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 skills add FirenzeClaw/kimi-session-orchestrator --skill xmind-orchestratedgit clone --depth 1 https://github.com/FirenzeClaw/kimi-session-orchestratorWrote 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/skills/firenzeclaw/kimi-session-orchestrator/xmind-orchestrated)<a href="https://agentmods.dev/skills/firenzeclaw/kimi-session-orchestrator/xmind-orchestrated"><img src="https://agentmods.dev/badge/skills/firenzeclaw/kimi-session-orchestrator/xmind-orchestrated/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/firenzeclaw/kimi-session-orchestrator/xmind-orchestrated"><img src="https://agentmods.dev/badge/skills/firenzeclaw/kimi-session-orchestrator/xmind-orchestrated.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00076 | $0.01743 |
| Opus 5 | $0.00038 | $0.00872 |
| Sonnet 5 | $0.00015 | $0.00349 |
| Haiku 4.5 | $0.00008 | $0.00174 |
Grade A, and why
xmind-orchestrated 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 10d 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 — 162 lines — stays where its author put it; the contents beside it link to each section on GitHub.
XMIND Orchestrated — 困境分析 + 独立上下文求解
⛔ 加载即执行
Auto 检测
若 auto permission mode → 纯文本提问,不用 AskUserQuestion。
Phase 0: 收集事实(三问)
Q1: 问题描述? 现象 + 发生时机 + 影响范围(必须具体) Q2: 相关文档与日志? 文件路径 + 关键日志/文档原文摘要(至少1个;路径必须可访问) Q3: 已尝试方案? 每次尝试的方案描述 + 可观测结果(可无;有则必写失败原因)
每问验证非空(Q3 可为空)。禁止在收集事实前开始分析。
Phase 1: 认知过滤
扫描 Q1-Q3 回答,移除含以下词的全句:
我觉得|可能是|怀疑|应该不是|好像|大概|估计|我认为|不太可能
转换示例:
- "之前试过改 timeout,好像没效果" → "已尝试增加 timeout 参数(无效果)"
- "我怀疑是 A 模块的问题" → 整句删除
过滤后的内容 = 纯事实包。
Phase 2: 路由
调用 get_tunnel_status:
wireConnected: true→ Phase 3A(task session 主路径)wireConnected: false或工具不可用 → Phase 3B(子 Agent 降级)- 3 秒内未确定 → 降级
Phase 3A: task session 路径
前提: memory_get(namespace="project/decisions") → 已知约束
① create_session(cwd, permission_mode="auto")
→ 失败 → 降级 Phase 3B
② execute_prompt(sid, prompt) → 注入模板:
[系统注入] 你是独立的问题分析 session。
以下均为客观事实,不含任何主观判断。
## 问题描述
<Q1 过滤后>
## 相关文档与日志
<Q2 原文>
## 已尝试方案
<Q3 过滤后,仅保留方案描述+可观测结果>
## 已知约束
<project/decisions 条目,仅事实型决策>
---
请按以下步骤分析:
1. 读取上述文档和日志,理解项目上下文
2. 调用 zoom-out(可用时)或手动绘制模块关系图:用项目术语描述全局架构,标注模块间依赖
- 若 zoom-out skill 不可用:手动绘制模块依赖图,用项目术语描述全局
3. 基于宏观理解,提出 1-3 个新方案
4. 每个方案标注: 可行性(高/中/低)+理由+风险+与已尝试方案的本质区别
⛔ 若缺少关键信息,列出需补充的内容。
③ Bash(run_in_background=true, command=poll_command)
④ 等 <notification> → list_io_records → 提取方案
⑤ 若空返回 → 降级 Phase 3B 重试一次
Phase 3B: 子 Agent 降级
Agent(subagent_type="coder", prompt=
"## 问题\n<Q1 过滤后>\n\n## 文档\n<Q2>\n\n## 已尝试\n<Q3>\n\n
请阅读文档理解项目架构,分析问题,提出 1-3 个新方案。"
)
保留原 xmind zoom-out 流程:读文档 → 分析 → 宏观视角 → 方案。
Phase 4: 迭代
| 用户决策 | 行为 |
|---|---|
| 接受 | 结束。方案交付。 |
| 拒绝(< 3轮) | 路径 A: memory_set 记拒绝原因 → 新 session;路径 B: 拒绝原因写入下轮 prompt 前缀 → 新子 Agent |
| 拒绝(≥ 3轮) | 输出 3 轮完整记录,标记"需人工决策" |
拒绝原因也需过滤主观词。
工具规范
| 工具 | 路径 | 用途 |
|---|---|---|
get_tunnel_status |
A+B | 路由检测 |
create_session |
A | 创建独立 session |
execute_prompt |
A | 下发分析任务 |
Bash(poll_command) |
A | 后台等结果 |
list_io_records |
A | 回收结果 |
memory_get |
A | 读取已知约束 |
memory_set |
A | 记录拒绝原因 |
Agent |
B | 降级子 Agent |
AskUserQuestion |
A+B | 三问(非 auto) |
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.
- 10d ago First seen · 162 lines · 76 tokens per session scan A 85e5772858e3
xmind-orchestrated is a skill published in the GitHub repository FirenzeClaw/kimi-session-orchestrator (0 stars, last pushed 14d ago), licensed MIT. It adds 76 tokens to every session and 1,743 once invoked, about $0.0004 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 2026-08-31.
Other skills, from other repositories
output-dev-skill-file
Create .md skill files for Output framework's lazy-loaded instruction system. Use when adding skills to prompts, configuring skill loading, or debugging skill resolution.
output-debug-workflow
Debug Output SDK workflow issues. Use when user reports a workflow failing, erroring, hanging, producing wrong results, or asks to debug, troubleshoot, or investigate a workflow execution.
output-services-check
Verify Output SDK development services are running. Use when debugging workflows, starting development, encountering connection errors, services may be down, or when you see "ECONNREFUSED" or timeout errors.
output-workflow-trace
Analyze Output SDK workflow execution traces. Use when debugging a specific workflow, examining step failures, analyzing input/output data, understanding execution flow, or when you have a workflow ID to investigate.
scientific-debugging
A method for debugging software by observing the problem, forming possible explanations, running small experiments, and then fixing and checking the result.
diagnose-from-context
Diagnose a problem from the explicitly materialized context view.