# Profile Builder

> >-

- **Type:** Skill
- **Install:** `agentstack add skill-stanislavnianko-product-discovery-claude-skills-profile-builder`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [stanislavnianko](https://agentstack.voostack.com/s/stanislavnianko)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [stanislavnianko](https://github.com/stanislavnianko)
- **Source:** https://github.com/stanislavnianko/product-discovery-claude-skills/tree/main/plugins/discovery-phase/skills/profile-builder

## Install

```sh
agentstack add skill-stanislavnianko-product-discovery-claude-skills-profile-builder
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Profile Builder

> **Foundation skill** of the **discovery-phase** skill pack · `foundation` group · writes `discovery-context.md`, the file every other skill in the pack reads.

Produces `discovery-context.md` — the single source of truth that every other skill in this pack reads. Without it, other skills will warn and offer to bootstrap inline (a 2–3 question mini-interview held in conversation, not written to disk). Running this skill first gives the best, most consistent results.

## Activation rule

The moment this skill triggers, **go straight to the interview**. Do not summarise the skill, do not ask permission, do not explain the structure. The user already knows what they activated.

## What this skill writes

A single file at the project root: **`discovery-context.md`**.

It contains 7 sections:
1. Client
2. Product / Initiative
3. Engagement
4. Access & Data
5. Expected deliverable
6. Constraints
7. Open questions about the engagement itself

See `./template.md` for the exact structure.

## Procedure

### Step 1 — Check for existing context
If `discovery-context.md` already exists in the project root:
- Show the user a 5-line summary of what's there.
- Ask: "Do you want to **refresh** (re-interview from scratch), **update** (only the section I name), or **keep** (just confirm and exit)?"
- Route accordingly. Default to "update" if unclear.

If the file doesn't exist, proceed to Step 2.

### Step 2 — Interview, in 4 batches
Use AskUserQuestion. Group questions to minimize round-trips.

**Batch A — Client**
- Client name and short description (1 sentence)
- Domain / industry
- Business model: B2B SaaS / B2C product / marketplace / services / internal tool / other
- Stage: pre-launch / early traction / scaling / mature / declining
- Decision-makers: who signs off on what (names/roles)
- Existing relationship: new logo / repeat client / staff aug / pre-sale prospect

**Batch B — Product / Initiative**
- What's being explored (one sentence)
- Current state of the thing: idea / spec / prototype / live but failing / live but pivoting
- Prior research / artifacts available (links, file names, or "none")
- Has the client already proposed a solution or feature list? (yes / no — and if yes, summarize it)

**Batch C — Engagement**
- Mode: pre-sale discovery sprint / paid discovery sprint / delivery-phase discovery / staff aug / internal request / other
- Timeline available: days / 1-2 weeks / 3-4 weeks / quarter / open-ended
- Runner role: solo BA / BA + PM / lead consultant / part of agency squad / internal product
- Budget context: free pre-sale / fixed-fee / time-and-materials / discovery-sprint fee / staff aug rate / unknown

**Batch D — Access, Data, Deliverable, Constraints**
- End-user access: direct / client-mediated / proxy-only / none
- Client SMEs available: list names + roles, or "none"
- Data we can request: analytics / support tickets / call recordings / past interviews / sales notes / NPS / other / none
- Tools BA has access to: list (Linear, Jira, Notion, Confluence, Mixpanel, etc.) or "TBD"
- Primary deliverable expected: proposal / SoW draft / discovery handoff / go-no-go memo / estimation / other
- Secondary deliverables (optional): list
- NDA / confidentiality status
- Languages, geographies, regulatory (GDPR / HIPAA / SOC2 / other / none)

### Step 3 — Render the file
Render answers into the template. Keep it tight: each field one to three lines. No filler prose.

Add at the bottom:
```
## Open questions about the engagement itself
- (any answers the BA marked unknown / TBD become bullets here)
```

### Step 4 — Confirm + exit
After writing, show the user a 6-bullet summary:
- Client: , , 
- Initiative: 
- Mode: , 
- Access: ; SMEs: ; Data: 
- Deliverable: 
- Open questions logged: 

Then state: **"Discovery context written. Other skills in this pack will now run. Suggest starting with: `problem-framing`, `stakeholder-mapping`, or `discovery-conductor` if you want a guided sequence."**

## Anti-patterns

- **Skipping the engagement section.** "We'll figure out the engagement later" is how proposals end up unsigned. Force completion.
- **Vague decision-makers.** "Client leadership" ≠ name. Push for at least one named human per decision.
- **Filler.** If the BA doesn't know something, write "TBD — see open questions" and add to bottom list. Don't invent.
- **Refreshing too eagerly.** If the file exists, default to "update specific section", not "rewrite everything". Old context is hard-won.

## What downstream skills consume

Every other skill in the pack reads `discovery-context.md` and runs the soft-gate pattern: if the file is missing, ask 1–3 skill-specific bootstrap questions inline, tag any unverified output field as `[ASSUMED]`, and never block. See each skill's Step 1 for its specific bootstrap questions.

Specifically:
- **problem-framing** uses Client + Initiative + prior-solution sections
- **stakeholder-mapping** uses Client decision-makers + Engagement (two-tier map)
- **research-planning, user-interviews, sme-workshops, secondary-research, support-data-analysis** all branch on Access & Data
- **estimation, proposal, sow-draft** consume Engagement + Deliverable sections
- **All deliverable skills** read Constraints (NDA, regulatory) before writing

## `_log.md` — pack-wide log schema

Every downstream skill appends one line to `./discovery/_log.md` after writing its artifact, using a single canonical schema:

```
[ | YYYY-MM-DD] key1: value1; key2: value2; key3: value3
```

Rules:
- Brackets contain the skill name and the run date (ISO format, no time).
- Body is a flat list of `key: value` pairs separated by `;`.
- Keys are skill-specific but should be short, lowercase, and stable across runs.
- One line per run. Multiple runs of the same skill produce multiple lines (newest at bottom).

This schema lets the conductor read `_log.md` to see what's been run, and lets deliverable skills (`proposal`, `sow-draft`, `discovery-handoff`, `go-nogo-memo`) recover the evidence chain without re-reading every artifact.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [stanislavnianko](https://github.com/stanislavnianko)
- **Source:** [stanislavnianko/product-discovery-claude-skills](https://github.com/stanislavnianko/product-discovery-claude-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-stanislavnianko-product-discovery-claude-skills-profile-builder
- Seller: https://agentstack.voostack.com/s/stanislavnianko
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.