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 loop-contract-from-ideagit 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/loop-contract-from-idea)<a href="https://agentmods.dev/skills/firenzeclaw/kimi-session-orchestrator/loop-contract-from-idea"><img src="https://agentmods.dev/badge/skills/firenzeclaw/kimi-session-orchestrator/loop-contract-from-idea/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/loop-contract-from-idea"><img src="https://agentmods.dev/badge/skills/firenzeclaw/kimi-session-orchestrator/loop-contract-from-idea.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 | $0.00041 | $0.02099 |
| Opus 5 | $0.00020 | $0.01050 |
| Sonnet 5 | $0.00008 | $0.00420 |
| Haiku 4.5 | $0.00004 | $0.00210 |
Grade A, and why
loop-contract-from-idea 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 8d 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 — 194 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Loop Contract From Idea — 一句话需求到严格循环契约
概述
此 skill 将一句话需求转换为 .kimi-tunnel/loop-contract.yaml。核心原则:先追问,再锁定基线,再生成 Contract;未确认的推断不得伪装成事实。
共享模板只引用,不复制:skills/loop-contract-from-docs/templates/loop-contract.yaml。
何时使用
- 用户调用
/loop-contract-new - PM 只有一句话需求,尚无可直接执行的 Loop Contract
- 需要先从目标、约束、技术栈、方向边界推断 AC,并让用户确认后再进入循环编排
不用于用户已给出 SPEC/PLAN/PRD/TASK 文档集的场景。此时升级到 loop-contract-from-docs 路径。
加载即执行
Auto 检测
若系统提示含 Auto permission mode is active,用纯文本逐轮追问,不调用 AskUserQuestion。非 auto 模式可用结构化问题,但仍必须逐轮完成。
Phase 1 — 5 轮追问
加载 skills/loop-contract-from-idea/guide-clarify-chain.md,按顺序完成 Q1-Q5:
- Q1 最终目标:谁用、做什么、为什么重要。
- Q2 平台/环境约束:OS、浏览器、嵌入式、性能基线等。
- Q3 技术栈:语言、框架、数据库;用户明确“不限”时可跳过细化。
- Q4 方向约束:明确不做什么、不碰什么模块、不可降级边界。
- Q5 是否已有相关文档:若有文档,立即升级为
loop-contract-from-docs路径继续,不再走 from-idea 推断。
硬门:Phase 1 不可跳过。即使用户回答不充分,也必须按 guide 的追问策略继续追问,直到达到每轮退出标准或明确记录“不足但已追问”。
完成标准:Q1-Q5 均有记录;Q5 已判定继续 from-idea 或升级 from-docs。
Phase 2 — 基线锁定
加载 skills/loop-contract-from-idea/guide-baseline-lock.md。
- 从 Q1-Q4 推断 AC 草案,每条必须可验证。
- 从 Q4 和 Q1-Q3 的边界信息推断
direction_baseline。 - 向用户展示“PM 推断的 AC 草案 + 方向基线草案”。
- 用户确认“是”后写入 Contract 草案,此后 PM 不得自行修改。
- 用户要求修改时,走变更流程:改草案 → 重新展示 → 重新确认 → 再锁定。
完成标准:AC 与 direction_baseline 均已锁定,且锁定依据能追溯到用户确认。
Phase 3 — 复杂度评估
基于 Q2-Q4 评估 complexity.level 与 complexity.factors:
| Level | 判定线索 | Loop 参数 |
|---|---|---|
| low | 单模块、无协议/DB/跨平台约束,AC 少于 4 条 | min_cycles: 2, max_cycles: 4, consecutive_pass_required: 2 |
| medium | 2-3 个模块、存在接口/状态/兼容性约束 | min_cycles: 3, max_cycles: 5, consecutive_pass_required: 2 |
| high | 跨系统/协议/DB schema/安全/性能基线,或方向约束严格 | min_cycles: 4, max_cycles: 6, consecutive_pass_required: 3 |
完成标准:complexity.factors 至少 2 条;循环参数不得低于 low 的保障线。
Phase 4 — 生成 Contract
使用共享模板路径 skills/loop-contract-from-docs/templates/loop-contract.yaml 填充同格式 Contract,输出到 .kimi-tunnel/loop-contract.yaml。
必须保持与 from-docs 格式一致:
meta:
generated_by: loop-contract-from-idea
generated_at: <ISO timestamp>
source_docs:
- user-input
goal:
statement: "<一句话目标>"
acceptance_criteria:
- "<可验证条件>"
direction_baseline:
- "<范围边界>"
- "<不做什么或不可降级边界>"
complexity:
level: low | medium | high
factors:
- "<因素>"
loop_config:
min_cycles: 2
max_cycles: 5
degradation_detection: true
consecutive_pass_required: 2
verify:
primary: grade_step
cross_model: <must-differ-from-pm-model>
self_audit_rounds: 1
operational_brakes:
max_repeated_actions: 1
max_wall_time_minutes: 60
context_tokens_threshold: 36000
no_progress_escalation: "暂停当前轮 → 记录状态 → PM 诊断或退役接班"
harness:
required_tools:
- "kimi-session-orchestrator MCP"
- "loop-orchestrator"
- "grade_step"
allowed_paths:
- "<project-scope>"
forbidden_paths:
- "<out-of-scope>"
state_namespace: "session/<loop-id>"
kill_switch: "PM says stop, repeated no-progress, or delivery_gate cannot be made verifiable"
blockage_recovery:
max_retries: 3
strategies:
- "xmind-orchestrated → 多视角分析"
- "create_session(research) → 网络搜索或文档检索"
escalation: "3 次仍阻塞 → 暂停并报告"
pm_duties:
per_cycle:
- "对照 direction_baseline 核查方向一致性"
- "检测降级倾向:缩减范围/跳过验证/省略检查"
- "检查 operational_brakes 是否触发"
on_blockage:
- "先诊断后决策"
delivery_gate:
- min_cycles 满足
- consecutive_pass_required 满足
- direction_baseline 全部通过
- operational_brakes 未触发或已完成 PM 处置
- harness.allowed_paths / forbidden_paths 未被违反
- PM 确认:方向一致且未发现降级
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 8d ago First seen · 194 lines · 41 tokens per session scan A 3a01bcd75fe8
loop-contract-from-idea is a skill published in the GitHub repository FirenzeClaw/kimi-session-orchestrator (0 stars, last pushed 12d ago), licensed MIT. It adds 41 tokens to every session and 2,099 once invoked, about $0.0002 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
mp-implement
Use when the user provides any kind of feature spec (rough notes / system design / partial spec / canonical PRD) and wants it implemented end-to-end autonomously on Claude Code alone — no external execution engine — or invokes /mp-implement.
recipe-create-meet-space
Create a Google Meet meeting space and share the join link.
workthreads
SpecStory Workthreads - a weekly work-thread rollup across a team's repos from SpecStory coding histories (any agent - Claude Code, Codex, Cursor, Gemini, and more). It groups the window's sessions into threads of work per project and labels each new / open / recently closed, so a lead sees what shipped, what is still…
atmos-config
Atmos root configuration: atmos.yaml discovery, precedence, deep merging, basepath, imports, minimal bootstrap, and routing to narrower Atmos skills.
story-readiness
Validate that a story file is implementation-ready. Checks for embedded GDD requirements, ADR references, engine notes, clear acceptance criteria, and no open design questions. Produces READY / NEEDS WORK / BLOCKED verdict with specific gaps. Use when user says 'is this story ready', 'can I start on this story', 'is…
autotask-creator
Rules for automation CRUD from the group-chat commander. The commander does not call mutation tools and does not edit cloud/autotasks files directly. It emits one or more top-level ... containers in its final text; the bus parses and applies them after the turn.