find-engineering-firm

Drive the ServiceGraph API (https://api.servicegraph.co) to find, shortlist, and enrich US real-world engineering firms — buildings, infrastructure, energy, manufacturing — via the pro_services dataset. The catalog tags firms with industry:engineering_services and a ~23-tag service sub-taxonomy.

This is NOT for software engineering. "Software engineering firm", "engineering team", "engineering manager hire", "system architecture review" — those are all software-developer territory. The first thing this skill confirms is that the user means real engineering, not software.

Always pin `industry:engineering_services`. Note the `_services` suffix — older docs sometimes show industry:engineering but the literal pin returns zero results in the live catalog. Sub-disciplines are typically structured service_provided tags (civil-engineering, structural-engineering, mep-engineering, etc.) — confirm exact names via /v1/datasets/pro_services/fields?include_values=1. Surveying is surveying-mapping, not surveying.

Any HTTP client works (curl, fetch, requests). Examples below use curl.

When NOT to use this skill

The dominant failure mode is firing on software-engineering asks. Refuse those.

• Software / SaaS / app-development asks — even when phrased as "software engineering firm", "engineering team partner", "build our platform" → defer to find-software-developer.
• In-house "engineering manager" hires — in modern B2B usage this is almost always software-engineering recruiting. Refuse.
• System / cloud / database architecture questions ("Postgres vs DynamoDB", "monolith vs microservices") — software topics.
• Architecture (buildings) — find-architecture-firm covers that industry (when it exists). This skill covers engineering disciplines that work with architects (structural, MEP, civil), but the architect-of-record procurement is a different fire.
• Consumer/residential ("architect for a residential remodel", "engineer to inspect my house"). Catalog is B2B procurement only.
• Non-US firms / individual freelance engineers / DIY questions / engineering-software product comparisons (Revit, AutoCAD, SolidWorks, Ansys).

MCP server (preferred for authed calls)

If your harness has the ServiceGraph MCP server loaded (tools containing servicegraph), prefer those — OAuth 2.1 + PKCE keeps the token in the harness sandbox. Otherwise use the REST flow below.

API surface (dataset id: pro_services)

Every endpoint requires the bearer (Authorization: Bearer vk_…). No anonymous tier.

Cost model. Discovery / validation / search / brief reads are free. Detail (url, phone, email, social, address, full platforms map) costs 10 credits per firm and lasts 30 days.

Auth

vk_* API keys minted in the dashboard. Keep the token out of the LLM context — never read .env* into your context; dispatch via shell.

1. Try the call first through a shell wrapper that sources .env.local:

   ( set -a; [ -f .env.local ] && . ./.env.local; set +a;
     curl -sS -H "Authorization: Bearer $SERVICEGRAPH_API_KEY" \
          'https://api.servicegraph.co/v1/datasets/pro_services/fields' )

1. On `401` prompt the user:

"Open https://servicegraph.co/profile/api-keys, create a key, and add SERVICEGRAPH_API_KEY=vk_… to .env.local here (or export it). Tell me when done. Please don't paste the key into chat."

1. Retry after the user signals ready.

Filter DSL

GitHub-search-style.

filter   := orExpr
orExpr   := andExpr ("OR" andExpr)*
andExpr  := notExpr (("AND")? notExpr)*    # whitespace = implicit AND
notExpr  := ("NOT" | "-") notExpr | atom
atom     := "(" filter ")" | predicate
predicate:= IDENT op valueOrList | bareword
op       := ":" | "=" | ">=" | "<=" | ">" | "<"
valueOrList := value ("," value)*
value    := IDENT | NUMBER | tagAtEvidence
tagAtEvidence := IDENT "@" ("low"|"medium"|"high")
bareword := IDENT | NUMBER          # → keyword:<bareword>

Four rules that bite: AND binds tighter than OR (use parens); comma list = OR within one predicate; negation is -x or NOT x; bareword = keyword search (quote multi-word phrases).

Engineering-flavored examples (validate yours with /check):

industry:engineering_services service_provided:civil-engineering state:FL transportation
industry:engineering_services service_provided:structural-engineering high-rise
industry:engineering_services service_provided:mep-engineering hospital
industry:engineering_services service_provided:mechanical-engineering hvac industrial
industry:engineering_services service_provided:environmental-engineering remediation
industry:engineering_services service_provided:geotechnical-engineering state:CA
industry:engineering_services service_provided:civil-engineering@high has:clutch
industry:engineering_services service_provided:electrical-engineering manufacturing

Sub-discipline → tag mapping (verify exact names via /fields; ~23 sub-tags so this isn't exhaustive):

Verticals (hospital, high-rise, semiconductor, retail, datacenter) and credentials (PE-licensed, LEED, AICP) are keyword-only.

Identifying firms — apex

Firms are identified by their apex domain (aecom.com, not www.aecom.com/about).

Recipes

A. Civil engineering for transportation, in a state

GET /v1/datasets/pro_services/search?filter=industry:engineering_services+service_provided:civil-engineering+state:FL+transportation&limit=10
# Present, get pick of 3. "Unlocking 3 = 30 credits, 30-day TTL."
POST /v1/datasets/pro_services/unlocks
  { "apexes": ["firm-a.com", "firm-b.com", "firm-c.com"] }

B. Structural engineering + commercial high-rise

GET /v1/datasets/pro_services/search?filter=industry:engineering_services+service_provided:structural-engineering+(high-rise OR commercial)&limit=10

C. MEP for a hospital project

GET /v1/datasets/pro_services/search?filter=industry:engineering_services+service_provided:mep-engineering+hospital&limit=10

D. Mechanical / HVAC for industrial facility

GET /v1/datasets/pro_services/search?filter=industry:engineering_services+service_provided:mechanical-engineering+hvac+industrial&limit=10

E. Indirect intent — "stamp the drawings"

User: "We're building a 10-story office and need a structural engineer to stamp the drawings."

GET /v1/datasets/pro_services/search?filter=industry:engineering_services+service_provided:structural-engineering+commercial&limit=10

PE-license details surface from the unlocked platforms map / firm detail. If the user wants only PE-licensed firms, add pe as a keyword.

F. Geotechnical in California

GET /v1/datasets/pro_services/search?filter=industry:engineering_services+service_provided:geotechnical-engineering+state:CA&limit=10

Avoid pinning narrow keywords like seismic on top — engineering verticals tend to collapse below k=20 quickly. The full state pool of ~20 geotechnical firms is more useful than a 2-firm pool after keyword narrowing.

G. Quality threshold — be careful

Engineering firms aren't reviewed like agencies; rating>=N collapses civil/structural/MEP pools sharply. Prefer @high + size + geography as the quality proxy instead of a rating gate:

GET /v1/datasets/pro_services/search?filter=industry:engineering_services+service_provided:civil-engineering@high+state:CA+-company_size_signal:solo&limit=10

If the user insists on rated firms, add has:clutch (more lenient than rating>=4).

H. BYO apex list — enrich domains

User pastes 8–20 engineering firm domains:

1. GET /v1/datasets/pro_services/:apex per domain — free brief

(404 = not in catalog, no charge). A 404 often means the domain is a software firm tagged industry:it_services instead of real engineering.

1. User picks N to fully enrich. POST /unlocks = 10×N credits,

atomic, detail returned.

1. Re-runs within 30-day TTL are free.

Gotchas

• Always pin `industry:engineering_services`. Without it, civil-engineering / mep-engineering / electrical-engineering keywords could leak into other industries (architecture firms sometimes list MEP coordination as a sub-service).
• It's `engineering_services`, not `engineering`. The literal pin industry:engineering returns zero results. Same drift logic for surveying-mapping (not surveying).
• The hardest boundary is software-engineering. "Engineering firm" / "engineering team" in modern B2B usage almost always means software in tech contexts. Defer those to find-software-developer. Real engineering is for buildings, infrastructure, energy, manufacturing — not for SaaS apps.
• "Engineering manager" hires are recruiting, not procurement.
• Architecture firms are a separate industry. Architect-of-record procurement (form/aesthetics) is a different fire; this skill is for the engineering disciplines that work with architects.
• Consumer/residential is out of scope ("architect for my house remodel", "engineer to inspect my deck"). B2B-only.
• Engineering verticals are rating-sparse. Civil/structural/MEP firms aren't reviewed like agencies — rating>=N collapses pools below k=20. Use @high evidence + size + geography as the quality proxy.
• Narrow keywords artificially shrink pools. A keyword like seismic on top of a CA geotech filter yields only ~2 firms vs ~20 without it. Drop narrowing keywords for default-required verticals.
• Engineering-software product comparisons (Revit, AutoCAD, SolidWorks, Ansys, Bentley) are NOT procurement.
• Briefs DO include `apex`, `name`, location, ratings. They DON'T include url, phone_primary, email_primary, legal_name, address_full, full platforms — those require an unlock.
• `not_found` / `not_in_dataset` 404 = not in `pro_services`. Skip; not charged.
• Unlock is atomic. N apexes either all charge (up to 10×N credits) or none on 402.
• Within-TTL re-views are free (was_cached:true).

Errors

JSON envelope: {"error": {"code": "...", "message": "..."}}.

End-to-end example

User: "Three civil engineering firms in Florida focused on transportation infrastructure, ideally with PE-licensed engineers."

GET /v1/datasets/pro_services/fields?include_values=1
GET /v1/datasets/pro_services/check?filter=industry:engineering_services+service_provided:civil-engineering+state:FL+transportation+pe
GET /v1/datasets/pro_services/search?filter=...&limit=10
# Present briefs. "Unlocking 3 = 30 credits, 30-day TTL."
POST /v1/datasets/pro_services/unlocks
  { "apexes": ["firm-a.com", "firm-b.com", "firm-c.com"] }
GET /v1/me/credits