Overview

Find and extract the portfolio company list from an investment fund's website. The result is saved locally as a JSON file and returned to the caller.

Step 1 — Normalize the target URL

Take the input and produce a clean root URL:

• If it starts with http:// or https://, use as-is.
• If it's a bare domain (e.g., sequoiacap.com), prepend https://.
• Strip any path — use the root only (e.g., https://sequoiacap.com).

Also extract the bare domain (e.g., sequoiacap.com) for the output filename.

Step 2 — Discover the portfolio page

Try fetching these paths in order, stopping at the first one that returns meaningful company data:

1. [root-url]/portfolio
2. [root-url]/companies
3. [root-url]/investments
4. [root-url]/portfolio-companies
5. [root-url]/our-portfolio
6. [root-url]/founders

For each fetch, use this prompt: "List every company name mentioned on this page. Also return the page title so I can identify which fund this is."

If none of the above return a clear company list, fetch the homepage and look for any navigation links that suggest a portfolio or companies section. Then follow those links.

Step 3 — Fallback: web search

If WebFetch fails to find a usable portfolio page (e.g., JavaScript-heavy SPA, login wall, empty results), run a WebSearch:

[fund name] portfolio companies site:[domain]

Or if the fund name is unknown:

[domain] investment fund portfolio companies list

Use the search results to either find the direct portfolio URL to retry with WebFetch, or extract company names directly from search snippets.

Step 4 — Extract company names

From the fetched content, extract a clean list of portfolio company names:

• Include only company/startup names — not fund names, investor names, or team members
• Remove duplicates
• Normalize capitalization (use the company's own capitalization where visible)
• Do not include descriptions, sectors, or URLs — names only at this stage
• If the page is paginated or has "Load more", note how many companies were retrieved vs. total shown

Aim for completeness — capture every visible company name on the page.

Step 5 — Save the portfolio record

Write the result to a local audit file:

mkdir -p ~/.carta-crm/fund-portfolios
cat > ~/.carta-crm/fund-portfolios/[domain].json << 'ENDJSON'
{
  "fund": "[fund name]",
  "website": "[root url]",
  "portfolio_page": "[url used to retrieve data]",
  "retrieved_at": "[today's date as YYYY-MM-DD]",
  "company_count": [N],
  "companies": [
    "Company A",
    "Company B",
    "Company C"
  ]
}
ENDJSON

Confirm the file was written with echo $? (should be 0).

Step 6 — Return the result

Return the full JSON record to the caller. Then summarize:

"Found N portfolio companies for [Fund Name] from [portfolio_page]. Saved to ~/.carta-crm/fund-portfolios/[domain].json."

If the list appears incomplete (e.g., page was paginated, or only logos were shown without text names), add a note:

"Note: page may be incomplete — only N companies were visible as text. The fund may have more investments not listed."