Token导航 LogoToken导航

Agent Skill · 研究检索

context-continuity-code(上下文连续性代码)

context-continuity-code 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

context-continuity-code 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证为 MIT,暂无第三方扫描结果,按安装量在研究检索类的 73,675 个 Skill 里排第 69,087。

总安装 220周安装 9GitHub Stars 公开资料未说明下载 71核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:context-continuity-code(上下文连续性代码)
来源仓库:https://github.com/leegonzales/aiskills
仓库路径:skills/context-continuity-code
安装命令:
npx skills add leegonzales/aiskills --skill "context-continuity-code"
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

AgentSkills.tonpx skills
npx skills add leegonzales/aiskills --skill "context-continuity-code"

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

context-continuity-code 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。适用于需要从多个来源中筛选出与特定任务相关的信息片段的场景,如代码审查、文档整理或研究支持。

通过关键词匹配和来源仓库路径提供精准的结果定位,结合原始 README 可进一步验证具体功能和使用方法。安装命令为 npx skills add leegonzales/aiskills --skill "context-continuity-code",需确认权限范围及是否涉及联网或文件操作。

建议在使用前核实仓库维护状态,避免依赖长期未更新的技能实现。

来源仓库内路径:skills/context-continuity-code

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add leegonzales/aiskills --skill "context-continuity-code";命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

context-continuity-code 怎么安装?

在终端执行 npx skills add leegonzales/aiskills --skill "context-continuity-code"。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

context-continuity-code 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

context-continuity-code 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

context-continuity-code 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

context-continuity-code 用的人多吗?

公开数据是累计安装 220 次,近一周安装 9 次,下载 71 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

context-continuity-code 安全吗?

没有查到第三方扫描结果,安全性无法从公开资料判断,安装前请自行审查源码、依赖和命令行为。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 许可证明确为 MIT,商用和分发边界好判断。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 没有第三方扫描结果,安全性只能靠自己读源码判断。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 要求有安全扫描结论才肯安装的团队。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
context-continuity-code

上下文连续性代码 · 本页

220—接触敏感数据MIT未收录

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
context-continuity-code
description
Claude Code-optimized context transfer for development workflows. Preserves code context, git state, running services, and development environment when moving work between sessions. Works seamlessly with peer review skills (Codex/Gemini). Use when transferring development work to a new Claude Code session.

Context Continuity - Claude Code Edition

Transfer development context between Claude Code sessions with high fidelity, preserving code state, git context, and running services.

Core Concept

When development work needs to transfer between Claude Code sessions, this skill creates structured artifacts that capture:

  • Code Context - Files being worked on, functions/classes modified, pending changes
  • Git State - Branch, commits, staged/unstaged changes, merge status
  • Environment State - Running services, ports, environment variables, dependencies
  • Development Decisions - Technical choices made, alternatives rejected, tradeoffs
  • Open Loops - What's next, blockers, pending reviews

When to Use This Skill

Use when you need to:

  • Continue development work in a fresh Claude Code session
  • Hand off work to another developer (with AI context preserved)
  • Resume after context window fills (180K+ tokens)
  • Switch between different Claude Code instances
  • Document current state before major refactoring
  • Prepare for peer review (Codex/Gemini integration)

DO NOT use for:

  • General conversation summaries (use base context-continuity skill)
  • Non-development contexts (writing, research, analysis)
  • Simple task handoffs without code changes

Workflow: Single Mode (Development-Optimized)

This skill uses a single optimized mode for development contexts (~400-600 words):

═══════════════════════════════════════════════════════════════════
DEV CONTEXT TRANSFER
═══════════════════════════════════════════════════════════════════
Generated: [ISO timestamp] | Session: [ID if available]

**MISSION**: [What we're building/fixing + why it matters]

**STATUS**: [✓ complete | ⧗ in-progress | ⚠ blocked | ↻ iterating]

**PROGRESS**: [High-level summary of work completed this session]

───────────────────────────────────────────────────────────────────
§ CODE CONTEXT
───────────────────────────────────────────────────────────────────

**Active Files**:
- [file:line] - [What changed or what you're working on]
- [file:line] - [Status: implemented | in-progress | pending]

**Key Changes**:
- [Function/class]: [What changed + why]
- [Module/component]: [Refactoring/addition/fix]

**Code State**:
- Modified: [List files with uncommitted changes]
- Created: [New files added]
- Deleted: [Files removed]

───────────────────────────────────────────────────────────────────
§ GIT STATE
───────────────────────────────────────────────────────────────────

**Branch**: [current-branch-name]
**Base**: [main/master/develop]
**Commits**: [X commits ahead of base]

**Recent Commits**:
- [hash] [message]
- [hash] [message]

**Staged**: [Files in staging area | None]
**Unstaged**: [Modified files not staged | None]
**Untracked**: [New files not in git | None]

**Merge Status**: [Clean | Conflicts in X files | Pending PR #XXX]

───────────────────────────────────────────────────────────────────
§ ENVIRONMENT STATE
───────────────────────────────────────────────────────────────────

**Running Services**:
- [Service]: [localhost:PORT] - [Status: healthy | issue]
- [Database]: [postgres:5432] - [State: connected, X records]

**Dependencies**:
- Recently installed: [package@version, ...]
- Pending: [Packages needed but not installed]

**Environment Variables**:
- Critical vars set: [APP_ENV=dev, DATABASE_URL=localhost, ...]
- Missing/needed: [API_KEY (required for testing), ...]

**Terminal State**:
- Active shells: [X terminals open in /path/to/project]
- Background processes: [npm run dev, docker compose up, ...]

───────────────────────────────────────────────────────────────────
§ TECHNICAL DECISIONS
───────────────────────────────────────────────────────────────────

**Decisions Made**:
| Decision | Rationale | Alternatives Rejected | Tradeoff |
|----------|-----------|----------------------|----------|
| [Choice] | [Why] | [What we didn't do] | [Cost we're paying] |

**Architecture Notes**:
- [Pattern/approach chosen]: [Why it fits this context]
- [Constraint observed]: [Technical/business reason]

**Peer Review Integration**:
- Codex consulted: [Yes/No] - [If yes: key recommendations]
- Gemini consulted: [Yes/No] - [If yes: key recommendations]
- Agreements: [Where both AIs aligned]
- Disagreements: [Where perspectives differed + our choice]

───────────────────────────────────────────────────────────────────
§ OPEN LOOPS
───────────────────────────────────────────────────────────────────

**Next Actions**:
- [ ] [Immediate next step - be specific]
- [ ] [Following step]

**Blockers**:
- [What's blocking + why]: [Waiting for X | Need to solve Y]

**Pending**:
- Code review: [PR #XXX awaiting review]
- Testing: [Need to test X scenario]
- Documentation: [Need to update README/docs]

**Questions to Resolve**:
- [ ] [Technical question needing answer]
- [ ] [Design decision pending]

───────────────────────────────────────────────────────────────────
§ TESTING & VALIDATION
───────────────────────────────────────────────────────────────────

**Test Status**:
- Passing: [X/Y tests pass]
- Failing: [Test names that fail + why]
- Coverage: [X% coverage | Not measured]

**Manual Testing Done**:
- [Scenario tested]: [Result]
- [Edge case checked]: [Outcome]

**Still Need to Test**:
- [ ] [Test case pending]
- [ ] [Integration scenario]

───────────────────────────────────────────────────────────────────
§ CONTEXT NOTES
───────────────────────────────────────────────────────────────────

**Key Insights**:
- [Important discovery from this session]
- [Gotcha/caveat to remember]

**Developer Notes**:
- Communication style: [Preferences for next session]
- Assumed knowledge: [What doesn't need re-explaining]
- Sensitive areas: [Code that's fragile, requires care]

**Links/References**:
- Documentation: [URLs to relevant docs]
- Related PRs/Issues: [GitHub links]
- Design docs: [Figma, diagrams, etc.]

═══════════════════════════════════════════════════════════════════
§ TRANSFER READY
═══════════════════════════════════════════════════════════════════
Review for accuracy before sharing. Check git state and file paths.

After generating, ask: "Before you transfer—are there any sections that need further detail or refinement?"


Generating the Artifact

Step 1: Gather Context

File Context:

# Get modified files
git status --short

# Get recent commits
git log --oneline -5

# Check current branch and tracking
git branch -vv

# Show uncommitted changes summary
git diff --stat
git diff --cached --stat

Environment Context:

# Check running processes
lsof -i -P -n | grep LISTEN

# Check environment
env | grep -E '(PATH|NODE_ENV|DATABASE|API|APP_)'

# Recent package changes (if applicable)
git log --oneline -10 package.json

Code Context:

  • Note which files have been actively edited
  • Identify key functions/classes modified
  • Track new files created vs existing files changed

Step 2: Generate Artifact

Fill out sections systematically:

  1. Mission/Status/Progress - What and why
  2. Code Context - What files/functions changed
  3. Git State - Branch, commits, staging area
  4. Environment State - Services, dependencies, env vars
  5. Technical Decisions - Choices made (include peer review input)
  6. Open Loops - Next actions, blockers
  7. Testing - What's validated, what's pending
  8. Context Notes - Insights, gotchas, references

Step 3: Present & Refine

  1. Present artifact in full
  2. Add: "§ TRANSFER READY—Review for accuracy before sharing."
  3. Ask: "Before you transfer, are there any sections that need further detail or refinement?"
  4. Human reviews, requests expansions if needed

Step 4: Transfer to New Session

Receiving agent prompt (optional prepend):

[DEV CONTEXT TRANSFER]

The following is a development context transfer from a previous Claude Code session.

After reading, provide a handshake confirmation:

"I've reviewed the dev context. Quick confirmation:
- Mission: [Echo back what we're building]
- Code State: [Active files and key changes]
- Git: [Branch + commit status]
- Next: [Immediate next action]
- Environment: [Critical services/dependencies]

Ready to [next action]. What's your priority?"

Integration with Peer Review Skills

When using Codex or Gemini for peer review during the development session:

Before Peer Review

  1. Generate a lightweight context artifact (just Code + Git + Technical Decisions sections)
  2. Use as input to peer review request
  3. Get Codex/Gemini perspective

After Peer Review

  1. Update § Technical Decisions with peer review findings:

- What Codex/Gemini recommended - Where they agreed - Where they disagreed - Which advice you followed (and why)

  1. Include in transfer artifact so receiving agent knows:

- What external validation was done - What technical debates were resolved - What alternatives were already considered

Example Integration

§ TECHNICAL DECISIONS

**Decisions Made**:
| Decision | Rationale | Alternatives Rejected | Tradeoff |
|----------|-----------|----------------------|----------|
| Use Redis for session storage | Sub-ms latency required, peer reviews validated | PostgreSQL (too slow), In-memory (no persistence) | Added infrastructure complexity |

**Peer Review Integration**:
- **Codex consulted**: Yes - Recommended Redis over Postgres for session store, flagged potential memory limits
- **Gemini consulted**: Yes - Agreed with Redis choice, suggested Redis Cluster for scaling
- **Agreements**: Both AIs validated Redis for performance needs
- **Disagreements**: Codex suggested 1GB limit, Gemini suggested 2GB - we chose 1.5GB as middle ground
- **Implementation notes**: Added eviction policy (allkeys-lru) based on Codex warning about memory pressure

Best Practices

Do:

  • Run git status and git diff --stat before generating
  • Include specific file paths with line numbers (file.py:123)
  • Note running services and their ports
  • Capture peer review insights in § Technical Decisions
  • Include error messages or test failures verbatim
  • Mark files as "in-progress" vs "completed"

Don't:

  • Include secrets, API keys, credentials (redact them)
  • Paste entire file contents (link to files with line ranges)
  • Assume receiving agent has access to same environment
  • Skip git state (critical for resuming work)
  • Forget to note running services (easy to miss)

Examples

See references/examples.md for:

  • Full-stack feature development handoff
  • Bug fix mid-investigation transfer
  • Refactoring session continuation
  • Code review preparation
  • Post-peer-review implementation
  • Emergency context capture

Validation

Use the Python validator to check artifact quality:

python context-continuity-code/scripts/validate_dev_transfer.py artifact.md

Checks for:

  • Required sections present
  • Git state completeness
  • File paths formatted correctly
  • No secrets leaked
  • Peer review integration (if applicable)

Design Principles

Development-First: Optimized for code handoffs, not general conversation Git-Aware: Git state is mandatory, not optional Tool State Required: Environment and services are core context Peer Review Integration: First-class support for Codex/Gemini consultation Single Mode: One format optimized for dev workflows (no Minimal/Full choice) Antifragile: Critical info first (Mission → Code → Git → Environment)


Differences from Base Context Continuity

FeatureBase SkillClaude Code Edition
TargetGeneral conversationsDevelopment work only
ModesMinimal + FullSingle optimized mode
Tool StateOptional [T] tagsMandatory §§ sections
Git ContextNot includedRequired
Code ContextNot includedCore feature
Peer ReviewNot mentionedIntegrated workflow
Length200-1000 words400-600 words
EnvironmentAny Claude instanceClaude Code only

Use this skill when resuming development work. Use base context-continuity for general conversations.

平台分布

Claude Code

28.29%
按下载量换算20

windsurf

24.67%
按下载量换算18

OpenCode

17.77%
按下载量换算13

Codex

13.36%
按下载量换算9

Antigravity

7.77%
按下载量换算6

Gemini CLI

3.54%
按下载量换算3

安全审计

暂无安全审计结果可展示。

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills