A set of Skills that turn your startup profile into a full VC fundraising workflow: pipeline research, fund matching, partner profiling, outreach drafting, meeting prep, post-meeting debrief, term sheet analysis, and strategy.
Run commands in this order for a complete fundraising campaign:
# 0. Create your STARTUP_PROFILE.md
# Use the example file with your information.
# 1. Build the strategy
/vcraise
# 2. Research the landscape
/vclist
# 3. For each Tier 1 fund β legitimacy check first:
/vcposer <fund name> # Is this fund real and active? Drop if score < 40.
# 4. For each fund that passes vcposer (score 60+):
/vcmatch <fund name> # Does their mandate fit your startup?
/vcperks <fund name> # What do they offer beyond the check?
# 5. Stress-test your pitch against the funds you're pursuing:
/vcdevil # 10 lethal questions β know what to defend
/vcangel # 10 strongest signals β know what to lead with
# 6. For each fund with a vcmatch Pursue or Warm Up verdict:
/vcpartner <fund> <partner name>
/vcintro vcmatch-<fund>.md # send outreach
/vclp vcmatch-<fund>.md # attach the one-pager
# 7. When a meeting is booked:
/vcprep vcmatch-<fund>.md
# 8. After the meeting:
/vcdebrief vcmatch-<fund>.md # signal read, follow-up email, next step
# 9. When you receive a term sheet:
/vcterm termsheet-<fund>.md # clause analysis, score, negotiation priority list
# Anytime β review your full pipeline:
/vctrackFile naming convention:
| File | Purpose |
|---|---|
STARTUP_PROFILE.md |
Your startup data β update as traction grows |
vclist.md |
Master fund pipeline |
vcposer-<fund>.md |
Poser check β fund vitality, thesis authenticity, lead capacity |
vcmatch-<fund>.md |
Deep analysis per fund |
vcperks-<fund>.md |
Perks breakdown β value beyond capital, scored by relevance |
vcpartner-<fund>-<name>.md |
Partner brief |
vcintro-<fund>.md |
Outreach drafts |
vclp-<fund>.md |
Fund-specific one-pager |
vcprep-<fund>.md |
Meeting prep |
vcdebrief-<fund>.md |
Post-meeting debrief β signals, follow-up email, next step |
vcterm-<fund>.md |
Term sheet analysis β clause scores, negotiation priority list |
vctrack.md |
Pipeline dashboard β status of all funds at a glance |
vcraise.md |
Fundraising strategy |
vcdevil.md |
Adversarial stress-test β 10 lethal questions |
vcangel.md |
Champion's brief β 10 strongest signals to lead with |
These commands are available as the vcupid Claude Code plugin. Install with a single command β works in Claude Code web or CLI:
curl -fsSL https://raw.githubusercontent.com/maxoliverbr/vcupid-plugin/main/install.sh | bashThe script clones the plugin to ~/.claude/plugins/vcupid/, registers it in ~/.claude/plugins/installed_plugins.json, and enables it in ~/.claude/settings.json. It is idempotent β safe to re-run to update.
Restart Claude Code. All /vc* commands will be available in any directory that contains a STARTUP_PROFILE.md.
Local development install
If you've cloned the repo locally, you can install from the clone directly:
cd ~/dev/vcupid-plugin && bash install.shThis registers the local clone path so edits to skill files take effect immediately without re-running the installer.
Manual installation
Register in ~/.claude/plugins/installed_plugins.json:
"vcupid@local": [{
"scope": "user",
"installPath": "/home/<you>/dev/vcupid",
"version": "1.0.0",
"installedAt": "<ISO timestamp>",
"lastUpdated": "<ISO timestamp>"
}]Enable in ~/.claude/settings.json:
"enabledPlugins": {
"vcupid@local": true
}- Keep
STARTUP_PROFILE.mdcurrent. Every new traction milestone (signed LOI, new partnership, first revenue) changes what funds you can approach and what you can say. Update it and re-run/vcraisemonthly. - Run
/vcmatchbefore spending time on outreach. A 40/100 fit score is a no-go β don't write the email. - Use Variant B (
/vcintro) over cold email whenever a warm path exists. A forwarded intro converts at 5β10x the rate of cold outreach. - The vcprep is for rehearsal, not for the room. Print it, rehearse it, then leave it behind β the meeting is a conversation, not a recital.
- Run
/vcdebriefthe same day as the meeting. Memory fades fast and the follow-up email loses impact after 24 hours. Don't wait until the next morning. - Share
/vctermoutput with your attorney, not instead of one. The term sheet analysis tells you what to fight for β your lawyer tells you how to fight for it without blowing the deal. - Run
/vctrackbefore every weekly review. If you're working 5+ funds in parallel, the pipeline table is the only way to see what's stale and what needs a nudge. - Want to add or change a skill? See Contributing below.
All commands read STARTUP_PROFILE.md from your current working directory. Create this file before running any command.
Create a markdown file in your project root with the following structure. The more detail you provide, the more specific and accurate every command output will be. Vague inputs produce generic outputs.
# Startup Profile: [Company Name]
## I. Company Overview
| Field | Detail |
|-------|--------|
| **Startup Name** | [Legal company name] |
| **One-Liner** | [One sentence: what you do and for whom] |
| **Core Problem** | [The specific problem you solve β include scale/urgency] |
| **Solution** | [What you built and how it works β be technical if relevant] |
| **Key Value Prop** | [The headline outcome you deliver: time saved, cost reduced, etc.] |
| **Category** | [e.g., Deep Tech / Infrastructure / Climate / SaaS] |
| **Location & Founded** | [City, State, Country. Month Year.] |
| **Website** | [URL] |
## II. Market & Business Model
| Field | Detail |
|-------|--------|
| **Target Customer** | [Specific buyer β not "enterprises", but "grid operators at electric cooperatives"] |
| **Market Opportunity** | [Beachhead TAM with source. Expansion path.] |
| **Business Model** | [How you charge: SaaS, value-based, usage, marketplace, etc.] |
## III. Traction & Milestones
| Field | Detail |
|-------|--------|
| **Product Stage** | [Idea / MVP / MVBP / Revenue / Growth] |
| **Product Launch** | [Target date or "Live as of [date]"] |
| **Traction** | [Be specific: design partners, LOIs, revenue, pilots, grants, gov partnerships] |
| **Referral** | [Any warm contacts who can open VC doors β Name, email, relationship] |
## IV. Team
| Role | Name | Key Highlights |
|------|------|---------------|
| **CEO** | [Name] | [3β5 most relevant credentials for investors] |
| **CTO** | [Name] | [3β5 most relevant credentials for investors] |
| [Other] | [Name] | [Credentials] |
## V. Fundraise
| Field | Detail |
|-------|--------|
| **Raise Amount** | [$XM] |
| **Instrument** | [SAFE / Priced Seed / TBD] |
| **Use of Funds** | [Top 3 line items] |
| **Milestone this raise funds** | [What you'll have accomplished when the money runs out] |Tips for a strong profile:
- Use specific numbers everywhere: "$5M DOE proposal", "38,000-member LinkedIn group", "$30.2M transmission project" β not "large DOE proposal"
- Name your referrals with contact info β the outreach commands use them
- Describe the problem with the urgency a VC would feel, not just the solution you built
- List credentials that matter to investors, not your full resume
Start here. Synthesizes your profile and any existing vcmatch/vclist files into a full strategy.
/vcraise
Reads: STARTUP_PROFILE.md + any vclist.md and vcmatch-*.md in the current directory
Saves: vcraise.md
Output sections:
- Current State β honest assessment of what you have and what's missing
- Raise Parameters β recommended amount, instrument, valuation cap, timeline
- Sequencing Strategy β who first, parallel vs. sequential, how to create urgency
- Milestone Map β which proof points unlock which tier of investors
- Month-by-Month Timeline β from today through close
- Risk Factors & Contingencies β what derails the raise and the fallback for each
- This Week β Next 3 Actions β concrete, specific, doable in 7 days
Research the VC landscape and produce a ranked list of 15β20 funds to target.
/vclist
Reads: STARTUP_PROFILE.md
Saves: vclist.md
Output: Three tiers of funds with rationale and suggested contact angle per fund:
- Tier 1 β Pursue Now: Strong fit, correct stage, warm path exists
- Tier 2 β Warm Up First: Good thesis fit, missing proof point or intro
- Tier 3 β Monitor: Stage mismatch or unclear path β revisit later
Run this before /vcmatch to know which funds are worth the deep analysis.
Pre-filter funds from your pipeline before committing to a full match analysis. Runs 8 evidence-based checks on whether a fund is actually active and writing checks β independent of whether their thesis fits your startup.
/vcposer <VC Fund Name>
Example: /vcposer "Slow Ventures" or /vcposer a16z
Reads: STARTUP_PROFILE.md (for stage/sector calibration)
Saves: vcposer-<fund>.md
Output β Poser Score (0β100) across 8 checks:
- Fund Vitality β Last investment date. >18 months silent = zombie risk.
- Fund Health β Active fund with dry powder? Key partner departures?
- Thesis Authenticity β Does their portfolio actually match their stated sector thesis?
- Lead vs. Follow β Evidence of leading rounds, not just co-investing.
- Stage Honesty β Stated stage vs. actual entry stage of recent investments.
- Check Size Reality β Stated range vs. observable deal sizes.
- Signal-to-Check Ratio β Conference/content volume vs. actual new deals.
- Founder Sentiment β Public feedback patterns β ghosting, slow decisions, reneging.
Verdict tiers:
- Legit (80β100): Active fund, real checks. Proceed to
/vcmatch. - Probable (60β79): Yellow flags β prioritize if fit is strong, clarify open questions.
- Watch List (40β59): Material poser signals. De-prioritize unless a warm intro exists.
- Likely Poser (0β39): Don't waste cycles. Move to the next fund.
vcposer β vcmatch. Poser Score answers "is this fund real and active?" Vcmatch answers "does their mandate fit our startup?" You need both. A fund can be fully legit but wrong for your stage β and a fund can claim a perfect thesis but be a zombie. Run
/vcposerfirst, then/vcmatchonly for funds that score 60+.
Deep research on a single fund against your startup profile. Spawns 4 parallel domain research sub-agents, synthesizes the results into a structured report, and scores the match across 8 dimensions summing to /100.
/vcmatch <VC Fund Name>
Example: /vcmatch a16z or /vcmatch "Lowercarbon Capital"
Reads: STARTUP_PROFILE.md
Saves: vcmatch-<fund>.md
How it works: Four sub-agents run in parallel β Thesis & Mandate, Portfolio Intelligence, People & Network, and Deal & Process β each with domain-specific search queries. The main agent synthesizes all four returns and scores the fit.
Output sections:
- Firm Snapshot β founded, model, key GPs, standard deal, fund size, notable portfolio
- Investment Thesis & Current Focus β table with Theme / Signal / Verbatim Quote
- Portfolio Pattern Analysis β what the last 24 months of investments reveal about actual deployment priorities
- Key People β partner roster with background, sector focus, and access dynamics
- Fit Assessment β 8-dimension score table summing to /100
- Why It Works β positive signals with named portfolio analogs
- Key Risks & Objections β material mismatches with specific math
- Deal Process Intel β decision timeline, diligence style, known pass triggers
- Best Approach β intro strategy, lead-with bullets, avoid list
- Warm Intro Path β named intermediaries and specific actions, ranked by likelihood
- Diligence Questions to Prepare β 5β8 fund-specific questions
- Cold Email β subject + 4-sentence body with fund-specific hook
- LinkedIn Message β β€150 chars
- Gaps / Things to Verify
- Sources
- Verdict β table including Recommended Action
Fit Score thresholds: 75β100 β Pursue Β· 50β74 β Warm Up First Β· 0β49 β No-Go
Run /vcmatch before /vcpartner, /vcintro, /vclp, or /vcprep.
Understand what a fund delivers beyond the check before you sign. Researches the full value a fund delivers beyond the check β credits and services, but also brand signal, strategic expertise, media reach, policy access, and follow-on capital pathway β scored by relevance to your startup.
/vcperks <VC Fund Name>
Example: /vcperks a16z or /vcperks "First Round Capital"
Reads: STARTUP_PROFILE.md (for stage/sector calibration)
Saves: vcperks-<fund>.md
Part A β Financial Value (dollar estimates, [Confirmed]/[Reported] labels):
- Cloud & Infrastructure β AWS/GCP/Azure credits, API platform discounts
- Legal & Finance β discounted counsel, accounting, cap table tools (Carta, Pulley)
- Talent & Recruiting β job boards, ATS discounts, executive recruiting relationships
- Go-to-Market β PR/comms support, CRM and sales tool discounts
- Network & Introductions β portfolio co. intros, LP relationships, co-investor paths
- Operational Support β office space, co-working, founder summits
- Formal Programs β structured accelerator tracks, office hours, mentorship
Part B β Non-Financial Value (impact ratings, qualitative assessment):
- Brand & Signal Value β what the fund's name does for customer credibility, talent attraction, and future round momentum
- Strategic Guidance & Pattern Recognition β named GP expertise, comparable investments they've led, operating partners
- Content & Media Amplification β publishing access, newsletter reach, event speaking, social amplification
- Policy & Regulatory Access β government relations team, compliance guidance, agency relationships
- Follow-on Capital Pathway β pro rata appetite, reserve pool, bridge capacity, Series A syndication partners, fund lifecycle
Closes with a Gaps section (benchmarked against named peer funds) and a Questions to Ask the Fund Directly section from unresolved research.
For most seed-stage companies, the non-financial value β brand signal, strategic depth, follow-on pathway β is worth more over a 5-year relationship than the credits.
/vcperkscovers both.
Know who's in the room before you walk in. Research a specific GP's personal thesis, deals they've led, public writing, and how they map to your startup.
/vcpartner <fund> <partner name>
Example: /vcpartner a16z "Erin Price-Wright"
Reads: STARTUP_PROFILE.md
Saves: vcpartner-<fund>-<lastname>.md
Output sections:
- Background β career path and pre-VC experience
- Personal Thesis β what they personally care about, with direct quotes
- Deals They've Led β companies, stage, sector, why they backed them
- Public Writing & Key Quotes β specific essays and clips relevant to your startup
- Personal Touchpoints β conferences, topics they post about, shared background
- Fit Assessment β how this partner specifically maps to your team and problem
- Conversation Starters β 3 specific, sourced things to bring up in the meeting
Turn a vcmatch report into ready-to-send outreach. Three variants for every situation.
/vcintro <vcmatch-report.md>
Example: /vcintro vcmatch-a16z.md
Reads: STARTUP_PROFILE.md + vcmatch report
Saves: vcintro-<fund>.md
Output β three variants:
- Variant A β Cold Email: 4-sentence direct email to the target partner. Subject line references their thesis.
- Variant B β Warm Intro Request: A note to your referral contact (from your profile), ready to forward to the partner.
- Variant C β LinkedIn DM: 3 sentences. No deck link. Opens with the fund-specific hook.
Each variant uses the verbatim opening hook from the vcmatch report β never generic. The skill recommends which variant to use based on your available warm paths.
β οΈ If the vcmatch Recommended Action is No-Go, this command will stop and tell you β don't send outreach to a no-go fund.
A one-page executive summary written entirely in the fund's language and framed around their thesis. Not a generic company description.
/vclp <vcmatch-report.md>
Example: /vclp vcmatch-a16z.md
Reads: STARTUP_PROFILE.md + vcmatch report
Saves: vclp-<fund>.md
Output sections (β€500 words total):
- Headline β one-liner reframed in the fund's thesis language
- The Problem β uses the fund's own published framing
- The Solution β mapped to their portfolio language
- Traction β only the proof points that scored highest for this fund
- Market β sized in terms this fund cares about
- Team β only the credentials that matter to this fund
- The Ask β amount + milestone unlocked + timeline
- Why Now β fund-specific urgency tied to their thesis
β οΈ Each one-pager is fund-specific. Do not send to other funds without running/vcmatch <fund>first.
Attach this to Variant A or B of /vcintro.
Full prep for a pitch meeting: timed agenda, verbatim talking points per segment, preemptive answers to every diligence question from the vcmatch report, red flag rebuttals, and the exact ask.
/vcprep <vcmatch-report.md>
Example: /vcprep vcmatch-a16z.md
Reads: STARTUP_PROFILE.md + vcmatch report
Saves: vcprep-<fund>.md
Output:
- 15-minute timed agenda (8 segments)
- Segment scripts: what to say, what to show, what NOT to say β for each segment
- Preemptive answers to every diligence question from the vcmatch report (with "if they push" follow-ups)
- Red flag rebuttals β one-liner responses for each vcmatch red flag
- The Ask β exact amount, 90-day plan, leave-behind action
- Meeting discipline β 3 rules for controlling the room
Before you pitch anyone, get destroyed first. A cocky, arrogant VC partner who's seen a thousand pitches like yours tears your startup apart with the 10 hardest questions he can find in your profile.
/vcdevil
Reads: STARTUP_PROFILE.md
Saves: vcdevil.md
Output: 10 lethal questions derived from your actual profile β not generic diligence, but the specific traps your startup walks into. Each question includes:
- A setup line in character (bored, contemptuous, or mocking)
- The question itself, phrased to sting
- Why this kills you β the exact vulnerability it exposes
- What a real answer looks like β the framework for a rebuttal that shuts him up
Run this before any outreach. The questions you can't answer in 30 seconds are the gaps to close first.
The other side of /vcdevil. A seasoned VC partner who champions deals in the partner meeting surfaces the 10 strongest signals in your profile β not cheerleading, but pattern recognition. The things that make skeptics go quiet and interested.
/vcangel
Reads: STARTUP_PROFILE.md
Saves: vcangel.md
Output: 10 insights derived from your actual profile β the specific assets most investors would overlook or underweight. Each insight includes:
- A setup line in character (warm, perceptive, strategically excited)
- The insight itself, phrased as the angel would say it in the partner meeting
- Why this wins the room β the specific strength it names, with evidence from your profile
- How to amplify this β a coaching prompt for making the point land harder
Run this alongside /vcdevil. Together they give you full-spectrum pitch prep: know what to defend, know what to lead with.
Turn your raw meeting memory into a structured debrief. Produces a signal read, an open diligence queue, a send-ready follow-up email, and a clear recommended next step.
/vcdebrief <vcmatch-report.md>
Example: /vcdebrief vcmatch-a16z.md
Reads: STARTUP_PROFILE.md + vcmatch report (+ asks you for meeting notes)
Saves: vcdebrief-<fund>.md
Output sections:
- Meeting Summary β attendees, duration, format
- Signal Read β green (interest), yellow (hesitation), red (pushback) signals from the meeting
- Diligence Prediction vs. Reality β which vcmatch questions came up, which didn't, what surprised you
- Open Diligence Queue β every item raised in the meeting with a draft answer and deadline
- Pitch Landing Assessment β what from the vcmatch angle resonated vs. missed
- Follow-Up Email β send-ready, subject line and body, references something specific from the meeting
- Recommended Next Step β one of: Request Second Meeting / Send Deck / Send Data Room / Give Space / Drop Fund
- Updated Fit Signal β revised read on interest level vs. the vcmatch prediction
Run this within 24 hours of every meeting. The follow-up email degrades sharply after that.
Benchmark every clause in a VC term sheet against current market norms, flag founder-unfriendly terms, and produce a ranked negotiation priority list with suggested pushback language.
/vcterm <termsheet-file.md>
Example: /vcterm termsheet-a16z.md
Reads: STARTUP_PROFILE.md + term sheet file
Saves: vcterm-<fund>.md
Output sections:
- Term Sheet at a Glance β instrument, amount, cap, discount, key dates
- Founder-Friendliness Score β /100 with breakdown (β15 per red flag, β5 per yellow flag)
- Clause Analysis Table β each term vs. market norm with π’π‘π΄ verdict
- Red Flag Deep Dives β real-world impact for each π΄ clause
- Negotiation Priority List β ranked by founder harm, with suggested pushback language and likelihood of success
- Questions to Ask Before Signing β 5β8 diligence questions for the fund or your attorney
Verdict tiers:
- 85β100 β Founder-Friendly: Standard or better. Sign with standard review.
- 65β84 β Acceptable: Negotiate the reds.
- 45β64 β Investor-Favorable: Push back on all reds, escalate yellows.
- 0β44 β Aggressive: Get a second legal opinion before proceeding.
β οΈ This skill is informational only β it does not constitute legal advice. Share the output with your attorney before responding to the fund.
Scan all vcupid output files in the current directory and generate a single status table showing every fund's stage, scores, and next recommended action.
/vctrack
No argument needed. Reads from the current working directory.
Saves: vctrack.md
Output:
- Summary counts (total, active, no-go, stale)
- Pipeline table: Fund | Poser Score | Fit Score | Recommended Action | Stage Reached | Next Step
- Urgent Actions list β funds with missing follow-through flagged with the specific command to run
Run this any time to get a bird's-eye view of where your pipeline stands. Especially useful before investor update calls or weekly reviews.
Contributions are welcome β bug fixes, clearer skill instructions, new commands, and documentation improvements.
We seek help testing the skills in other tools (Opencode, Codex, Cursor...) and llms.
- Fork and clone the repo:
git clone https://github.com/<your-username>/vcupid-plugin.git cd vcupid-plugin
- Install from your local clone (registers this path with Claude Code):
bash install.sh
- Restart Claude Code. Skill changes in
skills/are picked up from the registered install path; you do not need to re-run the installer after every edit.
To validate skill structure (optional, requires skills-ref):
skills-ref validate skills/vcmatch/install.sh runs the same validation automatically when skills-ref is on your PATH.
| Path | Purpose |
|---|---|
skills/<name>/SKILL.md |
Skill definition β frontmatter, invocation, execution steps, output rules |
skills/<name>/references/ |
Optional templates (e.g. output-format.md) linked from the skill |
assets/ |
Shared assets (logo, STARTUP_PROFILE_TEMPLATE.md) |
.claude-plugin/plugin.json |
Plugin metadata for Claude Code |
install.sh |
Registers the plugin in ~/.claude/plugins/ and enables it in settings |
Claude Code discovers skills from every skills/*/SKILL.md file. There is no separate manifest to update when you add a folder.
- Open
skills/<name>/SKILL.md. - Keep the YAML frontmatter consistent with sibling skills (
name,description,license,compatibility,allowed-tools,metadata). - If the skill references an output template, update
skills/<name>/references/output-format.mdin the same change when the on-disk report format changes. - Smoke-test in Claude Code: create a
STARTUP_PROFILE.mdin a test directory, run the/vc*command, and confirm the saved markdown matches the skillβs rules.
- Create
skills/<command-name>/SKILL.md(command name should match the folder, e.g.skills/vcmatch/β/vcmatch). - Follow the structure used by existing skills:
- Frontmatter with a one-line
descriptionthat includes usage (Usage: /vcfoo ...) - Invocation β syntax and examples
- Execution Steps β ordered steps the agent must follow
- Output Format / Output Rules β what to write and where to save (
vcfoo-<fund>.md, etc.)
- Frontmatter with a one-line
- Add optional
references/output-format.mdif the report has a fixed section layout. - Document the command in this README under Commands and, if it fits the fundraising flow, in Full Fundraising Workflow.
- Run
skills-ref validate skills/<command-name>/if available.
- Branch from
mainwith a focused change (one skill or one doc area per PR when possible). - Describe what you changed and how you tested it (which
/vc*command you ran, and whether output looked correct). - Open a PR against
mainon GitHub. Issues and ideas are also welcome via GitHub Issues.
By contributing, you agree that your contributions are licensed under the same terms as the project (LICENSE).
