# Linkedin Prospecting Loop > LinkedIn prospecting loop - Skill: `matteotitta/linkedin-prospecting-loop` (Agent Skill) - Install (CLI): `npx skillmds@latest add matteotitta/linkedin-prospecting-loop` - Raw SKILL.md: https://api.skillmd.com/api/skills/matteotitta/linkedin-prospecting-loop/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: matteotitta (https://skillmd.com/u/matteotitta) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/matteotitta/linkedin-prospecting-loop --- # LinkedIn prospecting loop One approved motion that turns an ICP into a sequenced LinkedIn outbound run. It does not add a new capability — it chains the ones you already own (find, qualify, personalize, sequence, multi-channel) behind a human gate at every phase, so a run stops being five skills stitched by hand. Built from the `/steal` of Gojiberry's "Connect Claude to LinkedIn via MCP" playbook. The playbook's execution layer was a paid vendor MCP; this skill runs the same motion on Extrovert + Apify + Apollo instead. See `.claude/discovery/0626-gojiberry-linkedin-mcp-steal-analysis.md`. --- ## Doctrine inherited Output complies with: - [`output-tenets.md`](../../../../../rules/output-tenets.md) — the seven tenets - [`output-simplicity.md`](../../../../../rules/output-simplicity.md) — length caps, robot-tells ban - [`outbound-research-hygiene.md`](../../../../../rules/outbound-research-hygiene.md) — dated signals (<12 months), current-company-only hooks, no invented stats - [`ai-speak-anti-patterns.md`](../../../../../rules/ai-speak-anti-patterns.md) — the 12 patterns DMs can't carry - [`linkedin-cold-dm-doctrine.md`](../../../../../rules/linkedin-cold-dm-doctrine.md) — the cold-start motion: 9 tactics, connection/InMail envelope, "easy no" cadence - [`marketing-psychology.md`](../../../../../rules/marketing-psychology.md) — Rule of 7 (touch count), loss aversion, jobs-to-be-done - [`pii-redaction.md`](../../../../../rules/pii-redaction.md) — mask end-client PII before model/store/share (fires on ClientCo/FCA data) --- ## When to run **Invoke when the user says:** - "run a LinkedIn prospecting loop for [ICP/client]" - "find, qualify, message and sequence [segment] on LinkedIn" - "build me a cold LinkedIn outbound run end to end" **Do NOT invoke when:** - One-off warm DM off a single signal → `linkedin-social-selling` - Pure prospect-find with no sequencing → `linkedin-engagement-prospects` - Writing LinkedIn content → the `linkedin-*` content skills - Email-only outreach → `outreach-emails` --- ## The loop ``` ICP ─► [0] Health + load ─► [1] Find + signal ─► [2] Qualify ─► [3] Personalize ─► [4] Sequence ─► [5] Multi-channel │ │ GATE │ GATE │ GATE │ │ GATE └─ Extrovert └─ Apify/Apollo └─ lead-scoring └─ cold-DM + └─ Day 0/4/9/30 └─ abm-campaign safety caps + 7 signals 0–10 route Extrovert voice + Rule of 7 + channel switch ``` Every gate is human approval. The loop **drafts and queues — it never auto-sends.** Sends route through Extrovert under its own LinkedIn safety caps. ### Phase 0 — Account health + ICP load Read the locked ICP (`icp-research`). Verify LinkedIn account safety before any sends — reuse the SSI / pending-request / acceptance-rate caps in [`linkedin-social-selling`](../social-selling/SKILL.md)& Limits (2026)". Stop here if the account is over caps. ### Phase 1 — Find + signal scan → GATE Pull prospects matching the ICP via `linkedin-engagement-prospects` (Apify modes), `apollo-io` enrichment, and the signal skills (`jobs-signal`, `niche-signal-discovery`). Score each prospect against the 7-signal taxonomy and flag the active-intent signals (see the premium reference). **Gate:** present the signal-ranked shortlist; user keeps the contact-now set. ### Phase 2 — Qualify → GATE Run `lead-scoring` (0–10 on fit / intent / accessibility → route: contact now / nurture / discard) on the kept set. **Gate:** user confirms the contact-now routing. ### Phase 3 — Personalize → GATE Draft per-prospect openers using [`linkedin-cold-dm-doctrine.md`](../../../../../rules/linkedin-cold-dm-doctrine.md) (notification-preview opener, feel-chosen, weightless ask, soft exit) and the warm plays in `linkedin-social-selling`. Voice via the Extrovert MCP. Every message passes the `ai-speak-anti-patterns.md` gate. **Gate:** user approves messages one by one (or in batches once the ICP is trusted). ### Phase 4 — Sequence Build the Day 0 / 4 / 9 / 30 cadence with signal-reset (see the premium reference), respecting Rule of 7 touch counts. Queue, do not send. ### Phase 5 — Multi-channel → GATE Apply the LinkedIn↔email channel-switch heuristic (in the premium reference) and hand non-responders to `abm-campaign` / `outreach-emails`. **Gate:** user approves the channel plan. --- ## Tool swap (Gojiberry → your stack) | Playbook step | This loop uses | |---|---| | Find + read signals | `linkedin-engagement-prospects` (Apify) + `apollo-io` + `jobs-signal` / `niche-signal-discovery` + Exa | | Qualify (0–10) | `lead-scoring` | | Personalize | `linkedin-cold-dm-doctrine` + `linkedin-social-selling` plays; draft via Extrovert MCP | | Sequence | the premium reference + Rule of 7 | | Multi-channel | `abm-campaign` + `outreach-emails` | No Gojiberry. Nothing here needs a new MCP — all four (Apify, Apollo, Extrovert, Exa) are already mounted. --- ## MCP credit gate This loop calls credit-bearing MCPs. Per [`apify-credits.md`](../../../../../rules/apify-credits.md) and [`apollo-credits.md`](../../../../../rules/apollo-credits.md): estimate cost and confirm before any Apify actor run or Apollo enrichment. Search is free; enrichment + scraping cost credits. State the estimate at the Phase 1 gate. --- ## Anti-hallucination guardrails 1. Never invent signal data — only reference real, dated engagement (per `outbound-research-hygiene.md`). 2. No fabricated metrics or fake social proof — use `[X]` placeholders if unknown. 3. Verify ICP fit per prospect — a signal alone is not a qualification. 4. Never auto-send — every phase gate is human approval. 5. Redact end-client PII before model/store/share for regulated clients (`pii-redaction.md`). --- ## Integration with other skills | Skill | Relationship | |---|---| | `icp-research` | Required upstream — supplies the ICP the loop runs on | | `linkedin-engagement-prospects` | Called in Phase 1 (find + signal) | | `lead-scoring` | Called in Phase 2 (qualify + route) | | `linkedin-social-selling` | Called in Phase 3 (plays + account-safety caps) | | `abm-campaign` / `outreach-emails` | Called in Phase 5 (multi-channel) | | `extrovert-sync` | Keeps the Extrovert voice seed current for Phase 3 drafting | --- ## Final ship gate Run `/premortem --output` before ship. See [`/premortem` skill](../../../../meta/orchestration/premortem/SKILL.md) for the 5 execution domains (will-it-resonate / will-it-convert / will-it-stay-on-brand / will-stakeholder-push-back / will-it-degrade-over-time) and output template. Trivial-case escape: `## Premortem\nNo failure modes — trivial change` satisfies the contract for genuinely trivial outputs. --- ## Persuasion & stickiness pass Output complies with [persuasion-and-stickiness.md](../../../../../rules/persuasion-and-stickiness.md) — Cialdini's 7 persuasion levers + Heath's SUCCESs. Deploy the 1-2 Cialdini levers that fit the reader's barrier (never all seven; every lever must be TRUE), run the SUCCESs diagnostic (Simple / Unexpected / Concrete / Credible / Emotional / Stories) over the near-final draft, then the rule's pre-ship gate. ---