network-finder
Skill14 starsChanged 9 months agoMITView on GitHub
**Network & Referral Finder**: Searches the user's LinkedIn connections and a referral spreadsheet to find potential contacts at a target company. Use this skill whenever the user wants to find connections at a company, look for referrals, check if they know anyone at a company, or find a warm intro path. Trigger on mentions of "referral", "connection", "know anyone at", "warm intro", "network", "who do I know at", or any company name combined with intent to find contacts there.
In this repository3 other files it gives its agents.
- job-analyzerSkill
- job-crawlerSkill
- job-form-fillerSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
24 rules. About 0 are common, 19 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈19
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
Before starting the workflow, check if config.yaml exists in the parent directory (one level up from this skill folder).
- Distinctive: line 17
Ask for the required fields using AskUserQuestion:
- Very specific: line 22
Write the responses to config.yaml in the parent directory
- Distinctive: line 26
Read it silently and proceed.
- Very specific: line 32
Read the URL from config.yaml → network.referral_sheet_url.
- Distinctive: line 37
Use browser tools to open the spreadsheet, then search for the target company.
- Distinctive: line 41
If the dialog says "no entries matching," the company isn't in the sheet
- Distinctive: line 42
If found, close the dialog and read the matching row for referrer contact info and details
- Very specific: line 44
Caching: If a local CSV copy exists at the workspace path and is less than 7 days old, search that first.
- Distinctive: line 48
Search the user's LinkedIn network for people currently at the target company.
- Very specific: line 50
Profile: Read from config.yaml → network.linkedin_profile_url
- Distinctive: line 56
Read the search results to find connections at the target company
- Distinctive: line 58
If no results with the exact company name, try common variations (e.g., "Block" vs "Block, Inc." vs "Square")
- Distinctive: line 70
If a referral sheet is configured, search it first — these are the highest-value contacts because they're explicitly offering referrals.
- Distinctive: line 78
If the company isn't found, note that clearly — "No referral found for [company] in your spreadsheet."
- Distinctive: line 80
If no referral sheet is configured, skip this step and note: "No referral spreadsheet configured — searching LinkedIn only."
- Distinctive: line 88
If there are results, present the top matches (up to 5).
- Distinctive: line 98
If found: Referrer name, contact info, any special instructions, date posted
- Distinctive: line 106
If referral found: "Reach out to [name] via [channel] — they're actively offering referrals"
- Distinctive: line 107
If LinkedIn connection only: "You're connected to [name] — consider reaching out for an intro"
- Distinctive: line 108
If neither: "No warm path found.
- Distinctive: line 112
Privacy: Don't screenshot or save LinkedIn profile data beyond names and titles.
- Distinctive: line 113
Referral sheet etiquette: If the spreadsheet has notes about referral conditions (e.g., can't refer if you already applied independently), …
- Distinctive: line 115
Freshness matters: If a referral sheet entry is months old, flag that the referrer may have moved on.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Network & Referral Finder
You help the user find warm intro paths to target companies by searching two sources: their LinkedIn connections and an optional referral/alumni spreadsheet.
First-Run Setup
Before starting the workflow, check if config.yaml exists in the parent directory (one level up from this skill folder). Read the user and network sections.
If config.yaml does NOT exist or the network.linkedin_profile_url field is empty:
- Tell the user: "This is your first time using network-finder. I need to set up your profile — this only happens once."
- Ask for the required fields using AskUserQuestion:
- Their LinkedIn profile URL (required — this is the core data source)
- Whether they have a referral/alumni spreadsheet (Google Sheet or other). If yes:
- The spreadsheet URL
- The column structure (e.g., "Column A = Name, Column B = Company, Column C = Notes")
- Write the responses to
config.yamlin the parent directory - Confirm setup is complete, then proceed
If config.yaml exists and has the required fields: Read it silently and proceed.
Data Sources
1. Referral Spreadsheet (Optional)
Read the URL from config.yaml → network.referral_sheet_url. If this field is blank, skip this source entirely.
The spreadsheet contains contacts who have offered referrals at their companies. Read the column structure from config.yaml → network.referral_sheet_structure.
Accessing the sheet: Use browser tools to open the spreadsheet, then search for the target company. The most reliable approach:
- Navigate to the spreadsheet URL
- Open Find and Replace via Edit menu > Find and replace (do NOT use Ctrl+F — it can get intercepted by cell edit mode in Google Sheets)
- Type the company name in the Find field and click "Find"
- If the dialog says "no entries matching," the company isn't in the sheet
- If found, close the dialog and read the matching row for referrer contact info and details
Caching: If a local CSV copy exists at the workspace path and is less than 7 days old, search that first. Otherwise, access the live sheet via browser. To create a cache, export via File > Download > CSV in the browser.
2. LinkedIn Connections
Search the user's LinkedIn network for people currently at the target company.
- Profile: Read from
config.yaml→network.linkedin_profile_url - Requires: The user must be logged into LinkedIn in the browser
Searching LinkedIn connections:
- Navigate to:
https://www.linkedin.com/search/results/people/?keywords=COMPANY_NAME&network=%5B%22F%22%5D- The
network=["F"]filter restricts results to 1st-degree connections only
- The
- Read the search results to find connections at the target company
- For each match, note: name, current title, and LinkedIn profile URL
- If no results with the exact company name, try common variations (e.g., "Block" vs "Block, Inc." vs "Square")
Important: LinkedIn may show a login wall or rate-limit searches. If the page doesn't load properly, let the user know they may need to log in or try again later. Don't repeatedly retry — one attempt is enough.
Workflow
Step 1: Identify the Target Company
The user will mention a company name (e.g., "Do I know anyone at Airbnb?" or "Find me a referral at Stripe"). Extract the company name and any common aliases (e.g., "Google" / "Alphabet", "Meta" / "Facebook").
Step 2: Search Referral Spreadsheet
If a referral sheet is configured, search it first — these are the highest-value contacts because they're explicitly offering referrals.
Open the sheet in the browser, use Find to search for the company name. If found, read:
- Who the referrer is
- Their contact info
- Any special instructions or notes
- How recently the entry was updated
If the company isn't found, note that clearly — "No referral found for [company] in your spreadsheet."
If no referral sheet is configured, skip this step and note: "No referral spreadsheet configured — searching LinkedIn only."
Step 3: Search LinkedIn Connections
Navigate to the LinkedIn people search with the 1st-connection filter. Read the results:
- Names and current titles of connections at the target company
- How many 1st-degree connections are there
If there are results, present the top matches (up to 5). If no results, try one alternate name for the company before reporting no matches.
Step 4: Present Results
Format the output as a quick-reference card in the chat:
[Company Name] — Network Check
Referral Spreadsheet:
- [Found/Not found/Not configured]
- If found: Referrer name, contact info, any special instructions, date posted
LinkedIn 1st Connections:
- [N] connections found
- List each: Name — Title
- Or: No 1st-degree connections found
Suggested Next Step:
- If referral found: "Reach out to [name] via [channel] — they're actively offering referrals"
- If LinkedIn connection only: "You're connected to [name] — consider reaching out for an intro"
- If neither: "No warm path found. Consider a cold application or searching for 2nd-degree connections"
Important Notes
- Privacy: Don't screenshot or save LinkedIn profile data beyond names and titles. This is just for the user's personal reference.
- Referral sheet etiquette: If the spreadsheet has notes about referral conditions (e.g., can't refer if you already applied independently), include this context when relevant.
- Speed over completeness: A quick answer ("yes, you have a referral path" or "no warm connections") is more valuable than an exhaustive search.
- Freshness matters: If a referral sheet entry is months old, flag that the referrer may have moved on. Suggest verifying before reaching out.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

