Token导航 LogoToken导航

Agent Skill · 研究检索

skill-update(技能更新)

skill-update 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。

skill-update 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,权限为只读;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 70,834。

总安装 198周安装 8GitHub Stars 11下载 62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
unknown

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:skill-update(技能更新)
来源仓库:https://github.com/parhumm/jaan-to
仓库路径:skills/skill-update
安装命令:
npx skills add https://github.com/parhumm/jaan-to --skill skill-update
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。

skills.shnpx skills
npx skills add https://github.com/parhumm/jaan-to --skill skill-update

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。

skill-update 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。适用于研究检索类任务,支持基于关键词或上下文进行信息聚合与过滤。

通过 npx skills add 命令从指定 GitHub 仓库安装,需确认权限范围和联网能力。建议结合原始 README 核验具体用法,注意维护状态及是否触发文件读写或命令执行。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/skill-update

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 https://github.com/parhumm/jaan-to --skill skill-update;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

skill-update 怎么安装?

在终端执行 npx skills add https://github.com/parhumm/jaan-to --skill skill-update。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

skill-update 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。

skill-update 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

skill-update 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

skill-update 用的人多吗?

公开数据是累计安装 198 次,近一周安装 8 次,下载 62 次,GitHub 11 星。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。

skill-update 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

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。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 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
skill-update

技能更新 · 本页

19811只读未明确有提醒

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

skill-update

Update existing jaan.to skills with specification compliance and documentation sync.

Context Files

  • docs/extending/create-skill.md - Skill specification (REQUIRED)
  • $JAAN_LEARN_DIR/jaan-to-skill-update.learn.md - Past lessons (loaded in Pre-Execution)
  • $JAAN_CONTEXT_DIR/config.md - Current skill catalog
  • ${CLAUDE_PLUGIN_ROOT}/docs/extending/language-protocol.md - Language resolution protocol

Input

Skill Name: $ARGUMENTS

The name of the skill to update (e.g., pm-prd-write or just prd-write).

If not provided, list available skills and ask which to update.


Pre-Execution Protocol

MANDATORY — Read and execute ALL steps in: ${CLAUDE_PLUGIN_ROOT}/docs/extending/pre-execution-protocol.md Skill name: skill-update Execute: Step 0 (Init Guard) → A (Load Lessons) → B (Resolve Template) → C (Offer Template Seeding)

Language Settings

Read and apply language protocol: ${CLAUDE_PLUGIN_ROOT}/docs/extending/language-protocol.md Override field for this skill: language_skill-update


PHASE 0: Git Branch Setup

Create feature branch for updates:

git checkout dev
git pull origin dev
git checkout -b update/{skill-name}

Confirm: "Created branch update/{name} from dev. All updates on this branch."


PHASE 1: Analysis (Read-Only)

Thinking Mode

ultrathink

Use extended reasoning for:

  • Analyzing existing skill structure
  • Planning updates carefully
  • Validating against specification
  • Ensuring backward compatibility

Questioning Guidelines

Ask up to 7 clarifying questions across Phase 1 if needed.

Skip questions when:

  • Information already provided in user input
  • Research answered the question
  • Context files contain the answer
  • Question is redundant with previous answer

Make questions smart:

  • Reference user's input: "You said '{X}' — does that mean...?"
  • Build on existing skill: "The current skill does Y — should that change?"
  • Probe specifics: "What should happen when Z?"

Step 1: Read Existing Skill

Read all files for the skill:

  • skills/{name}/SKILL.md - Current skill definition
  • $JAAN_LEARN_DIR/{name}.learn.md - Accumulated lessons
  • $JAAN_TEMPLATES_DIR/{name}.template.md - Output template (if exists)

Display current structure:

CURRENT SKILL: {name}
────────────────────
Command: /{name}
Name: {name}
Description: {description}

FILES
─────
□ SKILL.md ({line_count} lines)
□ LEARN.md ({lesson_count} lessons)
□ template.md ({exists/missing})

Step 2: Validate Against Specification

Check current skill against docs/extending/create-skill.md:

Frontmatter:

  • Has name matching directory
  • Has description with purpose and mapping
  • Has allowed-tools with valid patterns
  • Has argument-hint

Body:

  • Has H1 title matching skill name
  • Has tagline blockquote
  • Has ## Context Files
  • Has ## Input
  • Has # PHASE 1: Analysis
  • Has ## Step 0: Apply Past Lessons
  • Has # HARD STOP
  • Has # PHASE 2: Generation
  • Has ## Definition of Done

Trust:

  • Tool permissions are sandboxed

Show compliance status:

SPECIFICATION COMPLIANCE
────────────────────────
✓ Frontmatter: 4/4 fields
✗ Body: 8/9 sections (missing: Step 0)
✓ Trust: sandboxed

Step 2.1: v3.0.0 Compliance Check

Check the skill for v3.0.0 customization system compatibility:

Read and apply V3.1–V3.10 check patterns from: ${CLAUDE_PLUGIN_ROOT}/docs/extending/v3-compliance-reference.md (section: "skill-update: v3.0.0 Compliance Checks", subsections V3.1–V3.10)

Run each check (V3.1 through V3.10) against the skill. Then display:

v3.0.0 Compliance Summary

Display results:

v3.0.0 COMPLIANCE
─────────────────
V3.1 Frontmatter env vars:     ✓ / ✗
V3.2 Context paths:             ✓ / ✗
V3.3 Learning path:             ✓ / ✗
V3.4 Template path:             ✓ / ✗ / N/A
V3.5 Output path:               ✓ / ✗ / N/A
V3.6 Template variables:        ✓ / ✗ / N/A
V3.7 Tech integration:          ✓ / N/A

OUTPUT STRUCTURE COMPLIANCE
───────────────────────────
V3.8.1 ID generation:           ✓ / ✗ / N/A
V3.8.2 Folder structure:        ✓ / ✗ / N/A
V3.8.3 Index management:        ✓ / ✗ / N/A
V3.8.4 Executive Summary:       ✓ / ✗ / N/A

DESCRIPTION BUDGET
──────────────────
V3.9 Description budget:        ✓ / ✗

DISPLAY STRINGS
───────────────
V3.10 Display paths use vars:   ✓ / ✗

VERDICT: v3.0.0 Compliant / Needs Migration / Needs Output Migration / Needs Description Fix / Needs Display Fix

If any check fails (✗):

  • Add option [8] to Step 3: "Migrate to v3.0.0"
  • If V3.8 checks fail: Add option [9]: "Migrate output structure to ID-based folders"

Step 3: Ask Update Type

"What do you want to change?" [1] Add/modify questions (Phase 1) [2] Update quality checks (Phase 2) [3] Modify output format (template.md) [4] Add tool permissions [5] Incorporate LEARN.md lessons → SKILL.md [6] Fix specification compliance issues [7] Other (describe) [8] Migrate to v3.0.0 (if v3.0.0 compliance check failed) [9] Migrate output structure to ID-based folders (if V3.8 check failed) [10] Fix description budget (trim Auto-triggers/Maps-to lines, shorten description)

Step 4: Optional Web Research

For options [1], [2], [3], or [7], offer:

"Search for updated best practices? [y/n]"

If yes, use Task tool with Explore subagent:

Task prompt: "Research current best practices for {domain}:
1. Search '{domain} best practices {year}'
2. Search '{domain} checklist {year}'
Return: new practices, updated methodologies, changes since {skill_created_date}"

Step 5: Plan Changes

Based on selected option, plan specific changes:

Option 1 (Questions): Show current questions, propose additions Option 2 (Quality): Show current checks, propose updates Option 3 (Template): Show current template, propose modifications Option 4 (Tools): Show current permissions, propose additions Option 5 (LEARN→SKILL): Map lessons to skill sections:

LEARN.md SectionIncorporate Into
Better QuestionsPhase 1 Step 1 questions
Edge CasesPhase 2 quality checks
WorkflowProcess steps + Definition of Done
Common MistakesWarnings in relevant sections

Option 6 (Compliance): List missing sections, propose additions Option 7 (Other): Gather details, plan custom changes

Option 8 (Migrate to v3.0.0): Read and apply migration wizard from: ${CLAUDE_PLUGIN_ROOT}/docs/extending/v3-compliance-reference.md (section: "Migration Wizard (v2.x → v3.0.0)"). Offers 4 approaches: Auto-fix all, Interactive, Manual script, Guidance only.


HARD STOP - Human Review Check

Show diff preview:

PROPOSED CHANGES
────────────────
File: SKILL.md
───
- old line
+ new line
───

File: template.md (if applicable)
───
- old line
+ new line
───

COMPLIANCE AFTER UPDATE
───────────────────────
✓ Frontmatter: 4/4 fields
✓ Body: 9/9 sections
✓ Trust: sandboxed
"Apply these changes? [y/n/edit]"

Do NOT proceed to Phase 2 without explicit approval.


PHASE 2: Update (Write Phase)

Step 6: Update SKILL.md

Apply planned changes while preserving:

  • Two-phase workflow structure
  • HARD STOP section
  • Definition of Done section
  • Specification compliance

Step 7: Update template.md (if needed)

If output format changes requested:

  1. Update template structure
  2. Preserve required metadata section
  3. Update placeholders

Step 8: Update LEARN.md

If Option 5 selected (lessons incorporated):

  • Add workflow note: "Incorporated into SKILL.md on {date}"
  • Keep original lessons for reference

Otherwise, add any new workflow learnings:

  • "Updated {section} based on {reason}"

Step 9: Validate Updated Skill

Run full specification check:

  • YAML frontmatter complete
  • All required sections present
  • Two-phase workflow intact
  • HARD STOP section exists
  • Definition of Done present
  • Tool permissions sandboxed

If any check fails, fix before continuing.

Step 10: Preview All Changes

Show final versions of all modified files.

"Write these updates? [y/n]"

Step 10.5: Handle Output Structure Migration (If Option [9] Selected)

Read and apply output structure migration from: ${CLAUDE_PLUGIN_ROOT}/docs/extending/v3-compliance-reference.md (section: "Output Structure Migration (Step 10.5)")

Includes: migration plan display, HARD STOP approval, Step 5.5 insertion, output writing update, template update, and validation checklist.

Step 11: Write Updated Files

If approved:

  1. Write SKILL.md to skills/{name}/SKILL.md
  2. Write template.md to skills/{name}/template.md (if modified)
  3. Write LEARN.md to skills/{name}/LEARN.md

Confirm: "Skill files updated in skills/{name}/"

Step 12: Auto-Invoke Documentation Sync

Run /jaan-to:docs-update {name} to sync:

  • docs/skills/{role}/{name}.md

This ensures documentation stays in sync with skill changes.

Step 13: Commit to Branch

Stage and commit updated files:

bash scripts/prepare-skill-pr.sh
git add skills/{name}/ jaan-to/ docs/skills/{role}/{name}.md

prepare-skill-pr.sh regenerates + validates Codex skillpack artifacts and stages adapters/codex/skillpack/.

For full commit message template, read: ${CLAUDE_PLUGIN_ROOT}/docs/extending/git-pr-workflow.md (section: "Step 13: Commit to Branch") Commit message prefix: fix(skill): Update {name} skill


PHASE 3: Testing & PR

Step 14: User Testing

"Please test the updated skill in a new session. Here's a copy-paste ready example:" `` /{name} "{example_input_based_on_skill_purpose}" `` "Did it work correctly? [y/n]"

If issues:

  1. Help debug the problem
  2. Make fixes
  3. Commit fixes
  4. Repeat testing

Step 15: Create Pull Request

When user confirms working:

"Create pull request to merge to dev? [y/n]"

If yes, push and create PR. For full PR body template, read: ${CLAUDE_PLUGIN_ROOT}/docs/extending/git-pr-workflow.md (section: "Step 15: Create Pull Request")

git push -u origin update/{name}
gh pr create --base dev --title "fix(skill): Update {name} skill" --body "..."

PR summary must include:

Codex skillpack sync: ✅ generated via scripts/prepare-skill-pr.sh

Show PR URL to user.

If no:

"Branch update/{name} is ready. Merge manually when ready."

Step 16: Capture Feedback

"Any feedback on the skill update process? [y/n]"

If yes:

  • Run /jaan-to:learn-add skill-update "{feedback}"

v3.0.0 Migration Feedback (if Option 8 was used)

If skill was migrated to v3.0.0, capture migration-specific learnings:

Categories: migration approach effectiveness, missed patterns, template variable adoption, tech stack integration.

Auto-categorize: patterns to compliance checks, edge cases to Step 2.1, workflow to Migration Wizard.

Example:

/jaan-to:learn-add skill-update "Auto-fix missed pattern: `Read(jaan-to/docs/**)` in doc-generation skills."

Step 17: Auto-Invoke Roadmap Update

Run /jaan-to:pm-roadmap-update to sync the skill update with the roadmap.

This ensures the roadmap reflects the latest skill changes.


Skill Alignment

  • Two-phase workflow with HARD STOP for human approval
  • Single source of truth (no duplication)
  • Plugin-internal automation
  • Maintains human control over changes

Definition of Done

  • Existing skill files read and analyzed
  • Specification compliance validated
  • User-selected updates applied
  • Passes specification validation after update
  • Documentation synced via /jaan-to:docs-update
  • User tested and confirmed working
  • PR created (or branch ready for manual merge)
  • Roadmap synced via /jaan-to:pm-roadmap-update
  • User approved final result

平台分布

Codex

34.16%
按下载量换算21

Claude

28.78%
按下载量换算18

Cursor

19.33%
按下载量换算12

Gemini CLI

10.41%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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