菜单
Use when extracting structured company lists from public or authorized directories into JSON, CSV, CRM-ready lists, or research tables.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/dinglebear-ai/axon --skill company-directories命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Use Axon for RAG, web search, source indexing, extraction, memory, and grounded answers over indexed docs, repos, feeds, or web content.
Route web search, source indexing, extraction, RAG, jobs, and watch tasks through the local Axon CLI.
Use Axon site-scope source indexing to capture many pages from a site or scoped documentation section.
基于 SOC 职业分类
正在显示 SKILL.md
| name | company-directories |
| description | Use when extracting structured company lists from public or authorized directories into JSON, CSV, CRM-ready lists, or research tables. |
Use this to turn startup or company directories into structured lists.
Infer the directory, filters, result count, and output format from context. If the source is clear, proceed immediately.
Ask at most 1-3 concise questions only if blocked, such as the directory URL/name, required filters, or target result count.
Use axon map, axon scrape, and axon extract when listings are public and
static. Use separate browser automation for filters, pagination, infinite
scroll, or profile clicks only when the user confirms authorized access and
source terms permit automated extraction; feed discovered URLs/content back to
Axon.
Suggested sources include YC companies, Crunchbase, Product Hunt, G2 categories, or any custom directory URL.
For finished deliverables, run Axon commands in blocking mode: axon extract ... --wait true --json and axon <url> --scope site --wait true. Axon has no robots.txt enforcement flag — bound public-directory capture yourself with a tight start URL plus --max-pages, --max-depth, and --budget PATH=N, and honor the source's terms.
Stop at login walls, paywalls, CAPTCHAs, or contractual restrictions unless the user confirms authorized access and source terms allow automated extraction.
Capture fields that are visible:
Leave unavailable fields blank. Do not infer.
# Company Directory Export: [Source]
## Summary
[Filters, count extracted, limitations]
## Companies
[Table or link to JSON/CSV; each row includes sourceUrl, profileUrl, extractedAt, fieldsObserved, confidence, and limitations]
## Sources
[Directory pages and profiles used]
## Rerun Inputs
workflow: company-directories
directory: [source]
filters: [criteria]
max_results: [number]
output: [json/csv/markdown]
Use source, filters, extractedAt, totalResults, and companies[] with name, url, description, industry, stage, founded, location, teamSize, funding, tags, profileUrl, websiteUrl, sourceUrl, fieldsObserved, confidence, and limitations.