Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问许可证需确认审计提醒

draxarpdraxarp 搜索

Agent Skill

draxarp 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

1,672

周安装

67

GitHub Stars

公开资料未说明

下载量

541
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:draxarp(draxarp 搜索)
来源仓库:https://github.com/jorgemuza/orbit
仓库路径:skills/draxarp
安装命令:
npx skills add https://github.com/jorgemuza/orbit --skill draxarp
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/jorgemuza/orbit --skill draxarp

简介

LLM 辅助开发的第二大脑模块,管理九类资源。

  • 涵盖项目、任务、记忆、规格等全生命周期管理。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。
  • 提供 orbit CLI 别名 dx 进行高效资源操作。
  • 需初始化虚拟环境并安装依赖方可启用全部功能。
  • draxarp 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Draxarp Intelligence — orbit CLI

Alias: dx

The orbit draxarp (or orbit dx) command manages the Draxarp Development Intelligence module — a second brain for LLM-assisted development. It covers nine resource types:

Resources

ResourceAliasDescription
projectprojIntelligence projects — group tasks, specs, memories, and docs
taskDevelopment tasks with status, priority, activity logs, handoffs, focus, and dependencies
memorymemPersistent knowledge — patterns, bugs, decisions, domain context
specSpecifications with review workflow (draft → review → approved/rejected)
docTechnical documentation with slug-based addressing and wikilink support
sprintSprint planning with velocity tracking and AI content suggestions
capturecapContext captures — git commits, PRs, CI events, deployments
graphkgKnowledge graph — nodes, edges, impact analysis, subgraph, path finding
decomposeAI-powered task decomposition from descriptions

Tracking (via orbit tracking)

CommandDescription
tracking event pushPush workflow events (skill completions, Jira transitions, agent invocations)
tracking tkm pushPush token usage data from local TKM database

Tracking admin & diagnostics (via orbit dx tracking, alias trk)

CommandDescription
tool-versions / tvShow/set latest expected tool versions (orbit, claude-code, paybook-workflow)
doctor <user>Per-user pipeline diagnosis with verdict and actionable hints
pipelinesCluster-wide pipeline health + list of "stuck users"
tailTail ingest events across both pipelines; supports --follow
api-logsQuery the api_logs table (when request logging is enabled)
hooksInspect local ~/.claude/settings.json ingest hooks (offline)
hooks installIdempotently wire a Stop hook that runs tracking tkm push after every Claude Code response
hooks uninstallRemove tracking tkm push Stop hooks while preserving unrelated hooks
# Diagnose why events stopped flowing for a user
orbit dx trk doctor manuel.toala@paybook.me

# Cluster health + stuck users
orbit dx trk pipelines

# Real-time stream for a single user
orbit dx trk tail --user manuel.toala@paybook.me --follow

# Verify the user's local hooks still reference the ingest endpoints
orbit dx trk hooks

# First-run setup: wire the token-usage Stop hook (idempotent)
orbit dx trk hooks install --bake-profile paybook
Without hooks install, token-usage events only flow when a skill that embeds tracking tkm push (e.g. paybook-workflow:commit, paybook-workflow:prime) is explicitly used — producing false TOKEN_USAGE_BROKEN verdicts in doctor.

Quick Reference

Projects

orbit dx proj ls                              # List projects
orbit dx proj view <id-or-slug>               # View project details
orbit dx proj create --name "My Project"      # Create project
orbit dx proj delete <id>                     # Delete project
Slugs everywhere: All --project flags and resource IDs accept either UUIDs or slugs. Use slugs for readability: --project draxarp-portal instead of --project 019cf38c-...

Tasks — Core CRUD

orbit dx task ls --project <id>               # List tasks for a project
orbit dx task ls --status in_progress         # Filter by status
orbit dx task ls --type epic --project <id>   # List only epics
orbit dx task ls --parent <epic-id>           # List children of an epic
orbit dx task view <id>                       # View task details + last 5 activity logs
orbit dx task create --title "Fix bug" --project <id> --priority high
orbit dx task create --title "Wire frontend" --project <id> --depends-on <blocker-id>
orbit dx task create --title "Auth" --project <id> --type epic      # Create an epic
orbit dx task create --title "JWT" --project <id> --type story --parent <epic-id>  # Story under epic
orbit dx task update <id> --status testing --activity "Running PHPUnit"
orbit dx task assign <id> --assignee claude   # Assign task
orbit dx task start <id>                      # Set status to in_progress
orbit dx task complete <id>                   # Mark completed (with guard checks)
orbit dx task delete <id>                     # Delete task
orbit dx task epic <id>                       # Show epic progress (% complete, children)

Task types: task (default), epic, story

Tasks — Focus & Next (session management)

orbit dx task focus <id>                      # One command: assign + start + set focus
orbit dx task focus <id> --agent claude       # Specify agent (default: claude)
orbit dx task current                         # Show currently focused task
orbit dx task current --agent claude          # For a specific agent
orbit dx task next --project <id>             # Next available task by priority (skips epics & blocked)

Tasks — Activity Logs

orbit dx task log <id> "Implemented API endpoints" --tag progress
orbit dx task log <id> "Blocked on DB schema" --tag blocker --agent claude
orbit dx task log <id> "Chose approach A over B" --tag decision
orbit dx task logs <id>                       # List all logs (newest first)
orbit dx task logs <id> --tag blocker         # Filter by tag
orbit dx task logs <id> --limit 10            # Limit results
orbit dx task logs <id> -o json               # JSON output

Activity log tags: progress, blocker, decision, hypothesis, tried, result, handoff, note

Tasks — Structured Handoffs & Context

# Record a handoff (session continuity)
orbit dx task handoff <id> \
  --done "Implemented X" --done "Added tests for Y" \
  --remaining "Wire up frontend" --remaining "Add error handling" \
  --decision "Used approach A because of constraint B" \
  --uncertain "Not sure if Z is the right abstraction" \
  --files "backend/Services/Foo.php,frontend/Show.tsx" \
  --agent claude

# Get AI-optimized context for resuming work
orbit dx task context <id>                    # Human-readable
orbit dx task context <id> -o json            # Structured JSON

Tasks — Dependencies

orbit dx task dep <id> --on <blocking-task-id>    # Add dependency
orbit dx task deps <id>                            # Show dependency tree (blocks/blocked-by)
orbit dx task deps <id> -o json                    # JSON output

Task statuses: pending, in_progress, testing, completed, blocked, cancelled Task priorities: low, medium, high, critical

Memories

orbit dx mem ls --project <id>                # List memories
orbit dx mem ls --type architecture           # Filter by type
orbit dx mem view <id>                        # View memory content
orbit dx mem create --title "Pattern: Repo" --content "..." --project <id> --type patterns
orbit dx mem archive <id>                     # Archive memory

Memory types: architecture, patterns, bugs, domain, decisions, preferences, context

Specs

orbit dx spec ls --project <id>               # List specs
orbit dx spec view <id>                       # View spec content
orbit dx spec create --title "Auth Spec" --content "..." --project <id>
orbit dx spec submit <id>                     # Submit for review
orbit dx spec approve <id>                    # Approve spec
orbit dx spec reject <id>                     # Reject spec
orbit dx spec delete <id>                     # Delete spec

Spec statuses: draft, in_review, approved, rejected

Docs

orbit dx doc ls                               # List all docs
orbit dx doc ls --category architecture       # Filter by category
orbit dx doc toc                              # Table of contents (no content)
orbit dx doc view <id>                        # View doc by ID
orbit dx doc view architecture/overview       # View by slug (auto-detected)
orbit dx doc create --title "API Guide" --category api --slug api/guide
orbit dx doc sync docs.json                   # Batch sync from JSON file
orbit dx doc publish <id>                     # Publish draft doc
orbit dx doc archive <id>                     # Archive doc
orbit dx doc delete <id>                      # Delete doc

Doc categories: architecture, components, flows, security, api, infrastructure, general Doc statuses: draft, published, archived

Slug auto-detection: doc view automatically detects whether the argument is a UUID or slug — no --slug flag needed (though it's still supported for explicit override).

Batch sync: doc sync accepts a JSON file with an array of doc objects:

[{"slug": "api/guide", "title": "API Guide", "content": "...", "category": "api"}]

Existing docs (matched by slug) are updated; new slugs are created.

Sprints

orbit dx sprint ls --project <id>             # List sprints
orbit dx sprint view <id>                     # View sprint details
orbit dx sprint create --project <id> --name "Sprint 1" --starts-at 2026-03-16 --ends-at 2026-03-30
orbit dx sprint create --project <id> --name "Sprint 2" --starts-at 2026-03-30 --ends-at 2026-04-13 --goal "Ship auth" --velocity-target 20
orbit dx sprint start <id>                    # Start a sprint
orbit dx sprint complete <id>                 # Complete a sprint
orbit dx sprint suggest <id>                  # AI-powered sprint content suggestions
orbit dx sprint velocity --project <id>       # Velocity history (last 5 sprints)
orbit dx sprint velocity --project <id> --count 10  # More history

Context Captures

orbit dx capture ls --project <id>            # List captures for a project
orbit dx capture view <id>                    # View capture details + payload
orbit dx capture webhook --project <id> --source git_commit --payload '{"sha":"abc","message":"fix"}'
orbit dx capture persist <id>                 # Persist capture as a memory

Context sources: git_commit, git_pr, ci_pipeline, deployment, code_review, manual

Knowledge Graph

# Nodes
orbit dx graph nodes --project <id>           # List nodes
orbit dx graph nodes --type module            # Filter by type
orbit dx graph node <id>                      # View node details
orbit dx graph create-node --project <id> --type module --name "AuthService"
orbit dx graph delete-node <id>               # Delete a node

# Edges
orbit dx graph create-edge --from <id> --to <id> --relationship depends_on
orbit dx graph delete-edge <id>               # Delete an edge

# Analysis
orbit dx graph impact <node-id>               # Impact analysis (depth 3)
orbit dx graph impact <node-id> --depth 5     # Deeper analysis
orbit dx graph subgraph <node-id>             # Local subgraph (depth 2)
orbit dx graph paths --from <id> --to <id>    # Find paths between nodes
orbit dx graph paths --from <id> --to <id> --max-depth 8

Task Decomposition

orbit dx decompose create --project <id> --description "Build user auth with JWT"
orbit dx decompose create --project <id> --description "..." --spec <spec-id>  # With spec context
orbit dx decompose ls --project <id>          # List decompositions
orbit dx decompose view <id>                  # View decomposition + options
orbit dx decompose accept <id> --indexes 0,1,2  # Accept and create tasks from selected options
orbit dx decompose reject <id>                # Reject decomposition

Typical Agent Workflow

# 1. Orient
orbit dx context -p draxarp --project <id>

# 2. Pick up or create task
orbit dx task next --project <id> -p draxarp     # Or create new
orbit dx task focus <task-id> -p draxarp          # One command: assign + start + focus

# 3. Work (hooks auto-log progress)
# ... edit files, run tests ...

# 4. Log key decisions/blockers manually
orbit dx task log <id> "Chose X because Y" --tag decision -p draxarp
orbit dx task log <id> "Blocked on Z" --tag blocker -p draxarp

# 5. Handoff before session ends
orbit dx task handoff <id> --done "API done" --remaining "Frontend" --agent claude -p draxarp

# 6. Complete
orbit dx task complete <id> -p draxarp

# 7. Next session picks up with context
orbit dx task context <id> -p draxarp

Configuration

Add to ~/.config/orbit/config.yaml:

profiles:
  - name: my-project
    services:
      - name: draxarp
        type: draxarp
        base_url: https://your-draxarp-instance.com
        auth:
          method: token
          token: "your-api-token"
        headers:                              # Optional custom headers
          X-Tenant: my-tenant-slug            # Required for tenant-scoped access

For multi-tenant setups, the X-Tenant header scopes all API responses to that tenant's workspaces and projects. Tokens can be either platform admin (Sanctum) or tenant user (PAT) tokens.

Profile Selection

Use -p <profile> to target a specific profile:

orbit dx proj ls -p draxarp
orbit dx task ls -p draxarp --project <id>

JSON Output

All commands support -o json for machine-readable output:

orbit dx proj ls -o json
orbit dx mem view <id> -o json
orbit dx doc toc -o json
orbit dx task context <id> -o json
orbit dx sprint ls --project <id> -o json
orbit dx graph nodes --project <id> -o json

API Details

The orbit CLI communicates with the Draxarp API at:

  • Intelligence resources: /api/admin/v1/intelligence/* (tasks, memories, specs, docs, sprints, graph, captures, decompose)
  • Platform resources: /api/admin/v1/* (projects, workspaces)
  • Auth: Authorization: Bearer <token> (platform-api or tenant-api Sanctum guard)
  • Tenant scoping: X-Tenant: <slug> header (required for tenant-bound tokens)
  • Pagination: {"success": true, "data": [...], "meta": {"current_page", "last_page", "per_page", "total"}}
  • Single resource: {"success": true, "data": {...}}

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

36.21%
按下载量换算196

Claude

28.33%
按下载量换算153

Cursor

17.55%
按下载量换算95

Gemini CLI

9.78%
按下载量换算53

安全审计

Gen Agent Trust Hub

通过

Socket

可疑

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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