Skill instructions
enrich-contact7.26 KB
View saved version →
---
name: enrich-contact
description: "Look up a professional business contact and get their verified work phone numbers, work email, and company context. Use when the user says \"look up [name]\", \"get me the contact info for [person]\", \"find [name]'s phone number\", \"who is [name] at [company]\", \"enrich [email or name]\", or any request to retrieve a single person's professional business contact details.\n"
---
# Enrich Contact
Look up a professional business contact in Lusha and return a call-ready contact card. Phone numbers — direct line and mobile — lead the output.
Only return professional contact data supplied by Lusha. Do not invoke Lusha for home addresses, residential information, private personal phone numbers, or other non-business personal data.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `enrich-contact: ` (or with the calling skill's name, when another skill delegated here).
2. Reveal once per person, and say how much of the quota already included in the user's existing Lusha account it uses when the user hasn't asked for the reveal outright. This workflow cannot change the user's account entitlement.
3. If a response carries `next_step_hints` and the user picks one of its offers, pass its `ref` as `next_step_ref`.
## Step 1 — Parse Input
Extract all available identifiers from the user's request. `contacts_search` accepts three lookup paths:
- **Email** — standalone, strongest match
- **LinkedIn URL** — standalone, strong match
- **First name + last name + company name** — all three required together
A job title alone is not a lookup path. If the user gives only a title + company (e.g. "the CFO of Stripe"), there is no name to look up — first surface candidates with `prospecting_contact_search` (jobTitles + company), then enrich the chosen one. Only ask for clarification when no usable identifier is present at all.
## Step 2 — Look Up and Reveal
`contacts_search` has an `enrich` flag that controls whether the call reveals phones and email using quota already included in the user's existing Lusha account. Pick the path by how confident the match is — never reveal the same person twice because that may use the existing-account quota twice. This workflow cannot change the user's account entitlement.
**One-shot (preferred when the identifier is unambiguous — an email, a LinkedIn URL, or a clean name + company):**
Call `contacts_search` with `enrich: true` (the default). The response returns the profile *with* verified phones and email in a single call. You're done — do not call `prospecting_contact_enrich` afterward.
**Preview-then-reveal (when a complete identifier is available but the match may be ambiguous — for example, a common name with a company or multiple likely people):**
1. Call `contacts_search` with `enrich: false` — this returns a preview only and does not use reveal quota.
2. If multiple candidates come back, present the top 2–3 and ask the user to confirm.
3. Call `prospecting_contact_enrich` with the chosen result's `id` and `reveal` set from its `canReveal[].field` to reveal phones and email once.
**Restricted or do-not-call.**
- **`COMPLIANCE_RESTRICTED`** in place of the contact means it can't be shown, for compliance reasons, and no quota was used. Say so; don't retry with another lookup path.
- **`doNotCall: true`** marks a phone that is masked and not to be called. Show it as "Do not call".
- **A different company.** If the result shows a different company from the one the user named, it may be the same person after a job change, or someone else with the same name.
- Don't treat it as the same person until the user confirms. Show both companies and the LinkedIn URL on the card, and ask.
- When another skill delegated here (as `outreach-sequence` does), treat it as not this contact. Follow that skill's rule.
**Background reveals.** Either path can return a `job` block: some rev
Preview of the first 4,000 characters. Read full instructions →
lookalike-prospect8.16 KB
View saved version →
---
name: lookalike-prospect
description: "Find companies or contacts similar to a set of references, then enrich results with verified phone numbers. Use when the user says \"find companies like my best customers\", \"find more contacts like these\", \"expand from these accounts\", \"who else looks like [company]\", \"find similar companies to [list]\", or any request to discover lookalike targets from a reference set. Requires at least 5 reference companies or contacts for quality results.\n"
---
# Lookalike Prospect
Expand an ICP from a reference set of known-good companies or contacts. Requires a minimum of 5 references — the lookalike model degrades significantly below this threshold.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `lookalike-prospect: ` (or with the calling skill's name, when another skill delegated here).
2. State how much of the quota already included in the user's existing Lusha account each reveal uses, and wait for a yes on large batches. This workflow cannot change the user's account entitlement. On a 402: if the existing entitlement is insufficient, say so and offer to continue with previews or a smaller batch. Never suggest changing the user's account entitlement.
3. Keep the chain short: every call counts toward the user's AI usage.
4. If a response carries `next_step_hints` and the user picks one of its offers, pass its `ref` as `next_step_ref`.
## Step 1 — Validate Input Count
Count the number of reference companies or contacts provided.
**If fewer than 5 are provided**, stop and explain before doing anything else:
> "Lusha's lookalike model needs at least 5 reference [companies/contacts] to return quality results — fewer than that produces unreliable matches. You've provided [N]. Can you add [5−N] more?"
Do not proceed until the user has provided at least 5 references.
**If 5 or more are provided**, confirm the reference set with the user:
> "Running lookalike search using these [N] [companies/contacts] as the reference set: [list]. Shall I proceed?"
## Step 2 — Determine Mode
Based on the user's input, determine whether this is a **company lookalike** or **contact lookalike** search:
- References are companies (domains, LinkedIn company URLs, or names) → **company mode**
- References are people (emails, LinkedIn profile URLs, or name + company) → **contact mode**
- Mixed input → ask the user to clarify
A bare job title is not a valid seed — a lookalike needs concrete reference companies or people. If the user only has a persona/title in mind, route them to `prospect` (ICP search) or `signal-prospect` instead.
## Step 3 — Assemble the Seed Set
The lookalike tools accept raw identifiers directly as seeds — no enrichment or Lusha-ID resolution is needed in the common case. Pass the references straight through. The seed count (5–100) is the **total** identifiers across the seed arrays.
**Company mode** — `lookalike_companies.seeds` accepts:
- `domains` (e.g. `lusha.com`)
- `linkedinUrls` (company page URLs)
If the user gave company **names** rather than domains, resolve each name to a domain first with `companies_search` (`enrich: false` — you only need the domain, not reveal data), since the seed schema does not accept bare names. If a name can't be resolved, flag it and proceed only if ≥5 seeds remain.
**Contact mode** — `lookalike_contacts.seeds` accepts any mix of:
- `emails`
- `linkedinUrls` (profile URLs)
- `contacts` — `{ firstName, lastName, companyDomain | companyName }`
- `contactIds` — Lusha contact IDs, if you already have them
Pass whatever form the user provided directly. No lookup step is needed.
## Step 4 — Run Lookalike Search
**Company mode:** Use `lookalike_companies` with the seed set. `limit` up to 100 (default 25).
**Contact mode:** Use `lookalike_contacts` with the seed set. `limit` up to 50 (default 25).
Pass any known customers/won accounts in `exclude` (samPreview of the first 4,000 characters. Read full instructions →
outreach-research11.1 KB
View saved version →
---
name: outreach-research
description: "Builds a reusable outreach positioning brief: what the user's product does, its audience personas, with what proof, and against which competitors. Use when asked to build, review or update a positioning brief (including a pasted lusha-outreach-brief), run outreach research, or work out what to say to prospects before drafting (\"lusha outreach brief\"). Not for drafting emails: use outreach-sequence once a brief exists.\n"
---
# Outreach Research
## Purpose
Stage 1 of Lusha's outreach pair. Capture, once, what the user's own product does, its audience, with what proof and against which competitors, as a portable Markdown brief, lusha-outreach-brief.md. Stage 2, `outreach-sequence`, drafts copy from it. The user is a Lusha customer (a RevOps engineer, SDR, marketer or founder) preparing draft-only informational outreach about their own product: never promote Lusha. This workflow never sends, publishes, or executes external actions.
## Prerequisites
None beyond the conversation. This skill makes no Lusha calls. A host web-search tool is optional; without one, work from documents, the user's answers and prior knowledge.
## Instructions
### Rules
- **Text only.** Call no Lusha MCP tool.
- **Host tools only on the user's grant.** Read a file only when the user attaches, pastes or @-mentions it; never search the filesystem for a brief. Offer a web search and run it only once the user accepts. Write to disk only after the user confirms the path (Step 5).
- **No invention.** Every value comes from the user, a document they gave, a web search they accepted, or prior knowledge said as such. A value from a web search or prior knowledge carries ` (unconfirmed)` until the user confirms it (Step 4). A value nobody supplied is left blank with a `MISSING:` note, never guessed.
- **The user's own company.** The brief is for the company the user named as theirs. Never research a prospect's company, or a company that only appears in a contact list or an attached file, as if it were the user's.
### Step 1: Parse intent
- **A valid brief is already in the conversation and the user wants drafting** (contacts pasted, a CSV attached, "draft email 1 for this list"): run no intake. Point them to `outreach-sequence` with the brief and the audience; mention `prospect` only if they still need contacts.
- **The user names their company or product:** go to Step 2, then offer to draft the brief from prior knowledge or a web search rather than a blank page.
- **"Outreach brief for [Company]" with nothing that shows the user works there** (no attached material from that company, no "we provide", "I work at"): ask one question before drafting: *"Is [Company] your own company, or a target account you want to reach from a different company?"* For a target account, ask which company the user works at and build the brief for that one; the contacts come later, from their list or `prospect`. When the context already settles it, do not ask.
- **No intent:** open with one line, *"I'll help you build a positioning brief: what your product does, its audience, and how it differentiates, so outreach copy can be personalized later"*, and ask for the company name.
### Step 2: Detect an existing brief
Scan only what the user can see in this conversation: their messages, attached file pills, pasted blocks, @-mentioned files, and a brief rendered earlier in it. A brief is a Markdown block whose first non-blank line is `<!-- lusha-outreach-brief v2 -->` or `<!-- lusha-outreach-brief v1 -->`, followed by `# Outreach Positioning Brief`.
When one is found, parse it by the sections in `references/positioning-brief-schema.md`, echo one line, *"Loaded your brief: <company> · N personas · P proof points · M competitor displacements · default persona = `<persona>`. Looks current, or anything to change?"*, and enter Step 4. A v1 brief has no proof points or titles to match: offer once to add them. When no brief is found, go to Step 3.
Preview of the first 4,000 characters. Read full instructions →
Referenced files: 1 — file references are metadata, not package contents.
outreach-sequence20 KB
View saved version →
---
name: outreach-sequence
description: "Draft, but never send, personalized B2B outreach (Email 1, optional Email 2 and 3, and LinkedIn copy) for contacts the user already has, from an outreach-research positioning brief plus Lusha buying signals. Use when asked to draft outreach, write emails or LinkedIn messages for a contact list or CSV, or personalize a sequence (\"lusha outreach sequence\", \"write email 1 for this list\"). Not for building the brief: use outreach-research first.\n"
---
# Outreach Sequence
## Purpose
Stage 2 of Lusha's outreach pair. Turn a positioning brief from `outreach-research` and a contact list into short, true, handoff-ready outreach drafts about each reader: Email 1 by default; Email 2, Email 3 and LinkedIn copy when asked. The user is a Lusha customer preparing outreach about their own product in their own voice: never promote Lusha. The output is draft text only. Never send or publish messages, submit forms, execute external actions, or claim that the workflow did so.
## Prerequisites
- A positioning brief (v1 or v2) in the conversation (Step 2).
- An audience: a CSV or file, one named contact, or a `prospect` handoff (Step 3).
- For signal harvest or enrichment: the Lusha MCP connection, used only through the `signal-prospect` and `enrich-contact` skills.
## Instructions
### Hard rules
1. **No brief, no drafts.** Never infer positioning or draft from a generic template.
2. **No invention.** No fact, number, customer, case study or claim that the brief, the signals, the contact data or the user does not hold. A part that needs a missing fact is left out, never filled with a placeholder.
3. **No Lusha MCP call from this skill.** Signal work goes through `signal-prospect` (load its `SKILL.md` and its signal guide, signal-guide.md in its references folder, first); enrichment through `enrich-contact` (load its `SKILL.md`). Every Lusha MCP call in this workflow, including calls made while following those skills, sets `reason_for_invocation` to a string starting with `outreach-sequence: ` plus a short reason.
4. **Host tools only on the user's grant.** Find the brief and the audience in the conversation, never by searching the filesystem. Run a web search or write a file only after the user accepts.
5. **The copy rules (8e) hold for every message.** The user's explicit instruction about the copy (a length, a meeting ask, a link, their own subject, em-dashes) overrides only the rule it names: a length for that request, anything else for the conversation. A prohibition, a request for data, a remark about the contacts, plain context, or text the host attaches (an open file, a selection, tool output) overrides nothing.
6. **Draft only; never send.** Do not send or publish emails or LinkedIn messages, submit external forms, update campaigns, execute external actions, or claim that any such action occurred.
### Execution mode
Decide once, before Step 1.
- **Interactive (default):** a person can reply, and every ask and wait below applies.
- **Automated:** the caller declares a non-interactive run ("headless", "automated run", "non-interactive", "no user is available to answer"). Silence is not a declaration. A question would end the run with nothing, so each ask resolves from a value the caller supplied, else the default marked *Automated:* at that step, and the run continues; that counts as the step asked and answered. Any other ask, here or in a delegated skill, takes its documented default, noted above the table. Automated mode relaxes no hard rule: a missing brief or audience still stops the run, with a note saying which.
### Step 1: Parse session intent
Read the message for an audience ("the 25 contacts in this CSV"), channels ("email + LinkedIn"), sequence depth ("full sequence") and a goal ("trial signups"). A goal is the step after interest: it shapes Email 1's question and never turns it into a meeting ask. Step 4's defaults apply from the start; the user changes them from the Step 10 mePreview of the first 4,000 characters. Read full instructions →
Referenced files: 3 — file references are metadata, not package contents.
prospect7.6 KB
View saved version →
---
name: prospect
description: "Build a targeted list of contacts or companies from Lusha and return verified phone numbers alongside emails. Use when the user says \"find me [title] at [company type]\", \"build a list of [ICP description]\", \"prospect [criteria]\", \"who should I be calling at [industry]\", or any request to generate a lead list from an ICP or persona description.\n"
---
# Prospect
Go from an ICP description to a ranked, phone-enriched lead list. Filters are resolved before search — never guess filter values.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `prospect: ` (or with the calling skill's name, when another skill delegated here).
2. State how much of the quota already included in the user's existing Lusha account each reveal uses, and wait for a yes on large batches. This workflow cannot change the user's account entitlement. On a 402: if the existing entitlement is insufficient, say so and offer to continue with previews or a smaller batch. Never suggest changing the user's account entitlement.
3. Keep the chain short: every call counts toward the user's AI usage.
4. If a response carries `next_step_hints` and the user picks one of its offers, pass its `ref` as `next_step_ref`.
## Step 1 — Parse the ICP
Extract structured filters from the user's natural language description. Some filters take free-form text directly; others must be resolved to canonical values first.
**Contact filters (`prospecting_contact_search`):**
- Job titles → pass directly as `jobTitles` (free-form strings, e.g. "VP of Sales"). No resolution call needed.
- Department / seniority → resolve via `prospecting_contact_filters` (type: `departments`, `seniority`). Use these for broad role targeting when a specific title isn't given.
- Country → resolve via `prospecting_contact_filters` (type: `all_countries`); Location → type: `locations` (requires `locationSearchText`).
**Company filters (`prospecting_company_search`):**
- Industry → resolve via `prospecting_company_filters` (type: `industries_labels`)
- Size → resolve via `prospecting_company_filters` (type: `sizes`)
- Revenue → resolve via `prospecting_company_filters` (type: `revenues`)
- Location → resolve via `prospecting_company_filters` (type: `locations`, requires `q`)
- Tech stack → resolve via `prospecting_company_filters` (type: `technologies`, requires `q`)
- Buying intent → resolve via `prospecting_company_filters` (type: `intent_topics`)
Resolve every non-title filter to canonical values before searching — passing raw natural-language strings as structured filter values is the most common cause of search failures. Each `prospecting_*_filters` call resolves one filter type; run the independent lookups in parallel.
If the ICP is too vague to resolve (no title, no industry, no company size), ask one clarifying question before proceeding. At minimum, a title or department and at least one company-level constraint are required.
See `references/filter-guide.md` for filter resolution details.
## Step 2 — Search Companies
Use `prospecting_company_search` with resolved company filters. Request up to 25 results. This scopes the contact search to qualified accounts.
If the user only specified contact-level criteria (no company filters), skip this step and go directly to Step 3.
## Step 3 — Search Contacts
Use `prospecting_contact_search` with resolved contact filters. Scope to the company results from Step 2 where applicable. Request up to 25 results.
**No results?** Don't stop. Say exactly what was searched, then offer the widening that fits, most useful first:
1. An exact title (e.g. "CTO") → the department plus seniority (`departments` + `seniority` ids).
2. The narrowest company filter (size, location or industry) → drop it.
3. A location → its country or region.
## Step 4 — Enrich Top Results
Search results are previews — they carry no phones/emails but include a `
Preview of the first 4,000 characters. Read full instructions →
Referenced files: 1 — file references are metadata, not package contents.
save-or-export5.96 KB
View saved version →
---
name: save-or-export
description: "Save Lusha contacts or companies from this conversation to a Lusha Workspace table, or export them to the user's CRM (Salesforce, HubSpot and other supported CRMs). Use when the user says \"save these to a table\", \"add them to my Lusha list\", \"put this in a Workspace table\", \"export to HubSpot\", \"push these to Salesforce\", \"send to my CRM\", \"sync these leads\", or picks the save or export next step at the end of another Lusha skill. Works on Lusha ids already in the conversation; to find new people or companies, use prospect, signal-prospect or lookalike-prospect first.\n"
---
# Save or Export
The last step of every Lusha workflow: put the result somewhere the user will work from it. Either a Lusha Workspace table (a saved list in the Lusha platform) or the user's CRM.
Other skills hand off here with a set of records. When the user triggers this skill directly, collect the records from the conversation: the Lusha contact or company `id`s from earlier search, enrich, recommendation or scoring results. If there are none, say so and offer to find them first. Do not invent ids or search again for records the user already has.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `save-or-export: `. When another skill handed off here, use that skill's name as the prefix instead, so the whole workflow is attributed to the skill the user started.
2. If the last Lusha response carries `next_step_hints` and the user picked one of its offers, pass that hint's `ref` as `next_step_ref` on the call that carries it out.
3. Before any call that uses quota already included in the user's existing Lusha account, state the existing-account quota it will use, and wait for the user to agree. This workflow cannot change the user's account entitlement.
## Step 1 — Confirm what and where
Confirm in one line: how many records, whether they are contacts or companies, and the destination (table, CRM, or both). A table holds one entity type, so contacts and companies go to separate tables.
## Step 2a — Save to a Workspace table
1. **Pick the table.** If the user named an existing table, find it with `table_list` (`name` filter, uses no quota) and use its `tableId`. Otherwise create one with `table_create`: `entity_type` `contacts` or `companies`, and a name that says what the list is and when it was made, e.g. `Fintech VPs Sales · 2026-10-26`. Leave `visibility` private unless the user asks to share it.
2. **Add the records** with `table_add_entities`, up to 500 ids per call.
- **Contacts tables use no quota.** Pass `company_ids` index-aligned with `entity_ids` (each contact's Lusha company id) so the Workspace can enrich every row.
- **Companies tables use existing-account quota**: each newly added company uses the account's revealCompany quota rate (`account_usage` lists it). State the total first. The whole batch is reserved up front, so a batch the existing entitlement cannot cover fails with 402 and adds nothing. If the existing entitlement is insufficient, say so and offer to continue with a smaller batch. Never suggest changing the user's account entitlement.
3. Report `added`, `alreadyPresent` and `invalidIds`, plus the table name and id.
The table holds the records only. Columns such as tier or score cannot be written into it, so a skill that ranked the list keeps the ranking in its own output (chat and CSV).
## Step 2b — Export to the CRM
1. **Check the connection first** with `get_supported_integrations` (uses no quota). It lists every supported CRM; read `isUserConnected` / `isAccountConnected`.
- **A 403:** the session is an API key, which this check can't read (it needs a Lusha login). Don't conclude there is no CRM. Ask which CRM they use, and export directly to it; the export tools accept a personal API key.
- **One CRM connected:** use its `integrationName` as `platform_name`.
-
Preview of the first 4,000 characters. Read full instructions →
score-my-list7.04 KB
View saved version →
---
name: score-my-list
description: "Score and rank a list of companies or contacts the user already has (a CSV, a pasted list of domains or names, or a Lusha Workspace table) by fit to their ICP and by live buying signals, then split it into A, B and C tiers. Use when the user says \"prioritise this list\", \"score these accounts\", \"rank my target list\", \"which of these companies should I work first\", \"tier my accounts\", \"which of these leads are hot\", or brings a list and asks where to start. Not for finding new companies: use prospect, signal-prospect or lookalike-prospect for that.\n"
---
# Score My List
Most users start from a list they already own: a territory, an event export, a CRM view. This skill tells them which rows to work first, and why, using two Lusha scores:
- **ICP fit score** (0–1): how closely the company matches the user's own Target Profile in Lusha. It needs a profile.
- **Signal score** (0–1): how much buying-signal activity the company shows right now (hiring surges, funding, headcount growth and more). It needs no setup.
Together they make up the Score the user sees in the Lusha app.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `score-my-list: `.
2. Run the calls that use no quota first and branch on what they return. A missing profile gets a setup offer and a fallback, never an error. A tool missing from this session means the account's plan does not include it: take the fallback.
3. Before every quota-using call, state how much of the quota already included in the user's existing Lusha account it will use, and wait for a yes. This workflow cannot change the user's account entitlement. On a 402, if the existing entitlement is insufficient, say so and offer to continue with previews or a smaller batch. Never suggest changing the user's account entitlement.
4. Keep the chain short: every call counts toward the user's AI usage. Use the full batch size of each tool.
## Step 1 — Read the list
Accept any of:
- **A CSV or pasted list.** Use one identifier per row, in this order of preference: Lusha id, domain, email (its domain is used), company name. For a contacts list, use email, LinkedIn URL, or first + last name + company.
- **A Workspace table.** Find it with `table_list` (uses no quota). Get its row count with `table_status` (uses no quota), then read the rows with `table_get_entities`. Reading uses existing-account quota per block of rows (`tableRead` in `account_usage`), and quota for the full `page_size` is reserved up front, so set `page_size` to the row count (max 100 per page) and state the quota usage first.
Decide whether the list is **companies** or **contacts**. A list of people with company columns can be scored either way; ask which the user cares about if it isn't clear.
Drop exact duplicates and say how many rows will be scored.
## Step 2 — Check for an ICP profile (uses no quota)
Call `icp_profiles_get` with `status: "active"`.
- **Active profiles exist.** Use up to 3 (the score tools take 1–3 `targetProfileIds`). If there are more than 3, list them and ask which to score against.
- **No active profile, or the ICP tools are missing.** Score with signals only, and say: *"You don't have an ICP profile in Lusha yet, so I'll rank by buying signals only. Want to set one up with set-up-my-icp after this, so I can score fit too?"* Do not call the fit score tools: with no profile they fail and still give no score.
## Step 3 — State the quota usage and score
Estimate the quota for both scores from `account_usage` (`pricing`, the per-action quota rates; uses no quota), and give one total, e.g. *"Scoring 180 companies uses about 360 of the quota already included in your existing Lusha account (180 fit + 180 signal). Go ahead?"* If a score has no entry in `pricing`, say its rate isn't listed for this plan, and report the quota it actually used from the usage metadata afterwards. Of
Preview of the first 4,000 characters. Read full instructions →
set-up-my-icp6.39 KB
View saved version →
---
name: set-up-my-icp
description: "Set up the user's Ideal Customer Profile (an ICP Target Profile in Lusha) by asking the profile questions one at a time, drafting each answer from what is already known about their company and customers, then creating the profile and previewing 10 matching accounts. Use when the user says \"set up my ICP\", \"define my ideal customer\", \"who's my ideal customer\", \"create a target profile\", \"teach Lusha who my customers are\", \"my recommendations are off\", or when another Lusha skill sends them here because no profile exists. Not for a one-off search by criteria: use prospect.\n"
---
# Set Up My ICP
A Target Profile tells Lusha who the user's customers are: the companies (account criteria) and the people inside them (personas). Once it is active, recommendations, ICP fit scores and `score-my-list` all use it. This skill builds one with the user, one question at a time, and never asks a question it can answer with a draft.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `set-up-my-icp: `. When another skill sent the user here, keep that skill's name as the prefix.
2. Every profile tool uses no quota. The only quota-using calls here are the optional reads in Step 2 and the preview in Step 5; state how much of the quota already included in the user's existing Lusha account they will use first. This workflow cannot change the user's account entitlement.
3. Use only the criterion keys and values `icp_filters_get` returns. Anything else is refused.
## Step 1 — Load what exists (uses no quota)
Call `icp_profiles_get` and `icp_filters_get` together.
- **The ICP tools are missing from this session:** the account's plan doesn't include Target Profiles through this connection. Tell the user that profiles are set up in ICP Hub in the Lusha platform, and stop.
- **The user already has profiles:** list them (name, active or disabled). Ask whether to add a new one or change an existing one. Changes go through `icp_profile_update`. Updating a set of personas or the account criteria replaces that set in full, so show the full new set before saving. To pause or reactivate a profile, use `icp_profile_status_set`.
- **Limits** (in `icp_filters_get`): 10 profiles per user, 4 personas per profile. Near a limit, say so before starting.
`icp_filters_get` returns `profile_questions`: the order to settle a profile in. Follow it:
1. Account criteria
2. Each persona
3. Intent topics
4. Lusha-based exclusions
## Step 2 — Gather material for drafts
A good draft is what makes this fast: the user confirms or edits instead of starting from blank. Draft from, in order:
1. **What the user already said** in this conversation: product, customers, market.
2. **Their own company.** Use the `account_domain` from a `target_profile_suggestion` if a previous response carried one, or ask for their website. Look it up with `companies_search` (`enrich: false`, uses existing-account quota per successful lookup; say so first) for industry, size and location. The user's own company is a hint, not the answer: their customers may be in a different market or size than their own.
3. **Customers they name, or a saved list.** If the user mentions a list of customers or wins, find Workspace tables with `table_list` (uses no quota, by name). Reading rows with `table_get_entities` uses existing-account quota per block of rows (`tableRead` in `account_usage`), so state the quota usage and ask before reading. A sample of 25 is enough to see the common industries and sizes.
Skip any source the user doesn't want used.
## Step 3 — Ask one question at a time
Follow `profile_questions`. For each criterion:
- Show the draft as concrete values from `icp_filters_get`.
- Say in one line where the draft came from.
- Ask the user to confirm, edit, or skip.
For example: *"Industries — from your customers I'd target Financial Services and Software Deve
Preview of the first 4,000 characters. Read full instructions →
signal-prospect11.9 KB
View saved version →
---
name: signal-prospect
description: "Find companies or contacts triggered by a buying signal, then surface verified phone numbers for the right decision makers. Use when the user says \"find companies that just raised funding\", \"who's surging in sales hiring right now\", \"show me companies with headcount growth\", \"find contacts who just got promoted\", \"who changed jobs recently in [industry]\", \"show me [title] at companies that just [signal event]\", or any request that starts with a real-world trigger rather than a static filter.\n"
---
# Signal-Prospect
Start from a buying signal — company or contact level — and end with a call-ready list of enriched decision makers. This is Lusha's core differentiated workflow: trigger → identify → phone reveal.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `signal-prospect: ` (or with the calling skill's name, when another skill delegated here, as `outreach-sequence` does).
2. State how much of the quota already included in the user's existing Lusha account every quota-using call uses, and wait for a yes on large batches. This workflow cannot change the user's account entitlement. On a 402: if the existing entitlement is insufficient, say so and offer to continue with previews or a smaller batch. Never suggest changing the user's account entitlement.
3. Zero results is not the end of the session: say what was searched and what coverage exists, then offer to widen (Step 4a).
4. Keep the chain short: every call counts toward the user's AI usage.
5. If a response carries `next_step_hints` and the user picks one of its offers, pass its `ref` as `next_step_ref`.
## Step 1 — Identify the Signal Mode
Determine which mode applies based on the user's request:
**Company signal mode** — the trigger is something happening at a company (funding, hiring surge, headcount change, news event). Goal: find the right people at those companies and get their phones.
**Contact signal mode** — the trigger is something happening to a person (got promoted, changed company). Goal: find those people and get their updated contact details.
If unclear, ask: *"Are you looking for companies showing a specific signal, or individual contacts who recently changed roles or employers?"*
## Step 2 — Discover Available Signal Types and Sub-Filters
These tools are the authoritative source of valid signal identifiers — never assume a signal type exists without confirming it here, since invalid values are rejected with a 400.
**Company signals:** Call `signals_company_filters` with no `filterType` to get the directory: `{ signalTypes, availableFilters: [{ filterType, requiresQuery }] }`. To enumerate the values for a sub-filter, call it again with `filterType` set to `newsEventTypes`, `hiringByDepartments`, or `hiringByLocations` (`hiringByLocations` requires a `query`).
**Contact signals:** Call `signals_contact_filters` to get the supported contact signal types (e.g. `promotion`, `companyChange`, `allSignals`).
## Step 3 — Map User Intent to a Signal Type
Match the user's phrasing to a signal identifier returned in Step 2. The table is a starting point — always validate the identifier against the live directory before using it:
| User says | Signal type (`names`) | News sub-type (applied in Step 5) |
|-----------|----------------------|-----------------------------------|
| "raised funding / Series A/B/C / IPO" | `financialEventsNews` | `Funding Round`, `IPO`, `Strategic Investment` |
| "surging in hiring / lots of open roles" | `surgeInHiring` | — |
| "growing fast / headcount up" | `headcountIncrease3m` / `headcountIncrease6m` | — |
| "hiring sales reps" | `surgeInHiringByDepartment` (+ `filterByDepartment`) | — |
| "new partnership / new customer" | `commercialActivityNews` | `New Customer`, `Partnership`, `New Location` |
| "new product launch" | `productActivityNews` | `Product Launch`, `Product IntegrationPreview of the first 4,000 characters. Read full instructions →
Referenced files: 1 — file references are metadata, not package contents.
website-visitor-followup8.76 KB
View saved version →
---
name: website-visitor-followup
description: "Turn the companies visiting the user's own website into a call list: rank recent visitors by engagement, keep the hot ones, check them against the user's ICP, find the decision makers and reveal their phones. Use when the user says \"who visited our website\", \"which companies are on our site\", \"follow up on site visitors\", \"show me hot website visitors\", \"who's been looking at our product page\", or \"turn website traffic into leads\". Not for companies showing a buying signal elsewhere (funding, hiring): use signal-prospect for that.\n"
---
# Website Visitor Follow-up
A company that keeps coming back to the user's site, and reads the high-intent pages, is the warmest account they have. This skill finds those companies, keeps the ones that fit, and ends with the people to call.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `website-visitor-followup: `.
2. Run the calls that use no quota first and branch on what they return. No visit data means a fallback, never an error.
3. Before every quota-using call, state how much of the quota already included in the user's existing Lusha account it will use, and wait for a yes. This workflow cannot change the user's account entitlement. Get the quota rate for every quota-using call from `account_usage` (uses no quota; call it once per session). Each `pricing` entry (the per-action quota rates) names the tools it applies to. If a tool has no entry, say its rate isn't listed for this plan, and report the quota the call actually used from the usage metadata afterwards. Reveal only the people the user picks. On a 402, if the existing entitlement is insufficient, say so and offer to continue with previews or a smaller batch. Never suggest changing the user's account entitlement.
4. Keep the chain short: every call counts toward the user's AI usage.
5. If a response carries `next_step_hints` and the user picks one of its offers, pass its `ref` as `next_step_ref`.
## Step 1 — Rank the visitors (uses no quota)
Call `website_visits_search` with:
- `filters: { scoreBands: ["hot"] }`
- `sort: { by: "score", order: "desc" }`
- `pagination: { size: 25 }`
The default window is the last 30 days. Pass `startDate` / `endDate` if the user named a period; keep the range to 90 days or less, because a longer one is refused. Pass `domains` only when the user names one of their tracked sites.
- **No hot visitors, but the call worked:** say so, and run it once more with `scoreBands: ["warm"]`.
- **No warm visitors either:** run it once more with no `scoreBands` filter. That returns every band, cold included. Cold visitors are still visitors, so show them, labelled as cold. Offer a longer window, up to 90 days, if that call is empty too.
- **An error:** read it before deciding anything. An error is not "no tracking".
- **429 (rate limit):** wait, and retry once.
- **400 naming the date range, or a 500 on a long window:** retry once with 30 days.
- **Any other error, after one retry:** say the visit data couldn't be read right now, and offer the fallback. Don't say tracking isn't set up.
- **No visitors in any band, even over 90 days:** go to the fallback below.
Show the ranking:
| # | Company | Domain | Visit score | Band | Days visited | High-intent pages | Last visit |
|---|---------|--------|------------|------|--------------|-------------------|------------|
List up to three `highIntentPages` labels per company (e.g. "Product ×4, Demo ×1"). They tell the user what the visitor wants, and make a good opener.
## Step 2 — Keep the ones that fit (optional, uses quota)
Call `icp_profiles_get` with `status: "active"` (uses no quota).
- **Active profiles exist:** offer to drop visitors outside the ICP: *"Check these N companies against your ICP?"* Estimate its existing-account quota usage from the `icp_fit_score_companies` rate in `account_usage`. If Preview of the first 4,000 characters. Read full instructions →
work-my-accounts8.03 KB
View saved version →
---
name: work-my-accounts
description: "Tell the user which accounts to work next and why, from Lusha's recommendations ranked on their own ICP and live buying signals, then find and enrich the decision makers at the accounts they pick. Use when the user asks \"who should I go after\", \"which accounts should I prioritise\", \"what should I work on this week\", \"give me my best accounts\", \"any warm accounts for me\", \"where should I focus today\", or any targeting question that names no criteria of its own. Not when the user describes the companies or people to find (industry, size, title, location): use prospect for that.\n"
---
# Work My Accounts
Answer "who should I go after?" with ten accounts, a reason for each, and a call list for the ones the user picks. Lusha's recommendations are personalised to the user's Target Profiles and ranked by lead score (ICP fit plus live buying signals), so the user does not have to describe a market.
## Lusha call rules
1. Pass `conversation_id` from the most recent Lusha response on every call. Start every `reason_for_invocation` with `work-my-accounts: `.
2. Run the calls that use no quota first and branch on what they return. A missing profile gets a setup offer, never an error.
3. Before every quota-using call, state how much of the quota already included in the user's existing Lusha account it will use, and wait for a yes. This workflow cannot change the user's account entitlement. Get the quota rate for every quota-using call from `account_usage` (uses no quota; call it once per session). Each `pricing` entry (the per-action quota rates) names the tools it applies to. If a tool has no entry, say its rate isn't listed for this plan, and report the quota the call actually used from the usage metadata afterwards. Preview 10 before a full page. Reveal only the people the user picks. On a 402, if the existing entitlement is insufficient, say so and offer to continue with previews or a smaller batch. Never suggest changing the user's account entitlement.
4. Zero results is not the end: say what was searched, then offer to widen.
5. Keep the chain short: every call counts toward the user's AI usage.
6. If a response carries `next_step_hints` and the user picks one of its offers, pass its `ref` as `next_step_ref`.
## Step 1 — Recommend 10 accounts
Say: *"I'll pull your top 10 recommended accounts. That uses N of the quota already included in your existing Lusha account."* Work out N from the `recommendations_companies` rate in `account_usage`: the rate is per account returned, times 10. Then call `recommendations_companies` with `pagination: { size: 10 }` and no filters. Results come back sorted by lead score, which is right for this question.
Change the call only when the user asked for something specific:
- **"Who's hot right now"** → `sort: [{ by: "signalScore" }]`.
- **A named segment or one of their profiles** → get the ids from `recommendations_companies_filters` (uses no quota) and pass `filters.targetIcp`.
- **A named signal** (hiring, funding) → get the ids from the same filters call and pass `filters.signals`.
Never construct a profile or signal id.
- **Account-wide API key (403):** recommendations need a personal login. Say so, and offer `prospect` instead.
- **No results:** say the recommendations came back empty for this profile, and offer `prospect` with the user's own criteria.
## Step 2 — Say why each account is listed
| # | Company | Domain | Lead score | ICP fit | Signal score | Why it's here |
|---|---------|--------|-----------|---------|--------------|---------------|
"Why it's here" is one line built from the record: the target profile it matched, plus the signals that fired, e.g. *"Matches 'Mid-market fintech'; hiring surge in Sales, headcount +12% in 6 months."* Never write a reason the record does not support.
**No profile of their own:** if the matched profile is the *Lusha Calculated Target Profile*, or the response carries a `target_profile_suggePreview of the first 4,000 characters. Read full instructions →