Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问许可证需确认审计异常

slipboxslipbox 搜索

Agent Skill

slipbox 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。

总安装

212

周安装

9

GitHub Stars

26

下载量

74
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:slipbox(slipbox 搜索)
来源仓库:https://github.com/randroids-dojo/skills
仓库路径:skills/slipbox
安装命令:
npx skills add https://github.com/randroids-dojo/skills --skill slipbox
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/randroids-dojo/skills --skill slipbox

简介

用于快速查找、检索和筛选相关信息,支持关键词和任务场景定位。

  • 适合在知识库、文档或代码库中快速获取候选结果时使用。
  • 可结合来源仓库和原始 README 进一步核验具体用法和功能。
  • 安装前应确认权限范围和维护状态,避免触发不必要的联网或文件操作。
  • 建议在使用前先了解是否会执行命令或访问外部资源。

SKILL.md

SlipBox Skill

IMPORTANT: Before doing anything else, run the setup check below. Do not skip this step.

Setup Check

Run this command immediately upon skill invocation:

echo "SLIPBOX_API_KEY: ${SLIPBOX_API_KEY:+${SLIPBOX_API_KEY:0:6}…(set)}" | sed 's/^SLIPBOX_API_KEY: $/SLIPBOX_API_KEY: (MISSING)/'
echo "SLIPBOX_URL: ${SLIPBOX_URL:-(MISSING)}"
echo "SLIPBOX_PRIVATEBOX_REPO: ${SLIPBOX_PRIVATEBOX_REPO:-(MISSING)}"

If any show (MISSING):

  • STOP IMMEDIATELY. Do not attempt any further action.
  • Do not guess values, use defaults, search for shell config files, or attempt fallbacks of any kind.
  • Tell the user exactly which variables are missing and that they must set them in ~/.zshrc (or ~/.zprofile) and re-source their shell before trying again.
  • End your response there and wait for the user to fix the issue.

Once env vars are confirmed, verify the service is reachable:

curl -sL "$SLIPBOX_URL/api/health"
# {"status":"ok"}

If the health check fails or returns anything other than {"status":"ok"}:

  • STOP IMMEDIATELY. Do not attempt any further action.
  • Report the response to the user and tell them the service is unavailable.
  • End your response there and wait for the user.

API Error Handling

If any API call returns an error response (any JSON with an "error" field, or a non-2xx HTTP status):

  • STOP IMMEDIATELY. Do not attempt any further action.
  • Do not try to write notes directly to PrivateBox or any other fallback.
  • Do not retry with different parameters or modified requests.
  • Report the exact error response to the user and wait for them to resolve it.

About

Interact with the SlipBox semantic knowledge engine and browse your PrivateBox notes.

SlipBox service: $SLIPBOX_URL PrivateBox repo: $SLIPBOX_PRIVATEBOX_REPO

Configuration

Required environment variables (set in shell):

SLIPBOX_API_KEY=<shared-secret>          # Bearer token for API auth
SLIPBOX_URL=https://slip-box-rho.vercel.app  # SlipBox service base URL
SLIPBOX_PRIVATEBOX_REPO=Randroids-Dojo/PrivateBox   # GitHub repo for notes (owner/repo)

All other configuration (OpenAI, GitHub, PrivateBox) lives on the deployed Vercel service.


Quick Reference

All API calls require: Authorization: Bearer $SLIPBOX_API_KEY

# Health check
curl -sL "$SLIPBOX_URL/api/health"

# Add a note
curl -sL -X POST "$SLIPBOX_URL/api/add-note" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"content": "Atomic idea goes here."}'

# Add a typed note (type: "meta" or "hypothesis")
curl -sL -X POST "$SLIPBOX_URL/api/add-note" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"content": "## Cluster: ...", "type": "meta"}'

# Re-link all notes (recompute similarity links)
curl -sL -X POST "$SLIPBOX_URL/api/link-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

# Cluster notes into thematic groups
curl -sL -X POST "$SLIPBOX_URL/api/cluster-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"k": 5}'

# Detect conceptual tensions (contradictions within clusters)
curl -sL -X POST "$SLIPBOX_URL/api/tension-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

# Fetch theme data (clusters + note content + tensions) for agent synthesis
curl -sL "$SLIPBOX_URL/api/theme-data" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

API Reference

POST /api/add-note

Capture an atomic idea. SlipBox embeds it, links it to similar notes, and commits it to PrivateBox.

Optional type field sets a semantic type in the note's frontmatter. Valid values: "meta" (AI-generated cluster summary) or "hypothesis" (AI-generated research hypothesis). Omit for regular atomic notes.

# Regular note
curl -sL -X POST "$SLIPBOX_URL/api/add-note" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "The Zettelkasten method treats each note as a discrete, reusable idea."
  }'

# Meta-note (cluster synthesis)
curl -sL -X POST "$SLIPBOX_URL/api/add-note" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "## Cluster: Agentic Systems\n\nNotes in this cluster explore...",
    "type": "meta"
  }'

Response:

{
  "noteId": "20260222T153045-a1b2c3d4",
  "type": "meta",
  "linkedNotes": [
    {"noteId": "20260110T091200-b2c3d4e5", "similarity": 0.91},
    {"noteId": "20260115T143000-c3d4e5f6", "similarity": 0.85}
  ]
}

type is null in the response for regular notes.

POST /api/link-pass

Recompute semantic similarity links across all notes. Run after adding many notes in bulk.

curl -sL -X POST "$SLIPBOX_URL/api/link-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

Response:

{"message": "Link pass complete", "notesProcessed": 42, "totalLinks": 156}

POST /api/cluster-pass

Run k-means clustering on note embeddings. Omit k to auto-select cluster count.

curl -sL -X POST "$SLIPBOX_URL/api/cluster-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"k": 5}'

Response:

{
  "message": "Cluster pass complete",
  "noteCount": 42,
  "clusterCount": 5,
  "clusters": [{"id": "cluster-0", "size": 9, "noteIds": ["20260222T153045-a1b2c3d4", ...]}, ...]
}

POST /api/tension-pass

Detect conceptual tensions — notes with contradictory content that cluster near each other.

curl -sL -X POST "$SLIPBOX_URL/api/tension-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

Response:

{
  "message": "Tension pass complete",
  "noteCount": 42,
  "clusterCount": 5,
  "tensionCount": 8,
  "tensions": [{"id": "tension-0", "noteA": "...", "noteB": "...", "similarity": 0.68, "clusterId": "cluster-0"}, ...]
}

GET /api/theme-data

Returns clusters with full note content and tensions for local LLM agent synthesis. No embeddings — only human-readable data. Use this to read your knowledge graph and synthesize meta-notes per cluster, then POST them back via /api/add-note.

Requires a current clusters index (run cluster-pass first).

curl -sL "$SLIPBOX_URL/api/theme-data" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

Response:

{
  "clusters": [
    {
      "id": "cluster-0",
      "noteIds": ["20260222T153045-a1b2c3d4", "20260110T091200-b2c3d4e5"],
      "notes": {
        "20260222T153045-a1b2c3d4": {"title": "Optional title", "body": "Atomic idea content."},
        "20260110T091200-b2c3d4e5": {"body": "Another idea."}
      }
    }
  ],
  "tensions": [
    {"id": "tension-0", "noteA": "20260222T153045-a1b2c3d4", "noteB": "20260110T091200-b2c3d4e5", "similarity": 0.65, "clusterId": "cluster-0"}
  ],
  "clusterCount": 1,
  "noteCount": 2,
  "tensionCount": 1,
  "computedAt": "2026-02-23T01:33:00.000Z"
}

If no clusters exist yet, returns {"message": "No clusters found. Run cluster-pass first.", "clusters": [],...}.


Note Format

Notes in PrivateBox are Markdown files with YAML frontmatter:

---
id: 20260222T153045-a1b2c3d4
title: "Optional title"
type: meta
tags: ["tag1", "tag2"]
source: "URL or origin"
created: 2026-02-22T15:30:45.000Z
updated: 2026-02-22T15:30:45.000Z
links:
  - target: 20260110T091200-b2c3d4e5
    similarity: 0.91
  - target: 20260115T143000-c3d4e5f6
    similarity: 0.85
---

Atomic idea content in Markdown.

type is omitted for regular notes. Valid values: meta, hypothesis.

Note ID format: YYYYMMDDTHHMMSS-<8hex> (timestamp + content hash)

Index files (in index/ directory of PrivateBox):

  • index/embeddings.json — noteId → vector + model + timestamp
  • index/backlinks.json — noteId → array of linking notes
  • index/clusters.json — thematic groups of notes
  • index/tensions.json — pairs of contradictory notes

Reading Notes from PrivateBox

Use the gh CLI to read notes without needing direct GitHub API calls.

# List all notes
gh api "repos/$SLIPBOX_PRIVATEBOX_REPO/contents/notes" \
  --jq '.[].name'

# Read a specific note by ID
gh api "repos/$SLIPBOX_PRIVATEBOX_REPO/contents/notes/20260222T153045-a1b2c3d4.md" \
  --jq '.content' | base64 -d

# Read the backlinks index
gh api "repos/$SLIPBOX_PRIVATEBOX_REPO/contents/index/backlinks.json" \
  --jq '.content' | base64 -d | jq '.'

# Read the clusters index
gh api "repos/$SLIPBOX_PRIVATEBOX_REPO/contents/index/clusters.json" \
  --jq '.content' | base64 -d | jq '.'

# Read the tensions index
gh api "repos/$SLIPBOX_PRIVATEBOX_REPO/contents/index/tensions.json" \
  --jq '.content' | base64 -d | jq '.'

Search notes by content

# Search PrivateBox notes for a keyword (uses GitHub code search)
gh api "search/code?q=<keyword>+repo:$SLIPBOX_PRIVATEBOX_REPO+path:notes" \
  --jq '.items[].path'

# Or clone locally for fast full-text search
gh repo clone "$SLIPBOX_PRIVATEBOX_REPO" /tmp/privatebox
grep -r "keyword" /tmp/privatebox/notes/ --include="*.md" -l

Find notes by tag

gh api "search/code?q=tags+keyword+repo:$SLIPBOX_PRIVATEBOX_REPO+path:notes" \
  --jq '.items[].path'

Workflows

Capture an idea

  1. Write the atomic idea as a single focused thought.
  2. POST to /api/add-note with the content.
  3. Note the returned noteId and linkedNotes to see what concepts it connects to.
  4. Optionally run /api/link-pass afterward if adding many notes in bulk.

Browse and explore

  1. List notes with gh api "repos/$SLIPBOX_PRIVATEBOX_REPO/contents/notes".
  2. Read individual notes by ID.
  3. Follow links in frontmatter to explore connected ideas.
  4. Use the clusters index to browse thematic groups.
  5. Use the tensions index to identify areas of conceptual conflict worth investigating.

Synthesize theme meta-notes

After running a full analysis cycle, use GET /api/theme-data to read every cluster with its note contents, then write one atomic synthesis note per cluster and POST each back via /api/add-note.

# 1. Fetch theme data
curl -sL "$SLIPBOX_URL/api/theme-data" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

# 2. For each cluster, synthesize a meta-note and POST it back with type "meta"
curl -sL -X POST "$SLIPBOX_URL/api/add-note" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "## Cluster: Agentic Systems\n\nNotes in this cluster explore...",
    "type": "meta"
  }'

The type: meta field is written to the note's frontmatter and can be used by future passes to filter AI-generated notes from atomic notes.

Run a full analysis cycle

After adding a batch of notes or to refresh the graph:

# Step 1: Recompute links
curl -sL -X POST "$SLIPBOX_URL/api/link-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

# Step 2: Recluster
curl -sL -X POST "$SLIPBOX_URL/api/cluster-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

# Step 3: Detect tensions
curl -sL -X POST "$SLIPBOX_URL/api/tension-pass" \
  -H "Authorization: Bearer $SLIPBOX_API_KEY"

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

37.84%
按下载量换算28

Claude

29.57%
按下载量换算22

Cursor

19.31%
按下载量换算14

Gemini CLI

10.6%
按下载量换算8

安全审计

Gen Agent Trust Hub

未通过

Socket

可疑

Snyk

可疑

权限和风险

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills