About this skill
This skill defines a sophisticated operational framework for AI agents, enabling them to execute tasks in a continuous, autonomous loop. It integrates critical software engineering best practices, including explicit quality gates (e.g., `plankton-code-quality`), iterative evaluation mechanisms (e.g., `eval-harness`), and systematic recovery controls to address common failure modes. The skill provides various loop selection processes—such as `continuous-pr` for strict CI/PR control, `rfc-dag` for structured decomposition, `infinite` for exploratory parallel generation, and `sequential` as a default—allowing agents to adapt their operational strategy to specific project needs. It also outlines recommended production stacks and strategies for handling common issues like "loop no-ops" or "cost drift." Essentially, it empowers an AI agent to self-manage its iterative progress, ensuring quality and resilience in complex, ongoing projects, particularly in software development contexts.
Best use case
Managing and orchestrating complex, long-running AI agent tasks, especially in software development, code generation, or iterative problem-solving where continuous progress, quality assurance, and self-correction are critical. It's ideal for projects requiring a structured approach to continuous delivery or development cycles.
具有质量门、评估和恢复控制的连续自主代理循环模式。
An AI agent that can execute tasks continuously, maintaining quality, evaluating its own progress, and autonomously recovering from predefined failure states. It leads to more robust, reliable, and efficient long-term operations, reducing human oversight for iterative development cycles and complex projects.
Practical example
Example input
An instruction to the agent to operate within this loop pattern, potentially specifying loop selection and integrated sub-skills. For example: "Operate in a continuous development loop. Use `rfc-dag` for task decomposition and integrate `plankton-code-quality` for code reviews. Prioritize stable releases, and in case of stalled progress, initiate a `/harness-audit` for recovery."
Example output
The "output" is the agent's ongoing behavior and the results of its continuous operation, including progress reports, code artifacts, test results, and self-correction logs. Example: "Agent initialized in `continuous-agent-loop` mode, using `rfc-dag` for planning. Phase 1: RFC decomposition complete for feature X. Phase 2: Code generation initiated for module A. Quality Gate Check (via `plankton-code-quality`): Pass. Evaluation (via `eval-harness`): 85% test coverage, 2 minor bugs found. Recovery initiated: Identified root cause for bug #123, retrying code generation for affected component. Progress: Feature X is 30% complete, moving to module B development."
When to use this skill
- When an AI agent needs to operate autonomously over an extended period.
- For software development tasks where continuous integration, pull request controls, or RFC-based decomposition are desired.
- When robust quality assurance, automated evaluation, and self-healing capabilities are paramount for an agent's output.
- For iterative problem-solving that requires measurable progress and recovery from common failure modes.
When not to use this skill
- For simple, one-shot tasks that don't require continuous iteration or complex state management.
- When the overhead of quality gates, evaluation, and recovery mechanisms is unnecessary for the task's scope or criticality.
- In highly constrained environments where advanced orchestration or the use of multiple sub-skills is not feasible.
- For tasks where human intervention is explicitly preferred at every step rather than autonomous operation.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/continuous-agent-loop/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How continuous-agent-loop Compares
| Feature / Agent | continuous-agent-loop | Standard Approach |
|---|---|---|
| Platform Support | Claude | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | medium | N/A |
Frequently Asked Questions
What does this skill do?
具有质量门、评估和恢复控制的连续自主代理循环模式。
Which AI agents support this skill?
This skill is designed for Claude.
How difficult is it to install?
The installation complexity is rated as medium. You can find the installation instructions above.
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.
Related Guides
AI Agents for Coding
Browse AI agent skills for coding, debugging, testing, refactoring, code review, and developer workflows across Claude, Cursor, and Codex.
Best AI Skills for Claude
Explore the best AI skills for Claude and Claude Code across coding, research, workflow automation, documentation, and agent operations.
Top AI Agents for Productivity
See the top AI agent skills for productivity, workflow automation, operational systems, documentation, and everyday task execution.
SKILL.md Source
# 持续代理循环 这是 v1.8+ 的规范循环技能名称。它在保持一个发布版本的兼容性的同时，取代了 `autonomous-loops`。 ## 循环选择流程 ```text Start | +-- 需要严格的 CI/PR 控制？ -- yes --> continuous-pr | +-- 需要 RFC 分解？ -- yes --> rfc-dag | +-- 需要探索性并行生成？ -- yes --> infinite | +-- default --> sequential ``` ## 组合模式 推荐的生产栈： 1. RFC 分解 (`ralphinho-rfc-pipeline`) 2. 质量门 (`plankton-code-quality` + `/quality-gate`) 3. 评估循环 (`eval-harness`) 4. 会话持久化 (`nanoclaw-repl`) ## 故障模式 * 循环空转，没有可衡量的进展 * 因相同根本原因而重复重试 * 合并队列停滞 * 无限制升级导致的成本漂移 ## 恢复 * 冻结循环 * 运行 `/harness-audit` * 将范围缩小到失败单元 * 使用明确的验收标准重放
Related Skills
autonomous-loops
自主Claude代码循环的模式与架构——从简单的顺序管道到基于RFC的多智能体有向无环图系统。
continuous-learning-v2
フックを介してセッションを観察し、信頼度スコアリング付きのアトミックなインスティンクトを作成し、スキル/コマンド/エージェントに進化させるインスティンクトベースの学習システム。
workspace-surface-audit
Audit the active repo, MCP servers, plugins, connectors, env surfaces, and harness setup, then recommend the highest-value ECC-native skills, hooks, agents, and operator workflows. Use when the user wants help setting up Claude Code or understanding what capabilities are actually available in their environment.
safety-guard
Use this skill to prevent destructive operations when working on production systems or running agents autonomously.
repo-scan
Cross-stack source code asset audit — classifies every file, detects embedded third-party libraries, and delivers actionable four-level verdicts per module with interactive HTML reports.
project-flow-ops
Operate execution flow across GitHub and Linear by triaging issues and pull requests, linking active work, and keeping GitHub public-facing while Linear remains the internal execution layer. Use when the user wants backlog control, PR triage, or GitHub-to-Linear coordination.
manim-video
Build reusable Manim explainers for technical concepts, graphs, system diagrams, and product walkthroughs, then hand off to the wider ECC video stack if needed. Use when the user wants a clean animated explainer rather than a generic talking-head script.
laravel-plugin-discovery
Discover and evaluate Laravel packages via LaraPlugins.io MCP. Use when the user wants to find plugins, check package health, or assess Laravel/PHP compatibility.
design-system
Use this skill to generate or audit design systems, check visual consistency, and review PRs that touch styling.
click-path-audit
Trace every user-facing button/touchpoint through its full state change sequence to find bugs where functions individually work but cancel each other out, produce wrong final state, or leave the UI in an inconsistent state. Use when: systematic debugging found no bugs but users report broken buttons, or after any major refactor touching shared state stores.
ck
Persistent per-project memory for Claude Code. Auto-loads project context on session start, tracks sessions with git activity, and writes to native memory. Commands run deterministic Node.js scripts — behavior is consistent across model versions.
canary-watch
Use this skill to monitor a deployed URL for regressions after deploys, merges, or dependency upgrades.