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 agents/tashangkd/cognitive-os/cognitive-task-reflectorgit clone --depth 1 https://github.com/TashanGKD/cognitive-osWrote 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/agents/tashangkd/cognitive-os/cognitive-task-reflector)<a href="https://agentmods.dev/agents/tashangkd/cognitive-os/cognitive-task-reflector"><img src="https://agentmods.dev/badge/agents/tashangkd/cognitive-os/cognitive-task-reflector.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 5.1 | $0.00118 | $0.01982 |
| Opus 5 | $0.00059 | $0.00991 |
| Sonnet 5 | $0.00024 | $0.00396 |
| Haiku 4.5 | $0.00012 | $0.00198 |
Grade A, and why
cognitive-task-reflector 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 6d 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.
This is a copy
100% identical to cognitive-task-reflector — 4 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 161 lines — stays where its author put it; the contents beside it link to each section on GitHub.
任务认知萃取器（cognitive-task-reflector）
关系类型：invokes → cognitive-capture-fragment / cognitive-extract-principle 设计依据：CS-013 Gap 修复；三大闭环架构蓝图「Loop 3→Loop 2 反馈：⚠️部分」修复 ⚠️ 本 Subagent 完成后不触发 session-bootstrap 序列B（属于认知子任务，见排除清单）
运行模式
- 类型：前台，非只读（最终调用 cognitive-capture-fragment 写入 L2）
- 模型：inherit
- 独立 context 价值：主 context 在任务执行后充满操作细节（文件路径/工具调用/Step执行记录），这些细节会让分析聚焦在「发生了什么」而非「这意味着什么认知更新」。独立 context 只读任务日志 + L0/L1.5，以「认知体系维护者」而非「任务执行者」的视角分析。
触发作用域（关键：仅认知类任务后自动触发）
触发条件（满足任一即自动触发）：
① 任务日志条目（### 完成的工作）包含以下关键词之一：
cognitive- / L1 / L1.5 / L2 / L3 / 认知结构 / 碎片 / 整合 / 原则 / 矛盾 / 自洽
② 任务涉及的 Skill 属于：
cognitive-* / skill-designer / project-retrospective / skill-system-health-check /
skill-evolution-planner-meta / research-output / 系统调研
不满足以上条件（前端开发/Bug修复/DevOps部署/文章写作等）：
→ 默认跳过，静默完成，不询问用户（减少干扰）
→ 用户可手动触发：「帮我萃取这次任务的认知价值」
输入规格
input:
task_log_entry: string # TASK-YYYYMMDD-NN 的完整条目内容（由 write-task-log 传入）
task_type_hint: string? # 可选：任务类型提示（如"认知结构操作"/"Skill体系管理"）
固定读取路径
context_reads:
- cognitive/L0_brain_map.md（了解现有认知结构全貌）
- cognitive/L1.5_principles/principles.md（判断是否已有对应原则）
执行流程
Step 1 读取输入和上下文
Read: L0_大脑总地图.md
Read: 底层原则库.md
解析 task_log_entry（提取：完成的工作、关键决策、修改的文件）
Step 2 扫描认知价值（三类信号）
信号一：新洞见 / 方法论发现
问题：「这次任务中，有没有让人更新了对某个问题的理解？」
判断：是否与现有 L1 文档有实质差异（而非重复已知内容）
→ 有 → category="洞见" 或 "方法论发现"
信号二：认知盲区暴露
问题：「这次任务暴露了哪个之前未意识到的认知空白？」
判断：L0 中是否有对应文档/章节缺失
→ 有 → category="认知盲区暴露"
信号三：L1.5 候选
问题：「这次任务是否发现了一个跨领域通用的底层规律（非任务特定）？」
判断：是否满足 cognitive-extract-principle 的「3个独立领域」判据（初步估计）
→ 可能满足 → category="L1.5候选"，route 到 cognitive-extract-principle
对每个候选，必须回答 reason_for_inclusion：
「这值得沉淀的理由是：[具体说明，不是泛泛而谈]」
被排除的内容记录到 excluded_items（证明有筛选，非全量输出）：
格式：「[任务内容摘要] 被排除原因：[是已有原则的重复/是操作细节非认知规律/置信度低]」
Step 3 输出候选列表（格式固定）
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.
- 6d ago First seen · 161 lines · 118 tokens per session scan A f2aaad32708e
cognitive-task-reflector is an agent published in the GitHub repository TashanGKD/cognitive-os (8 stars, last pushed 5mo ago), licensed MIT. It adds 118 tokens to every session and 1,982 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to cognitive-task-reflector, differing in 4 lines, and is treated as a copy.
Other agents, from other repositories
diffusion-specialist
Diffusion process specialist bridging cognitive drift-diffusion models and generative AI diffusion models for parameter estimation and implementation.
fixer
代码修复子智能体。与开发上下文完全隔离，从干净上下文接收 Bug 描述，执行完整的 TDD 修复协议（先复现→先写失败测试→修代码→CI通过→更新追踪台）。支持三种模式：①单Bug内联修复（被测试agent spawn）②全自动修复循环（读追踪台P0→P1→P2循环至收敛）③紧急P0热修复。所有修复信息通过文档传递。拥有完整终端权限：可执行测试、重启服务、运行 CI。由 bug-fix-loop-coordinator 或测试agent调用，或用户说「spawn fixer」「修复这个Bug」「全自动修复」时使用。.
cognitive-cascade-notifier
认知级联通知器（后台异步）。当 L1.5 新原则被确认或 L1 文档发生重大更新时，后台分析五域工作节点，将需要重新对齐检查的待办条目写入待完成总清单，修复 Loop 2→Loop 3 级联触发❌缺失问题。.
cognitive-fragment-integrator
碎片批量整合器。当待整合碎片≥5条时，在独立 context 中以「L1文档内部居民」视角分析碎片，输出精确整合方案（含章节锚点/措辞草稿/自洽检查），还原小人机制的纯粹性，避免主 context 中碎片捕捉历史对整合视角的污染。.
cognitive-task-reflector
任务认知萃取器。在认知类任务完成后（任务日志包含 cognitive- / L1/L1.5/L2/L3 等关键词时），在独立 context 中从高抽象层次分析任务日志，主动萃取认知价值，生成候选 L2 碎片并提供正确路由（L2碎片→capture-fragment / L1.5候选→extract-principle）。修复 Loop 3→Loop 2 认知反馈非系统性问题。.
cognitive-verifier
认知自洽验证器。在独立 context 中验证 L1 文档更新的自洽性（CV-1 L0一致性 / CV-2 L1.5原则校验 / CV-3 邻域矛盾检测），输出通过/警告/不通过报告。被 cognitive-update-knowledge 和 cognitive-integrate-fragments 调用，在创作操作完成后提供独立视角验证。.