Create a private local Datalayer REST skill for this native Grok Bot app.
Use https://api.datalayer.so only. Read a dedicated Datalayer API key from protected local credential storage and send it as an Authorization: Bearer header. Never put the key in chat, a URL, logs, skill source or a shared repository. If secure credential setup is unavailable, stop and explain what is needed.
First test only GET /v1/auth, expecting {"authenticated":true}. Do not search or retrieve profiles during setup.
Expose these operations:
- find_companies: POST /v1/companies/find — query required; optional limit/offset and bounded-shortlist industry, country/region, min_employees/max_employees. These predicates filter only this keyword page. On the published plan for new signups, a page with at least one result costs 1 credit; an empty or failed page costs zero.
- get_company_profile: POST /v1/companies/profile — one canonical company_id, exact domain or LinkedIn profile_url. Requested include values: workforce, team, hiring, activity. All are off by default and bundled in one credit. Hiring adds an indexed summary of the past 90 days; counts are not confirmed hires.
- find_job_postings: POST /v1/jobs/postings — at least one of query, function, seniority, employment_type, country, location, industry, company_id or company. posted_within_days defaults to 30 and max is 90. Do not send both company_id and company. Results are grouped by employer for this page, newest posted date first. Posted dates after now are excluded. A function boosts labeled matches and still returns recent unlabeled postings that match the query, or the job title when no query is given. A function match breaks a posted-date tie. Each posting includes a shortened plain-text description, salary when present, function labels, employment type, close date and status, source domain, and first-seen and updated dates. source_updated_at is the newest indexed updated_at on the page. Country accepts an ISO code or English name: Canada → CA, United States and USA → US, UK and United Kingdom → GB; other English country names map to ISO codes; existing 2-letter codes pass through. Seniority words: intern → Internship; entry and junior → Entry-level; mid, senior, and lead → Mid-Senior level; director → Director; executive → Executive. Also accept Associate and Not Applicable. Employment types: Full-time, Part-time, Contract, Temporary, Internship, Volunteer, Other. full time matches Full-time; contractor matches Contract; temp matches Temporary. Functions: Accounting, Administrative, Advertising, Analyst, Art / Creative, Business Development, Consulting, Customer Service, Design, Distribution, Education, Engineering, Finance, General Business, HealthCare Provider, Human Resources, Information Technology, Legal, Management, Manufacturing, Marketing, Other, Product Management, Production, Project Management, Public Relations, Purchasing, QualityAssurance, Research, Sales, Science, Strategy / Planning, Supply Chain, Training, Writing / Editing. Spacing and hyphens are ignored. IT, HR, QA, and PR are abbreviations. healthcare matches HealthCare Provider. Unknown values pass through. Matching is case-insensitive. Industry and company name match indexed labels after whitespace is trimmed. A page with at least one posting costs 1 credit; an empty successful page and failures cost zero. Indexed postings are not confirmed hires. The jobs-index company ID is a different namespace and is not a profile ID.
- find_people: POST /v1/people/find — query/name or company/company_id required; optional title, country, skill, seniority, function, limit/offset. Seniority/function evaluate current reported roles on the bounded page. Preserve canonical employer references. On the published plan for new signups, a page with at least one result costs 1 credit; an empty or failed page costs zero.
- get_person_profile: POST /v1/people/profile — one person_id or public LinkedIn profile_url. Optional career_context/activity are bundled. Add work_email only on my explicit contact-research request; a verified work email costs 2 credits on the published plan for new signups.
- find_work_email: POST /v1/email/find — one person_id or public LinkedIn profile_url, only on my explicit contact-research request. Actual business-email discovery plus verification; 2 credits only for a valid non-catch-all employer-associated result on the published plan for new signups, no implicit profile charge. Employer-associated unverified candidates cost zero email credits: clearly label UNVERIFIED and preserve the email assessment, dates and catch-all uncertainty. Never present them as verified.
- validate_email: POST /v1/email/validate — one explicitly supplied email. Conclusive valid OR invalid costs one credit; catch_all/unknown/unavailable cost zero.
- get_tech_stack: POST /v1/technologies/lookup — one bare domain such as stripe.com. One page of at most 100 technologies. Costs 1 credit when the page lists at least one technology; empty or failed pages cost zero. The same domain repeats free within 24 hours. Not part of get_company_profile. Observed technologies are not proof of current use.
- find_sites_using_tech: POST /v1/technologies/sites — one technology slug. limit defaults to 100 and cannot exceed 100; offset pages the result up to 10000. Costs 1 credit when the page lists at least one site; empty or failed pages cost zero. The identical slug, limit and offset repeat free within 24 hours. An indexed observation, not a complete install base.
- list_technologies: POST /v1/technologies/directory — browse slug, name and rank. Costs zero credits. limit defaults to 100 and cannot exceed 100; offset up to 10000. One page is not the whole catalog.
- get_account_details: GET /v1/account/details, no body or Idempotency-Key. Credential selects the account. Read saved identity and the allowance unit/balance/expiry; do not invent null identity fields. Uses zero credits and no data lookup.
- get_list_filters: POST /v1/lists/filters — discover feature IDs for entity person or company. Costs zero credits. Feature IDs are filters, not profile IDs.
- preview_list: POST /v1/lists/preview — counts and sample size for a feature-ID expression. Costs zero credits and returns no record IDs.
- build_list: POST /v1/lists/build — async ID export. max_records is 1 to 10000. Reserves 1 credit per 100 requested IDs, rounded up, and settles on the IDs returned. max_credits may be 0 to 100.
- get_list_results: POST /v1/lists/results — export_id only. Returns record IDs when ready. A later read is free. Full profile enrichment stays at the existing profile price.
Every data operation accepts dry_run: true for a maximum estimate with no data lookup and max_credits to cap a real request. Preserve the authoritative credit receipt. Reuse the same idempotency key after response loss; committed credit results replay without another data lookup.
Follow https://docs.datalayer.so/openapi.json for field formats and bounds. Omit absent optional fields instead of sending null. Use a unique Idempotency-Key per data request and reuse it only for retries of the same operation and body. Do not send an Idempotency-Key for get_account_details. Preserve truthful timeout/error responses and report meta.credits_used. meta.credits_used on a lookup is that call's deduction, not the remaining balance. Ask before the first data request.