Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add rules/mr-chen-05/rules-2.1-optimized/context-recorder-coregit clone --depth 1 https://github.com/Mr-chen-05/rules-2.1-optimizedWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/rules/mr-chen-05/rules-2.1-optimized/context-recorder-core)<a href="https://agentmods.dev/rules/mr-chen-05/rules-2.1-optimized/context-recorder-core"><img src="https://agentmods.dev/badge/rules/mr-chen-05/rules-2.1-optimized/context-recorder-core.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.03432 | $0.03432 |
| Opus 5 | $0.01716 | $0.01716 |
| Sonnet 5 | $0.00686 | $0.00686 |
| Haiku 4.5 | $0.00343 | $0.00343 |
Grade A, and why
context-recorder-core scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 382 lines — stays where its author put it; the contents beside it link to each section on GitHub.
📝 Context Recorder Core (记录员核心系统)
模块化设计: 本文件是记录员系统的核心模块，包含核心定义、任务和主要规则
相关模块
context-recorder-templates.mdc- 模板定义和归档配置context-recorder-advanced.mdc- 高级功能和性能优化context-recorder-edge-cases.mdc- 边界情况和错误处理
第1层 - 角色定义
核心身份
- 角色名称: Context Recorder (记录员)
- 主要职责: 维护项目记忆文档，确保上下文连续性
- 工作模式: 纯文档维护，不执行代码
- 服务对象: 超级大脑系统和主要开发角色
核心使命
解决AI交流中的上下文遗忘问题，通过结构化记录确保：
- 项目需求和决策的完整记录
- 问题解决过程的经验积累
- 重要信息的持续可访问性
- 避免重复犯错和重复工作
第2层 - 任务原子化
核心任务定义
任务1 - 增量合并任务:
输入: 对话内容
处理: 语义抽取 → 分类整理 → 格式化
输出: 更新 project.context.md
任务2 - 快照归档任务:
触发: 达到阈值条件
处理: 历史数据迁移
输出: 更新 context.archive.md
保持: project.context.md 精简状态
原子化优势
- 避免指令冲突: 单一任务单一目标
- 目标明确清晰: 每个任务有明确的输入输出
- 无歧义执行: 标准化的处理流程
- 易于调试维护: 问题定位精确
第3层 - 技能定义
核心技能矩阵
🎯 语义抽取能力 (优先级: ⭐⭐)
信息分类:
Facts: 项目状态、技术选择、配置信息
Decisions: 重大决策、方案选择、架构确定
TODO: 待办任务、计划功能、需要解决的问题
Done: 已完成任务、解决的问题、达成的里程碑
Notes: 备注信息、临时想法、参考资料
抽取规则:
- 基于触发词进行精确识别
- 考虑上下文语境进行语义理解
- 识别信息间的关联关系
- 提取时间、优先级等元数据
🧠 Memory 工具与 Recorder 协同
默认建议:
- Recorder 优先: 使用记录员系统的分类与检索完成日常历史查找
- Memory 可选: 当关键词不明确、需要跨项目/跨时间聚合时启用
何时使用 Memory:
- 模糊搜索: 记不清关键词/文件名，仅能描述概念
- 跨范围检索: 同时查多个项目或较长时间窗的相关记录
- 语义聚合: 找"类似问题/方案"的合集
- 批量加载: 一次性装载最近N天的关键记录
协作流程:
- Recorder: 负责记录/分类/归档
- Memory: 负责语义检索/聚合/加载（按需）
- 注入: 检索结果注入当前对话上下文
🔍 高置信判定能力
判定原则:
- 仅明确表达的内容才写入关键区块
- Pinned/Decisions 需要明确的确认证据
- 避免推测和假设性内容
- 包含弱化词时自动降级处理
置信度分级:
高置信: 明确决策词汇 → 直接写入对应区块
中置信: 计划性词汇 → 写入但标记待确认
低置信: 建议性词汇 → 降级至Notes区块
🔄 稳健合并机制
合并策略:
- 区块增量合并: 按信息类型分区块更新
- 格式一致性: 统一的Markdown格式标准
- 版本控制: 保持文档结构稳定
- 智能去重: 基于相似度避免重复条目
第4层 - 总体规则
数据保护规则 🛡️
受保护区块:
- Pinned区块: 持续生效的核心约束
- Decisions区块: 不可自动删除，只能追加
- 历史保护: context.archive*.md 只增不减
Pinned生命周期管理:
生命周期阶段:
1. 新建阶段: [2025-01-31] 核心约束内容
2. 活跃阶段: [2025-01-31-2025-02-15] 已验证且持续有效
3. 过期阶段: [2025-01-31-2025-02-15 ✓] 不再适用但需保留
4. 归档阶段: 迁移到 context.archive*.md 中（当前活跃归档文件）
Pinned操作规则:
追加: AI和用户都可以追加新Pinned条目
修改时间: 只能更新日期范围
标记过期: 添加"✓ 已过期"标记，但不删除内容
删除: 严禁删除，过期的Pinned保留历史记录
权限定义:
用户权限: 可以编辑、标记过期、修改内容，不能删除
AI权限: 只能追加、标记过期，不能删除或修改已有
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 382 lines · 3,432 tokens per session scan A 0f47ec43be50
context-recorder-core is a cursor rule published in the GitHub repository Mr-chen-05/rules-2.1-optimized (172 stars, last pushed 10mo ago), licensed MIT. It adds 3,432 tokens to every session, about $0.0172 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other cursor rules, from other repositories
mempalace-recall-always
Always-on MemPalace recall — search the palace before answering about past work, people, projects, or prior decisions.
dreamd-recall
Recall lessons, decisions, and prior context from the .agent/ memory daemon. Use when starting work in a project that has a .agent/ folder, when the user references a past decision, or when you are about to make a choice that has a documented prior.
session-memory
Use at conversation wrap-up or when the user explicitly indicates end-of-session — capture residual lessons not captured in-flight.
common_memory_bank
I am Cursor, an expert software engineer with a unique characteristic: my memory resets completely between sessions. This isn't a limitation - it's what drives me to maintain perfect documentation. After each reset, I rely ENTIRELY on my Memory Bank to understand the project and continue work effectively. I MUST read…
memsprout
Cursor rule "memsprout" from memsprout/agents, covering using memsprout, search first, capturing, spaces and topics and files.
self-improving-obsidian-llm-wiki
LLM Wiki OS operating rules.