v2.0 to v3.0

99 added, 86 removed. Audit A to A.

---

name: vcmatch

- description: "Match the active startup profile against a VC fund. Usage: /vcmatch <VC Fund Name>. Reads STARTUP_PROFILE.md, researches the fund, and produces a structured VC Match Report with firm snapshot, thesis focus, key people, dimensional fit scoring, positive case, risks, pitch approach, and a verdict table."

+ description: "Match the active startup profile against a VC fund. Usage: /vcmatch <VC Fund Name>. Reads STARTUP_PROFILE.md, spawns 4 parallel domain research sub-agents, and produces a structured VC Match Report with 8-dimension fit scoring summing to /100, portfolio pattern analysis, deal process intel, and warm intro paths."

license: MIT

- compatibility: Requires Claude Code with WebSearch and WebFetch. Internet connection required.

- allowed-tools: WebSearch WebFetch Read Write

+ compatibility: Requires Claude Code with WebSearch, WebFetch, and Agent. Internet connection required.

+ allowed-tools: WebSearch WebFetch Read Write Agent

metadata:

author: 3Flux

- version: "2.0"

+ version: "3.0"

workflow-step: "4 — run after /vcposer scores 60+"

---

# /vcmatch — VC Match Analysis

- Match the active startup profile against a VC fund and produce a structured, scored VC Match Report.

+ Match the active startup profile against a VC fund using 4 parallel domain research sub-agents. Produces a 16-section report with an 8-dimension fit score summing to /100, portfolio pattern analysis, deal process intelligence, and warm intro paths.

## Invocation

```

/vcmatch <VC Fund Name>

```

Example: `/vcmatch a16z` or `/vcmatch Y Combinator`

## Execution Steps

- When invoked, execute these three steps in order.

+ Execute these five steps in order.

+ ---

### Step 1 — Load Startup Context

Read `STARTUP_PROFILE.md` from the current working directory. Extract and hold in context:

- Company name and one-liner

- - CEO name and profile

- - CTO name and profile

- - Core problem and solution

- - Target customer and market size

- - Key value proposition

- - Business model and pricing

- - Current stage, traction, and milestones

- - Team backgrounds and relevant credentials

- - Geography and founding date

- - Referral

- If `STARTUP_PROFILE.md` does not exist, stop and tell the user to create it first.

- ### Step 2 — Research the VC Fund

+ - CEO name and key credentials (3–5 highlights)

+ - CTO name and key credentials (3–5 highlights)

+ - Sector / category

+ - Stage (Idea / MVP / MVBP / Revenue / Growth)

+ - Geography

+ - Target customer (specific buyer, not generic)

+ - Business model

+ - Traction (exact metrics — numbers, not descriptions)

+ - Raise amount and instrument (SAFE / priced seed / TBD)

+ - Known competitors

+ - Referral contacts (name, email, relationship if listed)

- **Poser pre-check:** Before beginning research, look for a `vcposer-<fund-slug>.md` file in the current working directory (derive the slug the same way as the output filename). If it exists:

- - Read the Poser Score and verdict.

- - If score is **below 40**: stop and tell the user — "This fund scored [X]/100 on the poser check — the red flags suggest this fund may not be actively deploying capital. Investigate those findings or re-run `/vcposer <fund>` if conditions have changed before spending time on a full match analysis."

- - If score is **40–59**: continue with research but add a warning banner at the top of the output report: `⚠️ Poser Score: [X]/100 (Watch List) — this fund shows legitimacy concerns. De-prioritize unless a warm intro exists.`

- - If score is **60+** or no vcposer file exists: proceed normally.

+ Derive the fund slug: lowercase the fund name, strip punctuation and spaces (`Y Combinator` → `y-combinator`, `a16z` → `a16z`).

- Use WebSearch (and WebFetch on key pages like the fund's website, team pages, and portfolio pages) to gather:

+ If `STARTUP_PROFILE.md` does not exist, stop and tell the user: "Create `STARTUP_PROFILE.md` first — all /vcmatch research is personalized to your startup profile."

- - Investment thesis and mandate (verbatim quotes where possible)

- - Program model (accelerator, seed fund, multi-stage, etc.) and any mandatory time commitments

- - Stage focus (pre-seed / seed / Series A / growth)

- - Sector and vertical focus (deep tech, climate, enterprise SaaS, etc.)

- - Geography preference

- - Deal terms: standard equity %, check size structure, SAFE vs. priced round

- - Fund/batch size and cadence

- - Portfolio companies — especially sector analogs and recent investments in the last 2 years

- - Recent batch or portfolio composition (thematic trends, % AI, notable cluster sectors)

- - Key partners, their backgrounds, and their stated focus areas

- - Any thesis essays, blog posts, or public statements about what they invest in

- - Application/referral dynamics (open applications, referral advantage, partner assignment)

- - Any known portfolio conflicts or "too close to portfolio" sectors

+ ---

- Search queries to use:

- 1. `"<Fund Name>" investment thesis portfolio stage`

- 2. `"<Fund Name>" recent investments 2024 2025 2026`

- 3. `"<Fund Name>" partners focus sectors check size`

- 4. `site:<fund-website> thesis` (if website is known)

+ ### Step 2 — Poser Pre-Check

- ### Step 3 — Produce the VC Match Report

+ Before spawning any sub-agents, check for `vcposer-<fund-slug>.md` in the current directory.

- Output a structured markdown report with exactly these eight sections:

+ - **Score < 40:** Stop. Tell the user: "This fund scored [N]/100 on vcposer — likely a poser. Run `/vcposer <fund>` to review the evidence, or move to the next fund."

+ - **Score 40–59:** Continue, but flag for a warning banner in the final report. Tell the user: "Proceeding with caution — this fund scored [N]/100 on vcposer."

+ - **Score 60+ or file not found:** Proceed normally.

---

- ## VC Match Report: [Fund Name] × [Startup Name]

- ### Firm Snapshot

- Markdown table with these fields: Founded, Model, CEO, Key GPs, Standard Deal, Fund/Batch Size, Notable Portfolio/Alumni. Keep each cell concise.

+ ### Step 3 — Spawn 4 Parallel Research Sub-Agents

- ### Investment Thesis & Current Focus

+ Read the four agent prompt files from `references/`:

+ - `skills/vcmatch/references/agent-thesis.md`

+ - `skills/vcmatch/references/agent-portfolio.md`

+ - `skills/vcmatch/references/agent-people.md`

+ - `skills/vcmatch/references/agent-deal.md`

- Table with columns **Theme** | **Signal**. List the fund's active investment themes as rows — distinguish de facto focus (what they're actually funding right now) from stated mandate. Include verbatim partner quotes where available. Note any thesis drift from prior years.

+ Build a self-contained prompt for each sub-agent by substituting the following placeholders in the agent file text:

- ### Key People

+ | Placeholder | Value |

+ |-------------|-------|

+ | `{{FUND_NAME}}` | The fund name as entered by the user |

+ | `{{FUND_WEBSITE}}` | Best-guess domain (e.g., `a16z.com`) |

+ | `{{SECTOR}}` | Startup's sector/category from profile |

+ | `{{ADJACENT_SECTOR}}` | 1–2 closely related sectors |

+ | `{{STAGE}}` | Startup's current stage |

+ | `{{GEOGRAPHY}}` | Startup's geography |

+ | `{{ONE_LINER}}` | Startup's one-liner |

+ | `{{RAISE_AMOUNT}}` | Raise amount |

+ | `{{BUSINESS_MODEL}}` | Business model |

+ | `{{TRACTION}}` | Key traction metrics |

+ | `{{COMPETITORS}}` | Known competitors |

+ | `{{COMPETITOR_CATEGORY}}` | Generic competitor category |

+ | `{{CEO_NAME}}` | CEO name |

+ | `{{CEO_CREDENTIALS}}` | CEO key credentials |

+ | `{{CTO_NAME}}` | CTO name |

+ | `{{CTO_CREDENTIALS}}` | CTO key credentials |

+ | `{{REFERRAL_CONTACTS}}` | Referral names/orgs from profile |

+ | `{{REFERRAL_NAME}}` | Primary referral name |

+ | `{{REFERRAL_ORG}}` | Primary referral organization |

+ | `{{INSTRUMENT}}` | Raise instrument |

- Table with columns **Person** | **Role** | **Relevance to [Startup Name]**. List 3–6 decision-makers or relevant partners. After the table, add a plain-text note on HOW to approach (referral required? open application? partner assignment process?). Do not fabricate people or titles — caveat any uncertain data.

+ **Fire all 4 Agent tool calls in a single message** so they run in parallel. Each sub-agent must receive a fully self-contained prompt string — sub-agents have no access to the parent conversation context.

- ### Fit Assessment

+ Instruct each sub-agent to return only its named field blocks (as defined in the agent file). Sub-agents must not write files, score anything, or make recommendations.

- **Overall Fit Score: X/10**

+ Wait for all 4 sub-agents to complete before proceeding.

- Table with columns **Dimension** | **Score** | **Rationale**. Score each of these dimensions: Sector Fit, Stage Fit, Check Size Fit, Thesis Alignment, Program/Network Value. Scores are out of 10. Be calibrated — a 5/10 with clear reasoning is more useful than an inflated 8/10.

+ ---

- ### Why It Works

+ ### Step 4 — Synthesize and Score

- Bullet list of the strongest positive signals for pursuing this fund. Lead with concrete portfolio analogs (named companies + outcomes). Include narrative hooks, structural advantages, and timing factors that favor the startup right now.

+ Collect the returns from all 4 sub-agents. Reconcile any conflicting data points, noting the source conflict.

- ### Key Risks & Objections

+ **Score the 8 fit dimensions** based on the sub-agent evidence. Each dimension must cite specific evidence — no inflated or uncalibrated scores.

- Numbered list of material risks and mandate mismatches. Where relevant, include specific math (dilution %, valuation impact, timing lock-in). Flag structural concerns (batch schedule, mandatory program overhead, check size constraints, portfolio conflicts). Be direct — omit risks that don't materially affect this specific startup.

+ | Dimension | Max Points | Evidence Source |

+ |-----------|-----------|----------------|

+ | Sector Fit | 20 | Thesis sub-agent: Sector Clusters, De Facto Focus |

+ | Stage Fit | 20 | Thesis sub-agent: Stage Focus + Portfolio sub-agent: Entry Stage Reality |

+ | Thesis Alignment | 20 | Thesis sub-agent: Stated Mandate, De Facto Focus, Thesis Quotes |

+ | Team-Fund Fit | 10 | People sub-agent: Team-Fund Fit Signals |

+ | Traction Signal Fit | 10 | Portfolio sub-agent: Traction Bar |

+ | Check Size / Terms Fit | 10 | Deal sub-agent: Check Size Reality, Instrument Preference |

+ | Geographic Fit | 5 | Thesis sub-agent: Geography + Deal sub-agent |

+ | Network / Warm Path | 5 | People sub-agent: Warm Intro Paths, Referral Overlap |

- ### Best Approach

+ **Overall Fit Score:** Sum of all 8 dimensions. Maximum: 100.

- - **Intro strategy:** how to get a warm referral or bypass cold outreach (specific tactic for this fund)

- - **Lead with:** 3–4 specific bullets — what to put front and center in the application, pitch deck, or first meeting

- - **Avoid:** 1–2 things not to emphasize for this specific fund's current focus

+ **Recommended Action thresholds:**

+ - **75–100 → Pursue:** Strong mandate fit, viable path in. Proceed to `/vcintro`.

+ - **50–74 → Warm Up First:** Good fit signals but missing a proof point or intro. Address the gap, then return.

+ - **0–49 → No-Go:** Material mandate mismatch or insufficient traction for this fund's bar. Move to the next fund.

- ### Cold email

- Subject: [subject line]

---

- [4-5 sentence email body. Open with a specific reference to their thesis or a recent portfolio company. State what 3Flux does in one sentence. Explain why 3Flux fits their mandate using the match signals found. End with a clear, low-friction ask such as "Would a 20-min call make sense?"]

- ### LinkedIn message (≤150 chars, ready to send)

- [Reference one concrete thing — portfolio company, thesis, or recent post — and end with a soft ask]

+ ### Step 5 — Write Output and Report

- ### Gaps / things to verify

+ Read the output template from `references/output-format.md`. Populate all 16 sections using the synthesized evidence and scores from Step 4.

- ### Sources

+ Write the completed report to `vcmatch-<fund-slug>.md`.

- ### Verdict

+ The following strings must appear verbatim in the output (downstream skills vcintro, vctrack, and vcdebrief parse them by exact match):

+ - `Recommended Action:` followed by `Pursue`, `Warm Up First`, or `No-Go`

+ - `Fit Score:` followed by the numeric score

- Table with two columns — **Question** | **Answer**. Rows:

- - Should you pursue?

- - Stage fit verdict

- - Run a parallel raise?

- - Best-case outcome

- - Bottom line (one punchy sentence)

+ Then tell the user:

+ - The Fit Score and Recommended Action

+ - If **Pursue**: "Run `/vcpartner <fund> <partner name>` to profile the right GP, then `/vcintro vcmatch-<fund>.md` to draft outreach."

+ - If **Warm Up First**: "The gap is [state the specific dimension with the lowest score]. Address it, then re-run `/vcmatch`."

+ - If **No-Go**: "This fund is a poor fit — [state the primary reason]. Run `/vclist` to identify better-matched funds."

+ - "Run `/vctrack` to update your pipeline view."

---

- ## Output Format

+ ## Output Rules

- - Use markdown tables for Firm Snapshot, Key People, Fit Assessment, and Verdict

- - Cite sources inline where possible (e.g., "per their 2024 thesis post")

- - End with: `Email Subject: VC Match <fund-slug> - <fit-score>`

- `Suggested save filename: vcmatch-<fund-slug>.md`

+ - Every score in the Fit Assessment must cite specific evidence — not "seems like a good fit"

+ - The Warm Intro Path section must name specific people or companies — never generic ("ask a portfolio founder")

+ - The Cold Email must use `[Startup Name]` — never hardcode a company name

+ - The Verdict table must include a row labeled exactly `Recommended Action`

+ - If fewer than 3 of the 4 sub-agents return usable data, downgrade the Recommended Action by one tier (Pursue → Warm Up First, Warm Up First → No-Go) and note the data gap

+ - Word budget: no limit on the full report, but each individual prose section should stay under 200 words; use tables and bullets for density

+ - Do not fabricate portfolio company names, check sizes, or partner quotes — cite sources or mark as `Unknown`