1. KEEP A LINK-PREVIEW CARD OUT OF THE POST BODY. Put the launch link in the first comment (and say "link in the comments") — the low-risk default. The reach hit is mainly on the rich preview card, and the link penalty is contested/smaller in 2026 (see references/li-format-rules.md); don't over-engineer it — a native, dwell-worthy post matters more.
2. WIN THE HOOK. Only the first ~2 lines (~140–210 chars) show before "…see more" — that hook decides who expands. Lead with the strongest concrete line; no "I'm excited to announce…" throat-clearing.
3. ENGAGEMENT IS ALLOWED — BAIT IS NOT. You MAY ask your network and team to engage genuinely (employee amplification is fine and effective); there is no vote-ring rule. But never use engagement-bait that LinkedIn demotes ("comment YES below", "tag 3 people"), never use engagement pods, and don't buy engagement. Coach genuine asks.
4. DON'T EDIT RIGHT AFTER POSTING. Editing a post in the first ~hour can suppress its reach — proof the post (and the first comment) BEFORE publishing.
5. GROUND EVERY WORD IN THE REAL PRODUCT. Read .ulpi/launch/positioning.md if present, else the README, the live site (via browse), and the repo. Never invent features or metrics.
6. WIN THE GOLDEN HOUR. Early engagement (~first 60–90 min) decides reach — be present and reply to every comment fast, with substance. Bake this into PLAN.md.
7. COMPOSE THE SHARED SKILLS, DEGRADE GRACEFULLY. Invoke launch-copy (LinkedIn asset profile), launch-outreach (mode linkedin), and launch-analytics (source linkedin). If a companion isn't installed, give the one-line install (npx skills add https://github.com/ulpi-io/skills --skill <name>) and fall back to the built-in reference — never hard-fail. Run the pre-flight gate (references/preflight-gate.md) before posting, and WRITE the package to disk under .ulpi/launch/linkedin/. </EXTREMELY-IMPORTANT>

launch-linkedin

Inputs

• $request: the product to launch, or a specific LinkedIn goal (e.g. "write the post", "plan launch day"), plus optional links (live URL, demo, media).

Goal

Produce a paste-ready LinkedIn launch package under .ulpi/launch/linkedin/ (+ the shared .ulpi/launch/positioning.md):

• POST.md — the launch post: the hook (first ~2 lines), the skimmable body, the format/media plan, hashtags, and the product link as the first comment (UTM-tagged).
• PLAN.md — when to post, the golden-hour engagement runbook, team amplification, and the post-launch plan.
• CHECKLIST.md — the pre-flight gate.

Step 0: Discovery & companions

Resolve, asking only what you can't determine: the product (what it does, who for, live URL, demo/media); whether it posts from a personal profile (usually higher reach) or the company page; the user's network/team who'll engage. Note which companion skills are installed (launch-copy, launch-outreach, launch-analytics); install-hint if missing, but proceed. Success criteria: product, posting identity, and reach are known.

Step 1: Ground the product → .ulpi/launch/positioning.md

Reuse .ulpi/launch/positioning.md if present; else build it from the README, the live site (via browse), the repo, and project-context.md/.ulpi/design. Distill: one-line what-it-is, ICP, core value, top 3 differentiators, proof points, links — nothing invented. Save it as the shared source of truth. Success criteria: a brief no line will contradict.

Step 2: Mechanics, timing & targets

Load references/li-mechanics.md. Decide and record (into PLAN.md): when to post (a weekday morning window the user can attend live), realistic reach expectations, and the format that fits (text+image, native document/carousel, or native video — the higher-reach formats). Success criteria: a concrete post time, a chosen format, and an honest reach target.

Step 3: Write the post → POST.md

Hand the LinkedIn asset profile (references/li-format-rules.md + references/post-copy.md) to the launch-copy skill to draft the hook (first ~2 lines before "…see more") and the post body (hook → story/problem → what we built → proof → CTA), in a professional-but-human voice — no hype. Fallback: if launch-copy isn't installed, draft from those references. Then LinkedIn-owned: assemble POST.md with the formatted post (line breaks for skimmability), the format/media plan, 3–5 niche hashtags, the tags (people/company page), and the product link as the first comment (write it with utm_source=linkedin and a <launch> campaign placeholder — Step 6 finalizes the scheme and back-fills the exact tagged URL). Success criteria: a hook that earns the expand, link in the first comment not the body — paste-ready.

Step 4: Plan timing & engagement → PLAN.md

Load references/li-mechanics.md and references/timing-engagement.md. Write into PLAN.md: the post time, the golden-hour runbook (post → add the link as the first comment → reply to every comment with substance in the first ~60–90 min → don't edit the post → keep it alive), and the post-launch plan (repurpose, follow up). Success criteria: the user knows exactly when to post and how to run the first 90 minutes.

Step 5: Cross-promotion & team amplification → OUTREACH.md

Invoke the launch-outreach skill with compliance mode linkedin, positioning.md, and the post time — it writes .ulpi/launch/linkedin/OUTREACH.md with the genuine, compliant asks: notify your email list / other channels, ask your team/employees to engage early (employee amplification, genuine — not a pod), and a few personal shares. Fallback: if it isn't installed, draft these to .ulpi/launch/linkedin/OUTREACH.md yourself from references/antipatterns-compliance.md (engagement-allowed, bait-free). Success criteria: paste-ready, genuine engagement asks in OUTREACH.md — zero engagement-bait or pods.

Step 6: Wire measurement → tag links & events

Invoke the launch-analytics skill — channel linkedin (output dir) with utm_source linkedin (GA value) — to UTM-tag the product link (the one in the first comment) and wire signup/ activation tracking, writing .ulpi/launch/linkedin/analytics.md. Then update POST.md: replace the first-comment placeholder link with the final UTM-tagged URL so POST.md and analytics.md match. Fallback: if launch-analytics isn't installed, write a minimal .ulpi/launch/linkedin/analytics.md yourself — the UTM scheme (utm_source=linkedin&utm_medium=launch-post&utm_campaign=<launch>), the tagged link, and brief signup/activation notes — so the artifact exists either way. Success criteria: the link is attributable and signups are trackable — not just reactions.

Step 7: Pre-flight gate → CHECKLIST.md

Run references/preflight-gate.md end to end and write the result into CHECKLIST.md: hook earns the expand (counted ≤ the see-more cut), link is in the first comment (not the body), format chosen, hashtags 3–5, zero engagement-bait/pods, link UTM-tagged, post proofed (won't be edited after), post time set, golden-hour availability blocked. Fix any failing box and re-run. Success criteria: every box ticked.

Step 8: Post-launch plan → append to PLAN.md

Load references/timing-engagement.md (post-launch section): keep replying with substance through the day, repost/repurpose into other channels, follow up with new connections, and reuse the post as social proof. Success criteria: the user knows what to do after the golden hour.

Guardrails

• Keep the external link out of the post body — link in the first comment; UTM-tag it.
• Win the hook (≤ the see-more cut); never open with "I'm excited to announce…".
• Don't edit the post right after publishing; proof it first.
• Engagement asks are fine; engagement-bait, pods, and bought engagement are not.
• Ground all copy in the real product; never invent features or metrics.
• Compose the shared skills; if one is absent, install-hint + fall back — never hard-fail.
• Write the package to disk; don't leave it as ephemeral chat.

When To Load References

• references/li-mechanics.md — the algorithm, dwell time, comments-over-reactions, the link penalty, the golden hour, the edit penalty, realistic outcomes (Steps 2, 4).
• references/li-format-rules.md — char limit, the see-more hook, formats that earn reach, hashtags, the link-in-first-comment tactic (Step 3).
• references/post-copy.md — the LinkedIn asset profile passed to launch-copy, the hook + post skeleton, and the inline copy fallback (Step 3).
• references/timing-engagement.md — post timing, the golden-hour runbook, team amplification, and the post-launch section (Steps 4, 8).
• references/antipatterns-compliance.md — what LinkedIn penalizes, the engagement-bait/pod line, and the engagement-ask fallback (Steps 5, 7).
• references/preflight-gate.md — the blocking readiness gate (Step 7).

Output Contract

Write under .ulpi/launch/ and report:

1. .ulpi/launch/positioning.md — the shared grounded brief (created or reused)
2. linkedin/POST.md — the hook, the post body, the format/media plan, hashtags, the link as a UTM-tagged first comment
3. linkedin/PLAN.md — the post time, the golden-hour runbook, the post-launch plan
4. linkedin/CHECKLIST.md — the pre-flight gate result (every box ticked)
5. linkedin/OUTREACH.md — the cross-promo / team-amplification asks (from launch-outreach, mode linkedin)
6. linkedin/analytics.md — UTM map + signup/activation tracking (from launch-analytics)
7. a report of which companion skills were used vs fell back to built-ins