managing-community-presence

Manages FlashFusion's community intelligence and audience engagement using HubSpot (MCP connected), Notion (MCP connected), and LinkedIn (manual surface — no MCP). Common Room is NOT connected; this s

SKILL.md

GitHub

Community Presence Manager

Community signal → pipeline routing, wired through HubSpot (MCP ✅) and Notion (MCP ✅). LinkedIn is a manual surface — pull signals by hand, route decisions through connected tools. Common Room is not connected; see DEPENDENCY_DECLARATION for the manual fallback path.


DEPENDENCY_DECLARATION

Tool Status Operational Path Notes
HubSpot MCP ✅ CONNECTED Create / update contacts, log activity notes, pipeline stage changes Primary CRM path
Notion MCP ✅ CONNECTED Log community intel, update engagement tracker FlashFusion HQ root: cda29e98-3f08-4c83-8cfe-4c1e560836ee
LinkedIn MANUAL SURFACE Open browser, pull engagement signals by hand, paste here No MCP / API available
Common Room ❌ NOT_CONNECTED — Manual URL: https://app.commonroom.io — log in manually; see Section 5 for what to pull

Hard rule: No step in this skill calls Common Room via MCP. If the user references Common Room data, treat it as a manual paste and route it through HubSpot and Notion steps below.


FAIL_IF

  • FAIL_IF you query Common Room via MCP — it is not connected; any data returned would be fabricated.
  • FAIL_IF you present engagement signal data as "from Common Room" without the user confirming they pulled it manually from https://app.commonroom.io.
  • FAIL_IF you infer LinkedIn engagement signals from memory or training data — signals must be explicitly provided in this session.
  • FAIL_IF you create or update a HubSpot contact without user confirmation — all write operations require explicit approval per connector permissions.
  • FAIL_IF you embed any API key, HubSpot secret, or OAuth token in any output — all HubSpot writes go through the connected MCP only.
  • FAIL_IF you batch-create HubSpot contacts without per-contact confirmation.

"Just do it" override: if the user provides explicit signals in the conversation and says to proceed, skip clarification and label all assumptions inline.


Evals

Should trigger:

  1. "Who's been engaging most with my LinkedIn posts this month?"
  2. "I want to do a community check before I start my next outreach cycle"
  3. "Pull community signals — who looks like a warm prospect?"
  4. "Common Room — what's it showing?" (will surface NOT_CONNECTED status and offer manual path)
  5. "Who should I reach out to from my newsletter audience?"

Should NOT trigger:

  1. "Help me write a LinkedIn post" → use generating-social-posts
  2. "Find new prospects for FlashFusion" → use managing-ff-pipeline
  3. "Write my next newsletter edition" → use producing-ai-newsletter

Section 1: LinkedIn Signal Pull (Manual — 10 minutes)

LinkedIn has no MCP. Pull signals by hand, then paste them into this session to continue.

What to open:

  • LinkedIn → Notifications → filter by "reacted to your post" and "commented on your post"
  • Your 2–3 most recent posts → open each, check who commented (comments > likes for intent)
  • LinkedIn → your profile → Recent activity → who followed you this week

Signal tier weights:

Signal Weight
Commented on a post High
Shared your post High
Sent a direct message High
Reacted to 3+ posts in a week Medium
Reacted once Low — monitor
Followed your profile Low — monitor

Paste format for this session:

Name: [Full Name]
Title / Company: [Title at Company]
Signals: [e.g., "commented on workflow post + reacted to 3 others this week"]
LinkedIn URL: [url]
Already in HubSpot: [yes / no / unsure]
ICP fit: [yes / no / unsure — MSP, SaaS, tech team, content creator]

Paste one or multiple entries. The workflow below consumes them.


Section 2: Routing Signals Through HubSpot (MCP ✅)

Once signals are pasted, route each person through one of four HubSpot actions.

Decision tree

Signal received →
  ├─ Person already in HubSpot?
  │     Yes → update activity note with signal detail and date
  │     No  → ICP fit? (MSP, SaaS, tech team, content creator)
  │               Yes → create contact + tag COMMUNITY_SOURCE, stage = PROSPECT
  │               No  → skip, or tag WATCH_LIST if borderline
  ├─ Peer, collaborator, or adjacent creator?
  │     → create contact + tag PARTNER_POTENTIAL
  └─ Newsletter reader showing ICP signals?
       → update contact with NEWSLETTER_ENGAGED tag; note specific content engaged

HubSpot operations (via MCP — confirm before executing)

Search before creating — always check for duplicates by name or LinkedIn URL first.

Create a new contact:

  • Required: first name, last name, LinkedIn URL (in "LinkedIn Bio URL" field)
  • Source: "Community" (not "Inbound" or "Outbound")
  • Stage: PROSPECT (ICP fit) or WATCH_LIST (monitor)
  • Optional: company, job title

Update activity note on existing contact:

[DATE] Community signal — LinkedIn
Signal type: [commented / shared / reacted / DM]
Specific content engaged: [post title or topic]
Signal weight: High / Medium / Low
Action taken: [followed back / messaged / added to pipeline sequence / none]

Section 3: Logging Community Intel to Notion (MCP ✅)

Signals that don't produce a HubSpot action (non-ICP contacts, audience trends, strategy notes) go to Notion.

Target: FlashFusion HQ root → Community Intelligence log Parent page root: cda29e98-3f08-4c83-8cfe-4c1e560836ee — navigate from here to the correct sub-page.

Log entry format:

## [DATE] Community Pulse

### LinkedIn Signals This Week
- [Name] ([Title, Company]): [Signal] → [Action taken or "none — monitor"]
- [Name]: [Signal] → [Action]

### Trends Observed
- [Recurring topic in comments]
- [Type of post driving highest engagement]

### Newsletter (manual pull from Substack)
- Open rate: [%] | Click rate: [%]
- Top-performing issue: [topic]

### Community Threshold Check
Active subscribers: [number]
Status: [ ] Pre-threshold  [ ] Approaching 1k  [ ] Ready for dedicated platform
Recommendation: [distributed strategy / beginning to evaluate / ready to act]

Section 4: Engagement Response Patterns

Response is proportional to signal weight. Community trust is the asset — every engaged reader who becomes a sales target without a genuine relationship is a trust withdrawal.

Engagement level Response
1–2 low signals No action — add to watch list if ICP fit
3–5 mixed signals Follow on LinkedIn if not already; no message yet
6–10 signals, ICP fit Direct LinkedIn message — curiosity opener, not a pitch
10+ signals, ICP fit Add to HubSpot as warm PROSPECT; reference specific content in opener
Any level, peer / collaborator Outreach for partnership or guest content conversation

DM opener template (not a sales message):

"I've noticed you've been reading the FlashFusion content — what are you working on right now?"

The opener exists to start a conversation, not to convert. Conversion comes later if fit is real.


Section 5: Common Room Manual Path (When Browser Session Open)

Common Room is not connected via MCP. If you have a live browser session at https://app.commonroom.io, pull these signals manually and paste them into this conversation:

What to pull from Common Room:

  1. Activity feed → members with 3+ signals in the last 7 days
  2. Member list → sort by engagement score descending → top 10
  3. For each: name, platform, signal type, count, LinkedIn URL if available

Paste format:

Name: [Full Name]
Platform: [LinkedIn / Newsletter / other]
Signals: [e.g., "3 newsletter opens, 2 LinkedIn interactions"]
LinkedIn URL: [if available]

Route these through Section 2 (HubSpot) and Section 3 (Notion) exactly as you would LinkedIn signals.

When Common Room connects via MCP: replace Sections 1 and 5 with a live query step, remove this manual bridge.


Section 6: Community Building Strategy

FlashFusion's community is currently distributed — newsletter subscribers, LinkedIn followers. Current strategy: newsletter as anchor, LinkedIn as discovery surface, individuals treated as individuals rather than an audience.

Threshold to evaluate a dedicated community platform (Slack / Discord / Circle):

  • Newsletter list exceeds 1,000 active subscribers
  • Consistent inbound conversation volume that benefits from group discussion rather than 1-on-1 answers
  • A defined audience segment (e.g., MSP technicians learning AI) with sufficient density to sustain peer interaction

Do not create a community platform before the audience density exists. A Slack or Discord with 12 members creates maintenance burden and signals low traction to anyone who joins.

Track the threshold check in Notion monthly (Section 3 log entry).


Self-Verification Checklist

Weekly:

  • LinkedIn signals pulled manually and pasted into session — FAIL if you skip this and infer signals from memory
  • High-signal ICP-fit contacts added or updated in HubSpot via MCP — FAIL if HubSpot is bypassed for contacts that warrant action
  • Community intel logged to Notion — FAIL if this step is skipped for three consecutive weeks
  • Warm contacts messaged within 48 hours of signal before they cool

Monthly:

  • Newsletter engagement trends reviewed (open rate, click rate) — manual pull from Substack
  • Community threshold check documented in Notion with current subscriber count
  • Common Room manual pull completed if a browser session is available

CLAIMS / COUNTEREXAMPLES / CONTRADICTIONS

CLAIMS

  • [GENERATED] Outreach referencing specific content the contact engaged with converts at higher rates than cold openers — documented principle in community-led GTM; FlashFusion-specific numbers are not yet measured.
  • [GENERATED] Newsletter engagement (clicks > opens) is a stronger intent proxy than LinkedIn follows alone, because clicks require deliberate action while opens can be triggered by email clients automatically.

COUNTEREXAMPLE

  • LinkedIn engagement can be skewed by algorithmic amplification — a post that went wide will attract commenters with no ICP fit. Weight comments on niche-topic posts (workflow breakdowns, MSP AI content) more heavily than comments on high-reach general posts.
  • Common Room open signals from the newsletter are unreliable due to auto-open behavior in email clients. Weight click and comment signals more heavily than opens in Common Room data.

CONTRADICTIONS

  • "Move high-engagement members to pipeline" conflicts with community relationship integrity. Resolve by moving to pipeline only when ICP fit is explicit AND a genuine interaction already exists. Converting every engaged reader to a sales target destroys the engagement signal source that makes this workflow valuable.

GAPS / BLINDSPOTS / UNKNOWN-UNKNOWNS

  • No Substack MCP. Newsletter engagement data (open rate, click rate, per-issue opens) must be pulled manually from Substack's dashboard. There is no live path for newsletter analytics in this skill.
  • Common Room not connected. All Common Room steps are manual bridges. When Common Room connects via MCP, Sections 1 and 5 should be rewritten to replace manual pulls with live queries.
  • LinkedIn API not available. Engagement signals require a manual pull every time this workflow runs. If LinkedIn surfaces an MCP integration, update Section 1 to automate signal ingestion.
  • HubSpot write operations require per-contact confirmation per connector permissions. Do not batch-create without user approval for each record.
  • SELF-REVIEW: Verified no Common Room MCP calls present. All HubSpot and Notion operations route through connected MCPs. LinkedIn steps are explicitly manual. DEPENDENCY_DECLARATION and FAIL_IF blocks added. Version bumped to 2.0.0 (major — gate logic and tool dependencies changed from original).