GTM Engineer Pulse

Generate complete editions of the GTM Engineer Pulse newsletter — the weekly briefing on go-to-market engineering, AI, and revenue systems. User provides categorized links; output is a polished, Substack-ready newsletter matching the established voice and structure.

Doctrine inherited (Step 7 — 0626 rollout, locked 2026-06-04)

Output complies with output-tenets.md, output-simplicity.md, ai-speak-anti-patterns.md. Step 6 calibration: see [[feedback_execution_doctrine_refinements_step6]]. Cohort framing per [[feedback_gtme_pulse_conventions]] (Cohort 5 current).

Refinements applied: R1 (newsletter is end-customer-facing — sources convert to inline links in prose; no appended sources block), R2 (multi-section newsletter ships as one doc with toggles per section type), R3 (commentary capability-led, never "thrilled"), R5 (Pulse's own thread-anchor opens every section), R6 (cohort enrollment primary CTA per cohort-conventions memory), R9 (verb-led section headings).

---

When to invoke

Yes: "Create a new Pulse edition", "Write GTM Engineer Pulse #[X]", "Generate the Pulse newsletter", "Help me with the GTME Pulse", "/gtme-pulse"

No: general LinkedIn content → linkedin-content. Job board management only → handle directly. Edit existing edition → handle as direct edit task.

Full trigger detail + edge cases → the premium reference.

---

Inputs (minimum)

User must provide:

• Edition number, publication date
• Recent News (4-5 URLs), LinkedIn Hot Takes (5-7 URLs), GTME Jobs (5-10 URLs), Top GTMEs (4-5 LinkedIn profiles), Recommended Resources (5-8 URLs)
• Optional: theme, cohort CTA status (active | waitlist | none), cohort number, cohort start month + enrollment deadline, cohort link, special announcement, spots remaining (only if Matteo gives a real current number — never inferred)

If counts are below minimum, ask for more before proceeding. Full table + format spec → the premium reference.

---

Process — three phases

Phase 1 — fetch

Parse input → categorize links → fetch LinkedIn posts via Apify (supreme_coder~linkedin-post) → fetch profiles via Apify (harvestapi~linkedin-profile-scraper) → fetch other URLs via WebFetch → supplement profiles with WebSearch for generalized bios.

API token: api-keys/.env → APIFY_API_TOKEN. Full curl commands + extraction-per-section spec → the premium reference.

Phase 2 — generate

Write intro (multiple paragraphs, each 2-3 sentences, casual, reference highlights) → Recent News (bold linked headline + body inline, ONE paragraph of 2-3 sentences, NO line breaks within the bullet) → LinkedIn Hot Takes (link on headline NOT author, same one-paragraph 2-3 sentence rule) → GTME Jobs (**[Role @ Company](url)** | salary | location — body of 2-3 sentences inline, NO line break between meta-line and body) → Top GTMEs (**[Name](url)** — Role @ Company. Bio sentence + last-30-days post activity sentence(s), 2-3 sentences total, ONE paragraph, NO "Follow for X" closer) → Recommended Resources (**[Title](url)** by Author (context). Body of 2-3 sentences inline, ONE paragraph).

Hard rules every section follows:

• Every bullet body is ONE paragraph of 2-3 sentences. No line breaks within a bullet. No indented sub-paragraphs. If a topic needs more depth, condense or split into a separate bullet.
• No ** (bold) inside the body paragraph. The bold-link headline at the start is fine; nothing else in the body gets bold for emphasis. Use italics with single * for quotes/titles only.
• No "pairs with X" phrasing. If the pairing is real, state the parallel directly without the meta-commentary. Most of the time, just say the thing.
• Full conversational sentences, no abrupt sentence changes. Connect clauses with commas, em-dashes, or semicolons so the paragraph flows as one thought. No fragment punchlines.
• Sound like Matteo. Direct, operator-first, contractions ("you're", "it's", "won't"), em-dashes with spaces, sentence case headers. The 100 Posts Test is the final check.

Top GTMEs section — special rule: instead of a generic "Follow for X" closer, mention something from the person's last 30 days of LinkedIn activity if active; if inactive, pull a freshness angle from their profile (latest role change, headline, current company traction). The bio should ground in their lineage; the closer should ground in what they're doing NOW.

Per-section instructions → the premium reference. Templates with link rules → the premium reference. Voice rules → the premium reference. Section-level templates → the premium reference.

Phase 3 — assemble + polish

GDrive manifest line (auto, line 1) → sections (NO editor-notes HTML comment block — strip if present from prior edits) → closing ("Keep shipping, Matteo") → image suggestions (2-3, single-line each, no headshot grids unless the bullet has a real reason) → format polish (em dashes with spaces, bullets -, sentence case, link spans full title) → 2 Cialdini-powered CTAs (mid + end).

Detailed steps → the premium reference. Cialdini CTA templates → the premium reference. Output structure → the premium reference. Process flowchart → the premium reference.

Phase 4 — asset bundle (REQUIRED)

Download GTME Jobs company logos (logo.dev) + Top GTMEs LinkedIn headshots (Apify) for the edition. Output to ~/Downloads/pulse-{N}-assets/ with subfolders logos/ and headshots/, plus a manifest.md. Run before Phase 5 follow-ups so assets are ready when the GDoc / Substack publish step fires.

Coverage scope (narrow, deliberate):

• Logos: the 5 hiring companies in GTME Jobs only. NOT Recent News, NOT MCP-of-the-week, NOT Hot Take authors' employers, NOT Top GTMEs' employers, NOT Resources publications. The Jobs companies anchor a section the reader scans for visual recognition; everything else carries on link + headline alone.
• Headshots: the 4 Top GTMEs to Follow only. NOT the 5 Hot Take authors. Top GTMEs is the only section where the person IS the subject and a face accelerates recognition.

Hit rate from Pulse #30: 5/5 logos via logo.dev + 4/4 headshots via Apify apimaestro/linkedin-profile-batch-scraper-no-cookies-required ($0.02 total spend, under the apify-credits no-gate threshold).

Full procedure → the premium reference.

---

Critical guardrails (do not violate)

1. Link matching — re-read original input, checklist every URL, match each to its section. Common miss: official blog links, podcast/Spotify, LinkedIn buried in long lists.
2. Inline link formatting — link MUST span the ENTIRE title text. **[Title](url)** not **Tit[le](url)**. Google Docs export breaks if partial.
3. Profile research — Apify alone is insufficient. Pair with WebSearch "[Name]" [Company] background for generalized bios. Never write "in their latest post about…".
4. Job format — **[Role @ Company](url)** | Salary | Location. Link spans full role-at-company text.
5. CTA requirements — exactly 2 Cialdini-powered CTAs (mid after Jobs header, end before "Keep shipping"). Each uses ≥2 of: social proof, authority, scarcity, concrete. Every lever must be TRUE — scarcity is the real deadline and cohort start date, never a spot count or percentage-filled nothing can source. A fabricated number is rule 7 + anti-hallucination #1, not persuasion.
6. Hyphens - for body list items — not •, not →. Applies to ALL body items including CTA action links and job board link. One carve-out: the LinkedIn promo block uses • (plain text, LinkedIn renders no markdown).
7. News source verification — never write a summary from fallback web search if the source URL failed to fetch. Flag the failed URL; never fabricate.

Plus 7 anti-hallucination rules: never invent metrics, fabricate quotes, guess salary, assume titles, invent expertise. Mark missing as [Not available]. Full guardrails + per-rule examples + pre-delivery quality checklist → the premium reference.

---

The GTM Engineer Pulse | #[X]

SEO title: [<60 chars] Meta description: [<155 chars]

[Intro paragraph]

---

Recent News [4-5 items]

LinkedIn Hot Takes [5-7 items]

GTME Jobs [Cohort CTA if active] [5-10 items] • Check our full job board: gtm-engineer-jobs.com

Top GTMEs to Follow [4-5 profiles]

Recommended Resources [5-8 items]

Keep shipping, Matteo

Recommended Images to Add [3-5 items]

Full structure with separators + formatting rules table → the premium reference. Section-level good/bad examples → the premium reference. Full reference editions → `examples/pulse-14-example.md`, `examples/pulse-15-reference.md`.

---

## Required follow-ups

After approval:

1. **LinkedIn promo post** — REQUIRED. Template + 12 rules → `examples/linkedin-promo-template.md`. Real reference → `examples/pulse-20-linkedin-promo.md`.
2. **Publish — Notion + Substack draft (default since #32)** — create the Notion page (sibling of previous edition, manifest line added to local md) and stage the Substack draft via Claude in Chrome: title `The GTM Engineer Pulse | #N`, subtitle = meta description, body via synthetic HTML paste into the Tiptap editor, then the 4 CTA elements (2 blockquoted cohort CTAs + 2 native `subscribeWidget` nodes). Full procedure → the premium reference. Legacy GDoc path: `create-pulse.mjs` (NOT `create-doc-unified.mjs`) → the premium reference.
3. **Self-improvement loop** — REQUIRED. Ask user the 4 capture questions (what worked, what to fix, new guardrails, new examples). Full prompt + auto-capture triggers + pattern detection → the premium reference.

Iteration prompts (refine / expand / quality) → the premium reference.

---

## Research substrate (Exa)

**Default:** Exa, per `.claude/rules/exa-protocol.md`. **Primary tools:** `web_search_exa` (date-filtered for "recent" claims) and the plugin `/search` slash command for parallel topic deep-dives. **Fallback:** WebSearch with data-gap flag.

Use case: weekly news sweep with date filter; profile background supplementation; resource fact-check. Plugin namespace `mcp__plugin_exa_exa__*` preferred; legacy `mcp__exa__*` still mounted. Citation: `[VERIFIED: exa_search, {url}, accessed {YYYY-MM-DD}]`. Quality gate: ≥3 sources per major claim, ≥50% `[VERIFIED]`, no silent WebSearch fallback.

Worked examples + full tool catalog → `.claude/skills/meta-skills/exa/`.

---

## MCP integration

| Source | What | Tool | When |
|--------|------|------|------|
| **Exa** | Newsletter topic research, profile background | `web_search_exa` | Always |
| **Apify** | LinkedIn posts + profiles | `atomus~linkedin-reactions-scraper-pro` (post-details mode), `harvestapi~linkedin-profile-posts` | When URLs are LinkedIn (`supreme_coder~linkedin-post` is BROKEN since ~June 2026) |
| **WebFetch** | Job boards, blogs, Substack, YouTube | built-in | Non-LinkedIn URLs |

**Fallback (no MCP):** WebSearch for GTM topics; user-provided topic list.

---

## Skill chain

| Skill | Relationship | Usage |
| ----- | ------------ | ----- |
| `linkedin-content` | Related | Standalone LinkedIn posts (not newsletter hot takes) |
| `outreach-emails` | Complementary | GTM Engineer School promotion emails |
| `aeo-content` | Complementary | Newsletter archive SEO optimization |

---

## 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.

Then run `/voice-reviewer` — the content ship gate: voice + brand quality (pm-loop.md). The LinkedIn readability dimension is N/A for the Pulse body and fires on the promo post.

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.