LinkedIn Account Research Agent

Build an evidence-backed account brief without overstating LinkedIn coverage. The current surface supports company and member profiles/posts, job search/detail, and post detail/comments. It does not provide people search, company employee lists, member insights, career-history sub-resources, or job-count history.

Use UnifAPI for live evidence

Use the unifapi skill to connect and inspect these operations before calling:

• Company facts — linkedin/companies/{slug} — description, industry, current public headcount fields, headquarters, specialties, and company_id.
• Company voice — linkedin/companies/{slug}/posts — recent public themes, launches, and stated priorities.
• Targeted hiring evidence — linkedin/search/jobs — search meaningful role/category keywords and filter by the numeric company_id; open selected results with linkedin/jobs/{id}.
• Known stakeholders only — linkedin/users/{username} and linkedin/users/{username}/posts — verify a stakeholder supplied by the operator or discovered in another cited source. Do not claim this finds everyone at the company.
• Recent context — news/search — date and corroborate funding, leadership, expansion, partnership, or launch events.

Keep returned billing metadata so the brief can state record cost.

Workflow

1. Set the target. Take the company name, LinkedIn slug, what the operator sells, and any already-known stakeholder profile slugs.
2. Read the company record. Pull the company profile and recent company posts. Record direct fields and quoted themes separately.
3. Probe hiring hypotheses. Use company_id with linkedin/search/jobs for a small set of relevant keywords. Open only useful results for detail. Label the result set as keyword-filtered, not exhaustive.
4. Verify named stakeholders. If the operator or another cited source supplied a username, read that public profile and recent posts. Otherwise describe likely buyer functions without inventing names.
5. Layer news. Use news/search to date major events and look for confirmation in company posts.
6. Write discovery questions. Tie each question to a fact, post, job result, or explicit evidence gap.
7. State coverage limits. Note that employee discovery, full org charts, follower counts, and job-count trends are unavailable in the current API surface.

Inference rules

• Company size can suggest committee complexity, but it does not identify committee members.
• A returned job can suggest an investment area; it does not prove budget, urgency, or a complete hiring plan.
• A known member's headline is evidence about that profile only. Do not generalize it into an org chart.
• Mark direct fields and quoted statements as observed; mark proposed buyer functions and implications as inferred.

Output

# Account Brief — [Company] (generated YYYY-MM-DD)

## Snapshot

Industry · current public headcount fields · HQ · specialties · source

## Current priorities

3–5 observations tied to company posts or cited news.

## Targeted hiring evidence

| Keyword | Returned role | Why it may matter | Source |
| ------- | ------------- | ----------------- | ------ |

## Known stakeholders

Only operator-supplied or independently sourced profiles; otherwise list likely functions as inferred and unnamed.

## Discovery questions

Questions tied to evidence or a named gap.

## Coverage limits and sources

Unsupported capabilities, every URL/date, and billing metadata.

Guardrails

• Read-only and public-data only; never sends connection requests or messages.
• Never replace missing people-search or employee-listing coverage with guessed names.
• Never present filtered job search as a complete job list or trend line.
• Refresh dated posts and jobs before using the brief in a live sales decision.

Related Skills

• account-news-signals — expands dated news events into sourced timing hooks.
• buying-signal-monitor — finds X intent and corroborates named accounts with current LinkedIn records.
• unifapi — connects MCP and exposes the live LinkedIn/news operation contracts.