draft-investor-outreach
Skill4 starsChanged 8 months agoMITView on GitHub
Draft personalized investor outreach emails and LinkedIn messages using the startup thesis, investor fit rationale, and specific personalization hooks. Saves a final ranked outreach list and exports to CSV. Triggers on: "draft outreach", "write emails", "draft messages", "step 6", "next step", "continue to draft", "write investor emails", "draft linkedin messages".
In this repository8 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- clarify-thesisSkill
- enrich-vc-contactsSkill
- export-csvSkill
- find-vcsSkill
- map-investorsSkill
- resetSkill
- score-vc-fitSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
13 rules. About 0 are common, 7 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 22
Read data/normalized/scored_funds.json.
- Distinctive: line 35
Avoid hype, generic praise, and fake familiarity
- Distinctive: line 79
No long background — get to the point immediately
- Distinctive: line 83
Do not fabricate traction, portfolio connections, or relationships
- Distinctive: line 84
Do not claim a warm intro if none exists
- Distinctive: line 85
If there is not enough personalization data for a meaningful hook, note it and draft a simpler but honest message
- Distinctive: line 86
Keep messages distinct — each should feel written for that investor
- Very specific: line 90
Write the final ranked outreach array to data/normalized/final_output.json conforming to schemas/final_output.schema.json.
- Very specific: line 96
After writing data/normalized/final_output.json, present the output:
- Very specific: line 134
After the founder approves the messages, write data/normalized/output.csv by flattening final_output.json.
- Very specific: line 149
Do not include fund_id in the CSV — it is internal only
- Distinctive: line 153
After writing the CSV, remind them:
- Very specific: line 156
All inferred emails are marked contact_tier: "inferred" — verify before sending
≈ “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
Draft Investor Outreach
Pipeline: clarify-thesis → find-vcs → map-investors → enrich-vc-contacts → score-vc-fit → draft-investor-outreach
You are step 6 of 6. Your job is to draft a tailored email and LinkedIn message for each investor in the ranked list, then save the complete output.
Tools
None. This step is reasoning + writing only.
Input
Read data/normalized/scored_funds.json. If the file does not exist, stop and ask the founder to complete score-vc-fit first.
Also read data/normalized/company_profile.json for the startup voice and thesis.
Draft outreach only for firms and contacts that the founder confirmed in step 5. Skip any firms that were removed.
Message principles
Every message must:
- Be tailored to this specific investor — not a template with a name swapped in
- Reference one real, specific reason this investor was selected
- Explain the company clearly in 1–2 sentences
- Make a simple, clear ask
- Avoid hype, generic praise, and fake familiarity
Tone
- Concise
- Confident but not arrogant
- Specific — no filler
- Founder-like — sounds like a person wrote it
- Respectful of the investor's time
Personalization rules
Good personalization hooks:
- "You invested in [portfolio company] — we're solving the same problem from the infrastructure layer"
- "Your thesis post on [topic] directly maps to what we're building"
- "You led [comparable company]'s seed — our model is similar but in [market]"
- Portfolio adjacency that is genuinely relevant, not superficial
Bad personalization:
- "I've long admired your work"
- "You're one of the top funds in [sector]"
- "I noticed you invest in B2B SaaS" (too generic)
- Anything that could apply to 50 other investors
Output format
For each investor, write:
| Field | Description |
personalization_hook | The specific angle used (1 sentence) |
email_subject | Short, specific, non-salesy |
email_body | 3–5 sentences: who we are, why this investor, traction signal, ask |
linkedin_message | Shorter version — 2–3 sentences max |
call_to_action | Clear and simple: "Happy to share a brief deck — would a 20-min call make sense?" |
optional_follow_up_angle | One angle for a follow-up if no reply in 1 week |
Email rules:
- Subject line: max 8 words, no clickbait, no ALL CAPS
- Body: 3–5 sentences. Don't over-explain. One traction signal if available.
- End with a simple, low-friction CTA
LinkedIn message rules:
- Max 300 characters
- No long background — get to the point immediately
- Still include one real reason for reaching out
Rules:
- Do not fabricate traction, portfolio connections, or relationships
- Do not claim a warm intro if none exists
- If there is not enough personalization data for a meaningful hook, note it and draft a simpler but honest message
- Keep messages distinct — each should feel written for that investor
Saving output
Write the final ranked outreach array to data/normalized/final_output.json conforming to schemas/final_output.schema.json.
Field names map directly to CSV columns — use exact field names from the schema.
Final summary to founder
After writing data/normalized/final_output.json, present the output:
Done. Here's your investor outreach list for [company_name]:
[N] personalized outreach drafts saved to data/normalized/final_output.json
Preview of top 5:
--- #1: [investor_name], [fund_name] (score: [score_total]/[score_max_possible]) ---
Subject: [outreach_subject_line]
Email:
[outreach_email_body]
LinkedIn:
[outreach_linkedin_dm]
Hook: [outreach_personalization_hook]
---
[Repeat for top 5]
Contact quality breakdown:
✓ [N] confirmed emails
~ [N] inferred emails (marked clearly in the file)
✗ [N] LinkedIn-only contacts
Next steps:
1. Review and edit messages in data/normalized/final_output.json
2. Use your email tool (Instantly, Apollo, Gmail) to send
3. For inferred emails — verify before sending or use LinkedIn as primary channel
4. For investors with no email — send via LinkedIn DM
Want any messages revised? Tell me which investor and what to change.CSV export
After the founder approves the messages, write data/normalized/output.csv by flattening final_output.json.
CSV column order:
priority_rank, fund_name, score_total, score_max_possible, investor_name, investor_title,
contact_tier, work_email, linkedin, outreach_subject_line, outreach_email_body,
outreach_linkedin_dm, outreach_personalization_hook, outreach_ctaRules:
- Wrap any field containing commas or newlines in double quotes
- Escape internal double quotes as
"" - Include a header row
- One row per investor (same as
final_output.json) - Do not include
fund_idin the CSV — it is internal only
Confirm to the founder: data/normalized/output.csv written — [N] rows.
After writing the CSV, remind them:
data/normalized/output.csvis ready to drop into Notion, Airtable, or an email tooldata/normalized/final_output.jsonhas the full structured output for re-processing- All inferred emails are marked
contact_tier: "inferred"— verify before sending - Re-run
clarify-thesisat any time to start a fresh campaign with a new fundraising profile
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.

