warm-followup
Skill0 starsChanged 7 months agoMITView on GitHub
Complex follow-up workflow for warm relationships: post-interview follow-up (positive or negative), timing-triggered re-engagement ('let's talk when X'), and multi-contact thread recovery (spoke to A, referred to B, B rejected, back to A). Prompts for missing context, researches news hooks with date validation, calibrates tone for warm relationships, and creates follow-up packages with database updates.
In this repository14 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- application-trackerSkill
- applySkill
- cold-outreachSkill
- company-deep-diveSkill
- contact-populationSkill
- job-searchSkill
- meeting-prepSkill
- onboard-userSkill
- resume-tailorSkill
- session-checkpointSkill
- session-endSkill
- session-startSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. About 0 are common, 16 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈16
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 90
If critical information is missing, ASK THE USER:
- Distinctive: line 161
If they gave a timeline ("over the next year", "Q1", "by end of year"):
- Distinctive: line 168
If significant time has passed (30+ days):
- Distinctive: line 194
If all fallbacks exhausted, explicitly inform the user:
- Distinctive: line 217
No apologising for reaching out
- Distinctive: line 218
No desperation or excessive enthusiasm
- Distinctive: line 219
No mention of avoided topics
- Distinctive: line 271
If They Respond Positively (topics to discuss, topics to avoid)
- Distinctive: line 272
If No Response: Follow-Up #2 (wait 7-10 business days, different angle, same warmth)
- Distinctive: line 279
Wait time: 5-7 business days
- Distinctive: line 282
Ask: "Any updates on timing?"
- Distinctive: line 285
Wait time: match their stated trigger
- Distinctive: line 288
Avoid: mentioning the rejection or other contact
- Distinctive: line 289
Ask: 15-minute catch-up, not job discussion
- Distinctive: line 320
After drafting, STOP and evaluate:
- Distinctive: line 327
If ANY check fails → ITERATE.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Warm Follow-Up
Specialised follow-up system for warm relationships that require nuanced re-engagement beyond cold outreach.
What This Skill Does
- Relationship audit — pull all DB history (Contact, Pipeline, Interactions)
- Context mapping — understand interview outcomes, referrals, rejections, commitments
- Missing-context prompts — ask user for information not in records
- News-hook research — find a current trigger justifying outreach (with strict date validation)
- Tone calibration — warm peer conversation (not job-seeker)
- Callback references — reference specific past conversations
- Face-saving framing — handle rejections gracefully
- Database updates — new Interaction, Pipeline stage updates
Scenarios Covered
- Post-interview positive — interview went well, waiting for next steps
- Post-interview negative (pivot) — interview with B didn't work, return to A
- Timing-based re-engagement — contact said "reach out when [event]"
- Post-rejection (same contact) — single contact who rejected, but relationship worth maintaining
- Dormant warm lead — previous positive interaction, conversation went cold
Prerequisites
- Existing database records (Contact, Pipeline, or Interactions)
- Previous message files or meeting notes
- Clear understanding of relationship history
MCP tools:
mcp__nocodb__*WebSearch
Quick Start
"Follow up with Jane Smith at Acme Renewables"
"Follow up with Jane Smith — she said reach out when the regional team expands"Complete Workflow
Phase 1: Relationship audit (~5 min)
Step 1 — Pull DB records:
mcp__nocodb__search_records({
base_id: "${NOCODB_BASE_ID}",
table_name: "target_contacts",
query: "[Person Name]"
})
mcp__nocodb__search_records({
base_id: "${NOCODB_BASE_ID}",
table_name: "sales_pipeline",
query: "[Person Name]"
})Step 2 — Read existing files:
Glob("**/*[person-name]*")
Read("outreach/{user-slug}/messages/company-person.md")
Read("outreach/{user-slug}/prep/company-person-prep.md")
Read("outreach/{user-slug}/analysis/person-meeting-analysis.md")Step 3 — Map the relationship timeline:
- First contact date
- Interview date(s)
- Outcome of each interaction
- Last contact date and method
- Any commitments made by either party
Phase 2: Context analysis (~5 min)
Questions to answer:
- What was the original opportunity?
- What happened since? (interview, rejection, referral, silence)
- What trigger makes NOW the right time?
- What's the face-saving framing for both parties?
- What did they explicitly say about future contact?
If critical information is missing, ASK THE USER:
I need additional context to craft the right follow-up:
1. Outcome of last interaction with [Person]?
□ Positive — waiting for next steps
□ Referred to someone else
□ Rejected / not a fit
□ Went silent
2. Did they make any specific commitments?
(e.g., "reach out when X", "stay in touch", "will introduce you to Y")
3. Trigger for reaching out now?
□ Specific timing trigger they mentioned
□ Company news matches their stated interest
□ Enough time has passed
□ Other: ___________
4. Topics to AVOID mentioning?
(e.g., other contacts, specific interview, rejection reason)Phase 3: News-hook research (~10 min)
CRITICAL — date validation rules.
Step 1 — establish cutoff date:
CUTOFF DATE = Last Interaction Date + 1 dayStep 2 — search with date constraints:
WebSearch("[Company] news <current month> <current year>")
WebSearch("[Company] announcement <prior month> <year>")
WebSearch("[Company] [target region] <current month> <year>")
WebSearch("[Company] hiring [role] <current year>")Step 3 — extract and validate dates: For EVERY news item found:
- Extract the exact date (month + year minimum).
- Compare to cutoff date.
- Score freshness:
| News Date vs Last Contact | Score | Action |
| Within 14 days of TODAY | Ideal | Use as primary hook |
| 15-30 days old | Good | Use as hook |
| 31-60 days old | Acceptable | Use if nothing fresher |
| Before last interaction | STALE | REJECT — already discussed |
| No date found | UNUSABLE | REJECT — cannot verify |
Step 4 — document date analysis:
## News Hook Analysis
| News Item | Date Found | Last Contact | Fresh? | Usable? |Hook validation criteria (ALL must pass)
- DATE CHECK: news occurred AFTER last interaction
- FRESHNESS: ideally within 30 days of today
- RELEVANCE: connects to their stated trigger/interest
- SPECIFICITY: specific enough to reference credibly
- NOT ALREADY DISCUSSED: wasn't mentioned in previous conversations
NO valid hook found — fallback strategies (in order)
Fallback 1: their timeline reference If they gave a timeline ("over the next year", "Q1", "by end of year"):
"You mentioned [timeline] back in [month] — wanted to check in as
that window approaches..."Fallback 2: time-based acknowledgment If significant time has passed (30+ days):
"It's been [X weeks] since we spoke about [topic]. Wanted to
reconnect as [reason]..."Fallback 3: value-led (no news) Lead with insight instead of news:
"Been thinking about [challenge they mentioned]. Saw a pattern
recently that might be relevant..."Fallback 4: reuse unused insight from past prep
Read("outreach/{user-slug}/prep/company-person-prep.md")
// Look for "Extended Insights" or insights marked "not discussed".Fallback 5: industry news affecting them
WebSearch("[Industry] [target region] news <current month>")Tell the user when no hook exists
If all fallbacks exhausted, explicitly inform the user:
NO FRESH NEWS HOOK FOUND
Last interaction: [Date]
News searched: [Date range]
Result: No company news after [cutoff date]
Recommended approach:
□ Wait for news (set alert)
□ Use time-based follow-up (Fallback 2)
□ Lead with value/insight (Fallback 3)
Proceed with a fallback approach? (y/n)Phase 4: Message crafting (~15 min)
Tone calibration:
- Peer checking in (not applicant)
- Reference specific past moments
- Lead with news/trigger (not "checking in")
- Clear, low-friction ask
- No apologising for reaching out
- No desperation or excessive enthusiasm
- No mention of avoided topics
Message structure (70-90 words):
- News Hook (10-15 words) — specific recent news matching their trigger
- Callback (15-20 words) — specific moment from past conversation (their words, shared insight)
- Bridge (15-20 words) — connect news to what they said about timing (quote their exact words if possible)
- Value Add (15-20 words) — brief insight or offer related to new context
- Soft Ask (10-15 words) — low-friction next step (15-min call, quick catch-up)
Phase 5: Database updates (~5 min)
Step 1 — create Interaction record:
mcp__nocodb__insert_record({
base_id: "${NOCODB_BASE_ID}",
table_name: "interactions",
data: {
interaction_type: "Message Drafted",
interaction_date: "<ISO timestamp>",
target_contact_id: <contact_id>,
pipeline_entry_id: <pipeline_id>,
interaction_summary: "Warm follow-up drafted — [trigger/hook]",
subject_topic: "[Email subject]",
details: "Follow-up after [X days]. Trigger: [news/timing]. Hook: [callback]. Avoided: [topics].",
user_id: ${ACTIVE_USER_ID}
}
})Step 2 — update Pipeline notes:
mcp__nocodb__update_record({
base_id: "${NOCODB_BASE_ID}",
table_name: "sales_pipeline",
record_id: "<pipeline_id>",
data: {
notes: "[existing notes]\n\n[Date]: Warm follow-up initiated. Trigger: [hook]. Context: [summary]."
}
})Phase 6: File creation (~5 min)
File location: outreach/{user-slug}/messages/[company]-[person]-followup.md
Include sections:
- Relationship Context (first contact, interviews, last contact, days since, exact quoted commitments)
- Trigger for This Follow-Up (news hook, connection to their words, why now)
- Topics to Avoid
- Email Subject Line + alternatives
- Message Draft (word count)
- Callback References Used
- If They Respond Positively (topics to discuss, topics to avoid)
- If No Response: Follow-Up #2 (wait 7-10 business days, different angle, same warmth)
Scenario-Specific Guidance
Post-interview positive (waiting)
- Wait time: 5-7 business days
- Tone: confident, not anxious
- Hook: reference specific conversation moment + news
- Ask: "Any updates on timing?"
Post-interview negative (return to sponsor)
- Wait time: match their stated trigger
- Tone: fresh start, no dwelling on past
- Hook: news matching their trigger statement
- Avoid: mentioning the rejection or other contact
- Ask: 15-minute catch-up, not job discussion
Timing-trigger re-engagement
- Quote their exact words if possible
- Connect news to trigger explicitly
- Tone: "You mentioned X, and I see X is happening"
Dormant warm lead
- Acknowledge time briefly ("It's been a few months")
- Lead with value (news/insight, not relationship maintenance)
- Lower friction (shorter ask, easier commitment)
Quality Checklist
Before sending:
- Relationship history reviewed
- News hook found and validated (date check passed)
- Their exact words/commitments referenced
- Tone is warm and peer-level
- No mention of avoided topics
- Ask is low-friction and specific
- Message is 70-90 words
- Database updated
- File created with full context
Mandatory: self-evaluation before delivery
After drafting, STOP and evaluate:
Check 1: hook freshness — is the news hook dated AFTER the last interaction? Check 2: would they care? — would they think "this is old news" or "oh interesting, this just happened"? Check 3: tone test — desperate job seeker vs. peer with relevant observations? Check 4: specificity test — vague references vs. specific dates and quotes?
If ANY check fails → ITERATE. Do not deliver. Return to Phase 3 or use a fallback approach.
Reference Documentation
- Operating procedures:
operating-procedures/ - Meeting analysis examples:
outreach/{user-slug}/analysis/ - Rules:
.claude/rules/
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

