social-selling-content-generator-skill

Produce a month of net-new LinkedIn content under Tim's name that speaks to Epiphan's ICP verticals (Higher Ed, Courts, Government, Corporate AV, Healthcare, Houses of Worship, K-12), carries only proof points verified through Epiphan Knowledge `search`, and passes the Epiphan Brand voice check before Tim posts it. Output is a calendar file plus a comment strategy, not a wall of chat text. Keywords: `social selling`, `LinkedIn content`, `thought leadership`, `content calendar`, `personal brand`

<quick_start> Tools in call order:

1. Inputs — target verticals (default: all seven, weighted by ICP score), posting frequency (default 3/week), Tim's own stories and wins he wants used, any prospect list for the comment strategy.
2. Gate Q (only if a prospect list is supplied for the comment strategy) — qualify_lead per name; drop excluded and surface STALE AE LEAD rows.
3. Gate S — Epiphan Knowledge search (and search_product_catalog for any product named) for every proof point, stat, or capability a post will cite. No knowledge result → the post uses Tim's first-hand observation framed as such, never a fabricated number.
4. Gate B — get_writing_style once; check_my_copy on every post body and every comment template. Apply corrections.
5. Deliver — write the calendar to ~/Documents/linkedin-calendar/<YYYY-MM-DD>-calendar.md (or a claude.ai Artifact when running there) with each post marked [brand-checked] or [brand check skipped]. Tim copies posts into LinkedIn by hand; there is no LinkedIn write tool.
6. Sidecar + run-log line. </quick_start>

<house_rules> Customer-facing skill. Before output, run the gates in order (details: ~/.claude/skills/epiphan-house-rules-skill/reference/gates.md): S: search_product_catalog (Epiphan AI) or Epiphan Knowledge search before any product claim. B: check_my_copy on every customer-facing string. Q: qualify_lead right after the candidate pull, before tiering/enrichment/drafts. D: outbound text becomes a Gmail draft, never a send. Identifiers (owner IDs, portal, emails, seats) come from reference/identifiers.md in that skill — do not restate them here. On any gate failure follow reference/self-healing.md (retry, degrade with a flag, alert Tim, halt) and append a line to ~/.claude/skill-runs/.jsonl. </house_rules>

<success_criteria>

• 30-day calendar with 3-5 posts per week, each tagged with vertical and content pillar
• Every stat, customer outcome, or product capability in a post traces to an Epiphan Knowledge search result or is framed as Tim's first-hand observation
• Every post and comment template passed check_my_copy (or is flagged [brand check skipped])
• Comment strategy covers prospects' posts and replies on Tim's own posts; prospect list (if any) was run through qualify_lead
• Calendar file written to the path above; sidecar and run-log line written </success_criteria>

Stage 1: Positioning and pillars

Tim's position: a Sr. BDR who talks to AV, IT, and academic-technology buyers every day about unattended capture, recording reliability, and getting video to the people who need it. Posts teach from that vantage point; they do not pitch.

Content pillars (rotate across the month):

1. Vertical insight — what buyers in one vertical are wrestling with this quarter
2. Problem/Reframe — the real root cause behind a common AV failure (Challenger teach)
3. Call stories — anonymized lessons from discovery calls (no org names without permission)
4. How-To — practical capture, streaming, and CMS workflow guidance
5. Proof — verified customer outcomes and product facts from Gate S

Post types: question, story, list, contrarian, data (verified only), case (verified only).

Stage 2: Gate S — proof-point sourcing

For each pillar-5 post and any post naming a product or capability: search_product_catalog (confirms the product is current; EOL products get an acknowledgement, not a claim) then Epiphan Knowledge search for the specific fact. Record the source line in the calendar entry as Source: <knowledge title> so Tim can defend it in comments. If nothing comes back, rewrite the post around an observation ("most campus teams I talk to…") and tag it [observation, not stat].

Stage 3: Gate B — write and brand-check

get_writing_style once. Draft each post from the vertical templates below, then check_my_copy per post and per comment template. Apply corrections; keep the hook line under 12 words.

Stage 4: Calendar file

# LinkedIn Calendar — <month> — Tim Kipper, Sr. BDR, Epiphan Video

Verticals: <list + weights> · Frequency: <n>/week · Brand check: <date>

## Week 1 — Establish authority
### Day 1 (Tue) · Higher Ed · Vertical insight · [brand-checked]
Hook: <≤12 words>
<post body>
Source: <knowledge title or "observation">
Why it works: <one line>

### Day 2 (Thu) · Courts · Problem/Reframe · [brand-checked]
…
(continue through Week 4; 12-20 posts total)

## Comment strategy
…
## Engagement checklist
…

Vertical templates (ICP score in parentheses)

Each template is a frame, not a script; fill it from Gate S results and Tim's own calls.

Higher Ed (90)

• Insight: "The lecture-capture failures I hear about aren't budget problems. They're reliability problems: a classroom PC doing three jobs during finals week."
• Reframe: software-on-a-PC vs dedicated capture; the cost of one lost recording vs the cost of the box.
• How-To: getting recordings into the CMS without a person in the loop (verify the integration via Gate S before naming it).
• Audience: CIO, VP/AVP IT, Provost, Director of Academic/Instructional Technology, Dean.

Courts / Legal (85)

• Insight: "A dropped recording in a courtroom isn't an inconvenience. It's a gap in the record."
• Reframe: eight-hour unattended sessions and why a shared PC is the wrong tool.
• Audience: Clerk of Court, Court Administrator, Director of Court Administration.

Government (80)

• Insight: "Council meetings get recorded whether or not the AV person is in that day — or they don't, and someone explains why."
• Reframe: coverage that depends on who's on shift vs capture that runs itself.
• Audience: City/County Manager, IT Director/CIO, Director of Procurement.

Corporate AV (80)

• Insight: town halls, training capture, and executive comms as a reliability problem.
• Reframe: fleet management across sites (verify Epiphan Edge capabilities via Gate S).
• Audience: VP Facilities, VP IT/CIO, VP Corporate Communications, Director of Corporate Events.

Healthcare (75)

• Insight: "Recording the procedure is easy. Doing it without pulling clinical IT into every session is the hard part."
• Reframe: med-ed capture that doesn't compete with clinical systems.
• Audience: CIO, COO, Director of Medical Education, Surgical Services Director.

Houses of Worship (70)

• Insight: volunteer-run streaming and the Sunday-morning failure mode.
• Reframe: one-button operation vs a vMix rig that needs a trained operator.
• Audience: Senior Pastor, Executive Pastor, Facilities Committee Chair.

K-12 (65)

• Insight: board meetings, graduation, and athletics streaming on a district IT budget.
• Audience: Superintendent, CTO/Director of Technology Services, Business Director.

Comment strategy

When someone comments on Tim's post: thank by name → add one insight → if they fit the ICP, ask one diagnostic question ("Where do you lose the most time: capture, switching, or getting recordings out?") → if qualified, move to DM. Run qualify_lead before any DM that mentions Epiphan.

When a prospect posts: insightful addition, value-add question, or shared experience. Never pitch in a comment. Three comments per day on ICP-vertical posts beats one post per day.

Both comment frames pass check_my_copy once per month as templates.

Engagement rules

• Tue-Thu 9-11 AM in the audience's timezone; Mon/Fri afternoons second-best; no weekends
• Hook line first, one idea per post, question at the end, no links in the body, at most two emojis
• Comment on your own post first; reply to every comment within the hour
• Track impressions, engagement rate, profile views, connection requests, inbound DMs weekly; double down on the pillar that drives inbound

<self_healing> Failure = no calendar file written; partial = a stage degraded (name it: S nothing found for a post, B skipped, Q unavailable for the comment list). Ladder: retry once → degrade (post marked [observation, not stat] or [brand check skipped]) → alert Tim in the calendar header → halt only if Brand MCP and product knowledge are both unreachable. Full contract: ~/.claude/skills/epiphan-house-rules-skill/reference/self-healing.md. </self_healing>

<outcome_sidecar> At the end of every run write ~/.claude/skill-analytics/last-outcome-social-selling-content-generator-skill.json and append one line to ~/.claude/skill-runs/social-selling-content-generator-skill.jsonl:

{"ts":"<UTC ISO8601>","skill":"social-selling-content-generator-skill","version":"<frontmatter version>","variant":"default",
 "status":"success|partial|error","runtime_ms":0,
 "metrics":{"posts_generated":0,"verticals_covered":0,"proof_points_verified":0,"observation_fallbacks":0,"brand_skipped":0,"calendar_path":""},
 "error":null,"session_id":"<date>"}

status partial/error must name the failing stage in error. The SessionStart sweep ingests it. </outcome_sidecar>