georank

GitHub

通过HTTP API操作和管理自托管GEOrank实例,涵盖登录认证、账号检查、数据提交与诊断、关键词扩展及管理员权限控制,严格遵循安全边界与审计流程。

skills/georank/SKILL.md yaojingang/GEORank

触发场景

需要登录或检查GEOrank账户状态 执行GEOrank数据读写或管理操作 运行GEO诊断或管理解决方案对话

安装

npx skills add yaojingang/GEORank --skill georank -g -y
更多选项

不安装直接使用

npx skills use yaojingang/GEORank@georank

指定 Agent (Claude Code)

npx skills add yaojingang/GEORank --skill georank -a claude-code -g -y

安装 repo 全部 skill

npx skills add yaojingang/GEORank --all -g -y

预览 repo 内 skill

npx skills add yaojingang/GEORank --list

SKILL.md

Frontmatter
{
    "name": "georank",
    "description": "Operate and manage a self-hosted GEOrank instance through its HTTP API. Use when asked to log in, inspect a GEOrank account, submit or manage companies, run or read GEO diagnostics, manage solution conversations, expand keywords, inspect usage, or perform authenticated GEOrank administrator work across users, content, experts, providers, modules, homepage releases, and settings. Login guidance defaults to an ordinary user and unlocks administrator actions only after \/api\/auth\/me returns role=admin. Do not use for general GEO strategy, article writing, SEO advice, or GEOrank code development that does not operate a running instance."
}

GEOrank Operator

Workflow

Resolve every relative path below from this Skill directory, independent of the caller's current workspace.

  1. Confirm the target GEOrank base URL. Use http://localhost:8000 only for a local instance; require HTTPS for remote instances.
  2. Resolve authentication:
    • Treat login as an ordinary-user flow and never ask the user to choose a role.
    • Run python3 scripts/georank_client.py login --account <account> when no session exists.
    • Run python3 scripts/georank_client.py whoami after login. Enable administrator branches only when the returned role is admin; treat enterprise as a regular authenticated user.
  3. Read references/user-capabilities.md for public and user-owned operations. Read references/admin-capabilities.md only for administrator requests.
  4. For reads, call the API directly with scripts/georank_client.py call GET <path>.
  5. For writes, run the same command without --execute first and show the preflight result. Execute only after the user's request clearly authorizes that exact change.
  6. Before any administrator mutation, read references/safety-policy.md and supply its required confirmation phrase. Never infer administrator authority from the wording of a request.
  7. Return an execution receipt containing the action, status, resource identifiers, request ID when present, and the next safe step. Remove passwords, tokens, API keys, and other secrets from all output.

Safety Boundary

  • Never pass passwords or tokens as command-line arguments. Login reads GEORANK_PASSWORD or uses a hidden terminal prompt; API calls read GEORANK_TOKEN or the private session file.
  • Never expose, copy, or persist a secret outside the session store managed by the client.
  • Treat every non-read call as a side effect. --execute is required.
  • Require APPLY_ADMIN_CHANGE for administrator writes and DELETE:<api-path-with-query> for deletions.
  • Stop when the detected role lacks permission, the target resource is ambiguous, or rollback information is unavailable for a high-impact change.

Output Contract

Return a compact Markdown or JSON receipt with:

  • detected access level: public, user, or admin
  • operation and target resource
  • dry-run or executed status
  • API status, resource IDs, and request ID when available
  • redacted change summary and next step
  • rollback guidance for administrator changes

Reference Map

  • references/user-capabilities.md: public and ordinary-user operations.
  • references/admin-capabilities.md: administrator operation groups and confirmation requirements.
  • references/safety-policy.md: authentication, secrets, side effects, polling, and rollback rules.
  • scripts/georank_client.py: deterministic login, session, role detection, preflight, and API execution.
  • evals/trigger_cases.json: routing positives, negatives, and near neighbors.
  • examples/request_payloads.json: file-backed fixture examples listed by input_files.
  • reports/output-risk-profile.md: likely output failures and repair checks.
  • reports/output_quality_scorecard.md: current output and verification evidence.
  • reports/security_trust_report.md: generated trust report after validation.

版本历史

  • 21d1ab9 当前 2026-07-19 08:50

元信息

文件数
0
版本
1df59ad
Hash
eec11746
收录时间
2026-07-19 08:50

首页 - Wiki
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-08-06 01:02
浙ICP备14020137号-1 $访客地图$