longrun-prompt

Generate a copy-paste GitHub Copilot CLI orchestrator prompt plus recommended launch command from a natural-language task. Use when the user wants prompt generation, orchestration templates, autopilot starter text, or a reusable Copilot CLI command instead of immediate execution.

9 stars

byizscc

View on GitHub Installation ↓

Best use case

longrun-prompt is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Teams using longrun-prompt should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

You only need a quick one-off answer and do not need a reusable workflow.
You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/longrun-prompt/SKILL.md --create-dirs "https://raw.githubusercontent.com/izscc/Copilot-longrun/main/skills/longrun-prompt/SKILL.md"

Manual Installation

Download SKILL.md from GitHub
Place it in .claude/skills/longrun-prompt/SKILL.md inside your project
Restart your AI agent — it will auto-discover the skill

How longrun-prompt Compares

Feature / Agent	longrun-prompt	Standard Approach
Platform Support	Not specified	Limited / Varies
Context Awareness	High	Baseline
Installation Complexity	Unknown	N/A

Frequently Asked Questions

What does this skill do?

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

当用户要的是**prompt 产物**而不是立即执行时使用它。

如果用户其实要立刻开始长跑，请改用 `/longrun`。

## 1. 先分析任务画像

必须先提取或推断：
- goal
- profile: `coding | research | office`
- complexity: `single-lane | parallel | fleet`
- termination mode: `complete-and-exit | checkpoint-and-stop | watch-until-deadline`
- deliverables
- constraints
- language（默认跟随用户，中文优先）
- evidence mode（research / office 默认为 `balanced`）
- model policy（默认 `latest-available-opus-first`）

## 2. 模板选择

- `single-lane` -> Template A
- `parallel` -> Template B
- `fleet` -> Template C

参考：
- `references/prompt-generation-guide.md`
- `references/template-library.md`
- `references/specialist-labels.md`

## 3. 输出顺序

默认输出：
1. 任务分析摘要
2. 画像与复杂度判定
3. 模板选择
4. `## 可直接执行 Prompt`
5. `## 推荐启动命令`
6. 注意事项

如果用户只要 prompt，可省略 1/2/5/6。

## 4. Prompt 构造要求

生成的 prompt 必须体现：
- 单入口 `/longrun`
- 先任务画像，再规划执行
- 先声明 `terminationMode`
- 默认能力边界：本地文件 + shell + 公开网页
- 默认不双语；仅在用户明确要求时双语
- research / office：每个一级章节至少 2 个来源，文末 `## Sources Appendix`
- coding：本地验证优先
- deliverable 已完成时，优先 finalize，不要继续高成本验证
- 只有 `terminationMode == complete-and-exit` 时，才允许 `task_complete`
- 出现 rate limit 时，优先收尾或恢复，不要无意义 thrash

另外必须满足下面的格式约束，便于 launcher 自动整理成本地文件：

- `## 可直接执行 Prompt` 下面必须紧跟一个 fenced code block
- 这个外层 fenced code block **必须优先使用 `~~~~markdown`**（四个波浪线），不要用三反引号；因为 Prompt 正文里可能还会嵌套代码块
- 这个 code block 里只放 **Prompt 正文**，不要把分析摘要、注意事项、命令说明混进去
- `## 推荐启动命令` 下面必须紧跟一个 fenced `bash` code block
- 如果还要输出其他补充说明，放在这两个区块之外

## 5. 推荐启动命令

优先推荐 launcher：
```bash
longrun "<任务描述>"
```

若用户明确要原生 Copilot CLI 命令，再给：
```bash
copilot --autopilot --yolo --no-ask-user --model <当前账号可用的最新 Opus> --max-autopilot-continues <N> -p "/longrun <任务描述>"
```

`N` 推荐：
- Template A -> `20`
- Template B -> `50`
- Template C -> `100`

若 prompt 中显式指定模型，则启动命令应反映该模型；否则：
- 若已知本机模型缓存结果，使用“当前账号可用的最新 Opus”
- 若未知，提示先运行 `longrun-doctor` 探测模型能力；`copilot-longrun doctor` 仅作为兼容入口

## 6. 注意事项

必要时提醒：
- raw `/longrun` 若在普通 Copilot session 内手动敲入，会继承当前 session 权限，不保证无确认弹窗
- `longrun` / `copilot-longrun run`（兼容入口）才是更稳妥的无人值守入口
- raw `/longrun` 默认更像“跑到完成就退出”，不是守护进程；若任务要持续监控 / checkpoint / watch，优先建议 launcher + detach
- autopilot + yolo 有高自治权限，建议在隔离工作区使用

Related Skills

longrun

from izscc/Copilot-longrun

Orchestrate a long-running autonomous Copilot CLI mission from one prompt. Use when the user explicitly wants a one-shot, autopilot-style workflow with planning, workstream artifacts, evidence capture, recovery, and resumable state under .copilot-mission-control/.

longrun-status

from izscc/Copilot-longrun

Inspect the latest or a specified long-running Copilot CLI mission under .copilot-mission-control/ and report state, profile, current phase, delivered artifacts, blockers, evidence coverage, and the next likely step.

longrun-resume

from izscc/Copilot-longrun

Resume the latest or a specified long-running Copilot CLI mission from .copilot-mission-control/ without restarting completed work. Use when the user asks to continue, resume, or cleanly converge a previous /longrun run.

copilot-longrun-bridge

from izscc/Copilot-longrun

Use this skill when the user wants Codex to hand off a task to GitHub Copilot CLI LongRun for unattended execution, prompt generation, status checks, or resuming previous long-running missions. Trigger on requests mentioning longrun, Copilot CLI long tasks, resumable missions, or asking Codex to launch Copilot as the execution backend.

world-sufficiency-prompt

from plurigrid/asi

First-interaction system prompt generator for Gemini, Codex, and Claude. Detects hierarchical user intent (implicit and explicit) and loads GF(3)-balanced skill triads to achieve World -> World' sufficiency before any model response. Bridges dynamic-sufficiency theory to concrete systemInstruction/system-message payloads across all three providers.

detecting-ai-model-prompt-injection-attacks

from plurigrid/asi

Detects prompt injection attacks targeting LLM-based applications using a multi-layered defense combining regex pattern matching for known attack signatures, heuristic scoring for structural anomalies, and transformer-based classification with DeBERTa models. The detector analyzes user inputs before they reach the LLM, flagging direct injections (system prompt overrides, role-play escapes, instruction hijacking) and indirect injections (encoded payloads, multi-language obfuscation, delimiter-based escapes). Based on the OWASP LLM Top 10 (LLM01:2025 Prompt Injection) and Simon Willison's prompt injection taxonomy. Activates for requests involving prompt injection detection, LLM input sanitization, AI security scanning, or prompt attack classification.

prompt-injection-defense

from 0xDarkMatter/claude-mods

Defend the agent's instruction surface against adversarial content - hidden-Unicode prompt injection (Trojan Source bidi reordering, U+E0000 tag-block ASCII smuggling, zero-width text), homoglyph confusables, and poisoned context that a human reviewer can't see but the model obeys. Scan CLAUDE.md / AGENTS.md / SKILL.md / .cursorrules and MCP tool descriptions; sanitize fetched web pages, issue/PR bodies, and dependency READMEs before they enter context. Triggers on: prompt injection, hidden unicode, invisible characters, zero-width space, bidi override, Trojan Source, ASCII smuggling, tag characters, homoglyph, confusable, unicode steganography, poisoned CLAUDE.md, malicious tool description, MCP tool poisoning, instruction injection, jailbreak in file, is this file safe, sanitize untrusted content, scan for hidden text.

customaize-agent:prompt-engineering

from Zerone-Agent/agent-use-skills

Use this skill when you writing commands, hooks, skills for Agent, or prompts for sub agents or any other LLM interaction, including optimizing prompts, improving LLM outputs, or designing production prompt templates.

empirical-prompt-tuning

from sushichan044/dotfiles

agent 向けテキスト指示（skill / slash command / task プロンプト / CLAUDE.md 節 / コード生成プロンプト）を、バイアスを排した実行者に動かしてもらい、両面（実行者の自己申告 + 指示側メトリクス）で評価して反復改善する手法。改善が頭打ちになるまで回す。プロンプトや skill を新規作成・大幅改訂した直後、またはエージェントの挙動が期待通りにならない原因を指示側の曖昧さに求めたいときに使う。

prompt-evaluator

from nguyenthienthanh/aura-frog

Evaluate how you use Claude Code — analyze prompt patterns, feature utilization, and get improvement suggestions. Trigger: /prompts:evaluate, prompt analysis, usage evaluation, how am I using Claude

prompt-engineering

from NickCrew/Claude-Cortex

Optimize prompts for LLMs and AI systems with structured techniques, evaluation patterns, and synthetic test data generation. Use when building AI features, improving agent performance, or crafting system prompts.

iblai-security-prompt-injection

from iblai/vibe

Audit applications for AI prompt injection, agent security, and LLM permission boundary vulnerabilities. Use when the user mentions 'prompt injection,' 'LLM security,' 'AI security,' 'jailbreak,' 'indirect prompt injection,' 'prompt leaking,' 'AI red team,' 'LLM vulnerabilities,' 'AI input validation,' 'system prompt extraction,' 'agent security,' 'MCP security,' 'AI permissions,' 'AI privilege escalation,' or needs to secure any application with AI features, AI agents, or LLM integrations.