-
-## CRITICAL: You Operate autosnippet_task — The User Doesn't
-
-You are the task operator. Users speak naturally; you translate to task operations.
-DO NOT tell users to call autosnippet_task. That's YOUR job.
-
-- WRONG: "你可以调用 autosnippet_task({ operation: 'create' }) 来创建任务"
-- RIGHT: *(你自己调用 create，然后告诉用户 "已创建任务 asd-42: 修复登录 bug")*
-
-## Task Tracking Rules (MANDATORY)
-
-1. **MUST prime on EVERY message** — `autosnippet_task({ operation: "prime" })` BEFORE doing anything else. No exceptions.
-2. **MUST create a task for non-trivial work** — If work involves ≥2 files OR ≥10 lines, create BEFORE starting.
-3. **MUST claim before coding** — `autosnippet_task({ operation: "claim", id })` then code.
-4. **MUST close when done** — `autosnippet_task({ operation: "close", id, reason })` with meaningful reason.
-5. **MUST handle unfinished tasks first** — If prime returns in-progress tasks, ask user: Continue, Defer, or Abandon? Do NOT auto-resume.
-6. **NEVER skip prime** — Even for follow-up messages.
-7. **NEVER start new work with open in-progress tasks** — Handle existing first.
-8. **NEVER leave tasks in in_progress when session ends** — Close or defer everything.
-9. **NEVER tell the user to run task commands** — You are the operator.
-
-When in doubt → create a task. When idle → `autosnippet_task({ operation: "ready" })`.
-
-### When NOT to Create Tasks
-
-Quick questions, single-file trivial fixes (<10 lines), code explanation, running existing tests.
-Rule: <2min AND ≤1 file AND <10 lines → no task. Everything else → create.
-
-## User Says → You Run
-
-| User Says | Your Action |
-|---|---|
-| "帮我修 bug" / "fix this" | `create` → `claim` → code → `close` |
-| "做功能" / "implement" | `create` → `claim` → code → `close` |
-| "继续" / "continue" | resume in-progress task → code → `close` |
-| "先不做了" / "pause" | `defer(id, reason)` |
-| "不做了" / "abandon" | `fail(id, reason)` |
-| "太大了" / "break down" | `decompose(id, subtasks)` |
-| "有什么要做的" / "what's next" | `ready()` → present |
-| "就这么定了" / "agreed" | `record_decision(...)` |
-| Quick question | No task. Just answer. |
-
-## Session Closing Protocol
-
-Before ending work, you MUST complete this checklist:
-
-- [ ] Close every claimed task: `autosnippet_task({ operation: "close", id, reason: "what was accomplished" })`
-- [ ] Defer any incomplete task: `autosnippet_task({ operation: "defer", id, reason: "why and what remains" })`
-- [ ] Verify: zero tasks in in_progress state
-- [ ] If prime showed ready tasks, mention them to the user for next session
-
-**Work is not done until all tasks are closed or deferred.**
-
-## Knowledge Rules
-
-- **禁止直接修改** `AutoSnippet/recipes/`、`.autosnippet/` 等知识库目录
-- **优先使用 Recipe**；搜索用 `autosnippet_search`
-- MCP 调用失败时**不要重试**，回退到已有上下文
-
-<!-- autosnippet:end -->
-<!-- REMINDER: If you have not yet called autosnippet_task({ operation: "prime" }), do it NOW. -->
-<!-- REMINDER: If user agreed/disagreed with a plan → autosnippet_task({ operation: "record_decision" | "revise_decision" | "unpin_decision" }) immediately. -->

package/templates/cursor-rules/autosnippet-conventions.mdc

@@ -1,172 +0,0 @@
----
-description: AutoSnippet conventions — Skills for semantics, MCP for capabilities; do not edit Knowledge directly
-alwaysApply: true
----
-
-# AutoSnippet Conventions
-
-These rules are aligned with `.github/copilot-instructions.md` and jointly provide the AI with basic project cognition and required guardrails.
-
-These rules are written when installing AutoSnippet Skills (`asd install:cursor-skill`) and should be followed in Cursor sessions.
-
-## 强制规则（必须遵守）
-
-1. **每次收到用户消息，第一步调 `autosnippet_task({ operation: "prime" })`** — 加载最新决策和任务上下文，跳过会导致回答与团队约定矛盾。
-2. **当用户明确同意/否决某方案时，立即调用 `autosnippet_task({ operation: "record_decision" })` 持久化**（record_decision/revise_decision/unpin_decision）— 先写入团队记忆，再继续后续动作。
-3. **禁止直接修改** 知识库目录内容（`AutoSnippet/recipes/`、`.autosnippet/` 等）。
-4. **优先使用 Recipe** 作为项目标准；源代码仅作补充。
-5. MCP 检索优先：使用 `autosnippet_search`（auto 融合模式）。
-6. MCP 调用失败时，**不要在同一轮重复重试**，回退到已读文档或静态上下文。
-7. Skills 负责语义与流程，MCP 负责能力与调用；不要在 Skill 内硬编码 URL/HTTP。
-
-## CRITICAL: You Operate autosnippet_task — The User Doesn't
-
-You are the task operator. Users speak naturally; you translate to task operations.
-DO NOT tell users to call autosnippet_task. That's YOUR job.
-
-- WRONG: "你可以调用 autosnippet_task({ operation: 'create' }) 来创建任务"
-- RIGHT: *(你自己调用 create，然后告诉用户 "已创建任务 asd-42: 修复登录 bug")*
-
-## Task Tracking Rules (MANDATORY)
-
-1. **MUST prime on EVERY message** — No exceptions. Even follow-up messages.
-2. **MUST create task for non-trivial work** — ≥2 files OR ≥10 lines → create BEFORE starting.
-3. **MUST claim before coding** — `claim(id)` then code.
-4. **MUST close when done** — `close(id, reason)` with meaningful reason.
-5. **MUST handle unfinished tasks first** — prime shows in-progress → ask user: Continue, Defer, Abandon? Do NOT auto-resume.
-6. **NEVER skip prime** — Even for simple questions.
-7. **NEVER start new work with open in-progress tasks** — Handle existing first.
-8. **NEVER leave tasks in in_progress when session ends** — Close or defer everything.
-9. **NEVER tell the user to run task commands** — You are the operator.
-
-When in doubt → create a task. When idle → `autosnippet_task({ operation: "ready" })`.
-
-**When NOT to create**: Quick questions, single-file trivial fixes (<10 lines), code explanation, running tests. Rule: <2min AND ≤1 file AND <10 lines → no task.
-
-## User Says → You Run
-
-| User Says | Your Action |
-|---|---|
-| "帮我修 bug" / "fix this" | `create` → `claim` → code → `close` |
-| "做功能" / "implement" | `create` → `claim` → code → `close` |
-| "继续" / "continue" | resume in-progress → code → `close` |
-| "先不做了" / "pause" | `defer(id, reason)` |
-| "不做了" / "abandon" | `fail(id, reason)` |
-| "太大了" / "break down" | `decompose(id, subtasks)` |
-| "有什么要做的" / "next" | `ready()` → present list |
-| "就这么定了" / "agreed" | `record_decision(...)` |
-| Quick question | No task. Just answer. |
-
-## Session Closing Protocol
-
-Before ending work, you MUST complete this checklist:
-
-- [ ] Close every claimed task: `close(id, reason)` with what was accomplished
-- [ ] Defer any incomplete task: `defer(id, reason)` with why and what remains
-- [ ] Verify: zero tasks in in_progress state
-- [ ] If prime showed ready tasks, mention them for next session
-
-**Work is not done until all tasks are closed or deferred.**
-
-## Semantics vs capabilities
-
-- **Skills provide semantics**: When to use, workflows, and comparisons are described in Skills. Use natural language prompts in Cursor (e.g. "scan targets", "find recipe examples", "batch extract candidates") to trigger the appropriate Skill; AI intelligently interprets your request and activates the corresponding Skill logic.
-- **MCP provides capabilities**: Connection and tools are provided by MCP. Do not hardcode URLs/HTTP in Skills.
-
-## Project basics (for AI)
-
-- **Project root**: directory containing `boxspec.json`.
-- **Knowledge base**: `AutoSnippet/` (read-only for AI).
-- **Recipes**: `AutoSnippet/recipes/*.md` (Markdown + Frontmatter + Snippet + Usage Guide).
-  - **Frontmatter required fields (7)**: `title` (English, ≤50 chars, verb-based), `trigger` (MUST start with `@`), `category` (MUST be one of: `View`, `Service`, `Tool`, `Model`, `Network`, `Storage`, `UI`, `Utility`), `language`, `summary_cn` (≤100 chars), `summary_en` (≤100 words), `headers` (complete import statements, e.g. `["import Foundation"]`, `["#import <UIKit/UIKit.h>"]`, `["import \"fmt\""]`, `["import express from 'express'"]`).
-  - **V3 required Cursor Delivery fields**: `kind` (rule/pattern/fact), `doClause` (English imperative ≤60 tokens), `dontClause` (what NOT to do), `whenClause` (when this applies), `coreCode` (3-8 line code skeleton), `usageGuide` (### section format guide), `knowledgeType` (code-pattern/architecture/best-practice/code-standard/code-relation/data-flow/module-dependency/boundary-constraint/code-style/solution/anti-pattern/event-and-data-flow), `reasoning` ({ whyStandard, sources, confidence }).
-  - **Placeholders**: Use IDE-appropriate placeholders in snippets — Xcode format `<#URL#>` (auto-converted to VSCode `${N:...}` on install). Explain them in Usage Guide.
-  - **Usage Guide depth**: Go beyond "When/Key Points"; include dependencies, steps/config, error handling, performance, security, pitfalls, and related Recipes when applicable. MUST use `###` headings and list format.
-- **Constitution**: `AutoSnippet/constitution.yaml` (权限宪法: roles + capabilities + governance rules).
-- **Runtime DB**: `.autosnippet/autosnippet.db` (SQLite index cache for recipes/candidates/snippets).
-- **Vector index**: `.autosnippet/context/` (`asd embed`).
-- **Recipe stats**: `.autosnippet/recipe-stats.json` (usage + authority).
-
-## Knowledge categories (kind)
-
-Recipes are classified into three kinds:
-- **rule** — Guard rules (boundary-constraint), used for code quality checks
-- **pattern** — Reusable code patterns (code-pattern, architecture, best-practice, etc.)
-- **fact** — Structural knowledge (code-relation, inheritance, call-chain, data-flow, etc.)
-
-## Knowledge base and submission
-
-- **Do not edit** `AutoSnippet/recipes/` or `.autosnippet/` directly; all Recipe/Snippet changes go through Dashboard or MCP tools and are saved after human review.
-- **Search**: Use MCP tools `autosnippet_search` (recommended, auto mode BM25 + semantic fusion) or `autosnippet_context_search` (context-aware with session history). Also available: `autosnippet_keyword_search` (exact match) and `autosnippet_semantic_search` (vector).
-- **Retry policy**: If an MCP call fails, do not retry within the same agent turn; fall back to static docs or already-read context.
-- **Decision persistence**: When the user explicitly agrees/disagrees with a plan, call `autosnippet_task({ operation: "record_decision" | "revise_decision" | "unpin_decision" })` **immediately** before continuing. Persist first, execute second.
-- **Adoption tracking**: Tool `autosnippet_confirm_usage` records that a Recipe was adopted or applied (affects usage stats and authority ranking). Show it when the user explicitly adopts a recipe, not immediately after presenting one.
-- **Batch scan**: Ask Cursor in natural language (e.g. "scan all targets in this project", "batch extract candidates") to trigger the batch-scan workflow. The workflow calls `autosnippet_get_targets` → `autosnippet_get_target_files` → extract per file → `autosnippet_submit_candidates` automatically.
-- **Cold start**: Use `autosnippet_bootstrap_knowledge` for full project knowledge initialization covering 9 dimensions. Phase 5.5 auto-generates 4 Project Skills (code-standard, architecture, project-profile, agent-guidelines) to `AutoSnippet/skills/`.
-
-## Project Skills
-
-- **Skills 目录**: `AutoSnippet/skills/<name>/SKILL.md` — 跟随项目 Git 管理
-- **发现**: `autosnippet_list_skills` — 列出所有可用 Skills（内置 + 项目级）
-- **加载**: `autosnippet_load_skill(skillName)` — 获取 Skill 完整文档
-- **创建**: `autosnippet_create_skill(name, description, content)` — 创建项目级 Skill
-- **推荐**: `autosnippet_suggest_skills` — 基于使用模式推荐创建 Skill
-- **优先级**: 项目级 Skill 同名覆盖内置 Skill；Skill > Recipe（宏观知识优先查 Skill）
-- **Bootstrap 自动生成**: 冷启动会自动生成 `project-code-standard`、`project-architecture`、`project-profile`、`project-agent-guidelines` 四个 Project Skills
-
-## MCP tools (36 total)
-
-### Search (prefer search / context_search)
-- `autosnippet_search` — Unified search (auto: BM25 + semantic fusion)
-- `autosnippet_context_search` — Context-aware retrieval (intent recognition + 4-layer funnel + session continuity)
-- `autosnippet_keyword_search` — SQL LIKE exact keyword
-- `autosnippet_semantic_search` — Vector semantic search
-
-### Knowledge lists & details
-- `autosnippet_list_recipes` / `autosnippet_get_recipe` / `autosnippet_recipe_insights`
-- `autosnippet_list_rules` / `autosnippet_list_patterns` / `autosnippet_list_facts`
-
-### Knowledge graph
-- `autosnippet_graph_query` / `autosnippet_graph_impact` / `autosnippet_graph_path` / `autosnippet_graph_stats`
-
-### Candidate submission & validation
-- `autosnippet_validate_candidate` — Pre-validate (5 layers)
-- `autosnippet_check_duplicate` — Similarity detection
-- `autosnippet_submit_candidate` — Single submission (reasoning required)
-- `autosnippet_submit_candidates` — Batch submission (dedup + rate limit)
-- `autosnippet_submit_draft_recipes` — Parse draft Markdown files
-- `autosnippet_enrich_candidates` — AI-fill missing semantic fields
-
-### Project scanning & bootstrap
-- `autosnippet_get_targets` / `autosnippet_get_target_files` / `autosnippet_get_target_metadata`
-- `autosnippet_scan_project` — Lightweight probe (file list + Guard audit)
-- `autosnippet_bootstrap_knowledge` — Cold-start knowledge initialization (9 dimensions + auto Project Skills)
-- `autosnippet_bootstrap_refine` — AI refinement for bootstrap candidates (summary/insight/relations)
-
-### Guard & governance
-- `autosnippet_guard_check` / `autosnippet_guard_audit_files` / `autosnippet_compliance_report`
-
-### Skills
-- `autosnippet_list_skills` — List all available Skills (builtin + project)
-- `autosnippet_load_skill` — Load a Skill's full document
-- `autosnippet_create_skill` — Create a project-level Skill in `AutoSnippet/skills/`
-- `autosnippet_suggest_skills` — Recommend Skills based on usage patterns
-
-### Other
-- `autosnippet_health` / `autosnippet_capabilities` / `autosnippet_confirm_usage`
-
-## Recipe over code
-
-When both a Recipe and the codebase have relevant implementations, prefer the Recipe. Recipes are the project's agreed standard; source code may be legacy or non-standard.
-
-## Guard 诊断响应
-
-当编辑器出现 "AutoSnippet Guard" 诊断时：
-1. 读取 ruleId → `autosnippet_search(query: ruleId)`
-2. 按 Recipe 的 doClause + coreCode 修改代码
-3. 保存并确认诊断消失
-
-## 上下文压力应对
-
-- `_contextHint` 包含 `CONTEXT_PRESSURE:WARNING` → 总结已完成工作，继续。
-- `_contextHint` 包含 `CONTEXT_PRESSURE:CRITICAL` → 立即 prime 恢复上下文。

package/templates/cursor-rules/autosnippet-workflow.mdc

@@ -1,76 +0,0 @@
----
-description: AutoSnippet Task Tracking — 任务管理 + 行为规则
-alwaysApply: true
----
-
-# AutoSnippet Task Workflow
-
-## CRITICAL: You Operate autosnippet_task — The User Doesn't
-
-You are the task operator. Users speak naturally; you translate to task operations.
-DO NOT tell users to call autosnippet_task. That's YOUR job.
-
-## Task Tracking Rules (MANDATORY)
-
-1. **MUST prime on EVERY message** — `autosnippet_task({ operation: "prime" })` BEFORE anything else. No exceptions.
-2. **MUST create task for non-trivial work** — ≥2 files OR ≥10 lines → create BEFORE starting.
-3. **MUST claim before coding** — `claim(id)` then code.
-4. **MUST close when done** — `close(id, reason)` with meaningful reason.
-5. **MUST handle unfinished tasks first** — prime shows in-progress → ask user: Continue, Defer, Abandon?
-6. **NEVER skip prime** — Even for follow-up messages.
-7. **NEVER start new work with open in-progress tasks** — Handle existing first.
-8. **NEVER leave tasks in in_progress when session ends** — Close or defer everything.
-9. **NEVER tell the user to run task commands** — You are the operator.
-
-When in doubt → create a task. When idle → `autosnippet_task({ operation: "ready" })`.
-**When NOT to create**: <2min AND ≤1 file AND <10 lines → no task.
-
-## User Says → You Run
-
-| User Says | Your Action |
-|---|---|
-| "帮我修 bug" / "fix this" | `create` → `claim` → code → `close` |
-| "做功能" / "implement" | `create` → `claim` → code → `close` |
-| "继续" / "continue" | resume in-progress → code → `close` |
-| "先不做了" / "pause" | `defer(id, reason)` |
-| "不做了" / "abandon" | `fail(id, reason)` |
-| "太大了" / "break down" | `decompose(id, subtasks)` |
-| "有什么要做的" / "next" | `ready()` → present list |
-| "就这么定了" / "agreed" | `record_decision(...)` |
-| Quick question | No task. Just answer. |
-
-## Decision Persistence
-
-**用户同意/否决方案时** → `autosnippet_task({ operation: "record_decision" | "revise_decision" | "unpin_decision" })` 先持久化决策再继续
-
-## Session Closing Protocol
-
-Before ending work, you MUST complete this checklist:
-
-- [ ] Close every claimed task: `close(id, reason)` with what was accomplished
-- [ ] Defer any incomplete task: `defer(id, reason)` with why and what remains
-- [ ] Verify: zero tasks in in_progress state
-- [ ] If prime showed ready tasks, mention them for next session
-
-**Work is not done until all tasks are closed or deferred.**
-
-## 上下文压力
-
-- `_contextHint` 包含 `CONTEXT_PRESSURE:WARNING` → 总结已完成工作，继续
-- `_contextHint` 包含 `CONTEXT_PRESSURE:CRITICAL` → 立即 prime 恢复上下文
-
-## 任务操作速查
-
-| 操作 | 用途 |
-|------|------|
-| `autosnippet_task({ operation: "prime" })` | **每次收到消息第一步调用** — 加载决策 + 就绪任务 + 规则提醒 |
-| `autosnippet_task({ operation: "record_decision" / "revise_decision" / "unpin_decision" })` | **用户同意/否决方案时立即调用** — 先持久化再继续 |
-| `create(title)` | 创建新任务 |
-| `claim(id)` | 开始处理任务 |
-| `close(id, reason)` | 完成任务 → 返回 newlyReady |
-| `fail(id, reason)` | 标记失败 |
-| `defer(id, reason)` | 延后处理 |
-| `progress(id, desc)` | 更新进度 |
-| `ready` | 获取就绪任务 |
-| `decompose(id, subtasks)` | 拆分大任务 |
-| `stats` | 项目统计 |