How to use the Brand Discovery Interview agent skill

What is an agent skill, and what does brand-discovery do?

An agent skill is a folder of instructions that an AI agent can load when a task matches the skill's description. The brand-discovery skill is built for the case where a brand is being created, repositioned, or needs a written identity reference to brief collaborators.

When the skill is active, the agent does not produce a quick list of taglines. It works through eight modules in order, asking one question at a time, probing answers with follow-up questions, and writing each module to a file. The goal is a complete master brandbook named 90_SYNTHESIS.md that designers and writers can use.

• 10_purpose-why.md: purpose and the core belief behind the brand (Sinek Golden Circle, Lencioni)
• 20_positioning.md: competitive frame and what makes the brand different (Dunford, Moore template)
• 30_audience-niche.md: a concrete portrait of the ideal client (Baker, ICP)
• 40_personality-archetype.md: personality and archetype (Mark and Pearson, J. Aaker dimensions)
• 50_voice-tone.md: voice and tone
• 60_narrative-story.md: narrative and story arc (Neumeier trueline)
• 70_founder-tension.md: founder brand versus studio brand (Enns)
• 90_SYNTHESIS.md: the master brandbook

Install the brand-discovery skill

Download the brand-discovery ZIP from the item page on this directory. Then choose one of the three install routes below, depending on how you use your agent.

For a personal setup, unzip the download into ~/.claude/skills/. To make the skill available only inside one project, unzip it into that project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into the project's .claude/skills/ folder
• Claude apps: upload the ZIP under Settings → Capabilities → Skills

How to start an interview with natural requests

You do not need a special command. The skill's description tells the agent when to use it, so a plain request about discovering or articulating a brand's identity is usually enough. Mention that the work will take several sessions, because the skill is designed for that.

Good openers name the goal and the setting. For example, you might say that you are a founder who needs a written brand identity, or that you are an agency running a discovery workshop with a client and want the interview done module by module. If you have several founders, say so at the start, since the skill then interviews each person separately before a reconciliation pass.

• Help me discover my brand identity through a structured interview. We will do it over several sessions.
• I am repositioning my consultancy. Run a brand discovery interview and save the results in the project folder.
• We have three co-founders. Interview each of us separately before we compare answers.

What happens in a session

At the start of every session, the agent checks for earlier module files and a state.json checkpoint in the brand-identity directory. If none exist, it confirms the brand name, the participants and where to save files, then begins with the first module. If earlier work exists, it reports which module is in progress and asks whether to continue or switch.

During a module, the agent paraphrases each answer and then asks one deepening question, or closes the topic when the answers stop producing new information. It asks 'Why does that matter to you?' repeatedly to find core values, and it asks for a concrete example when an answer is vague. Once a module is done, it writes a module file with a Raw section of verbatim quotes and a Synthesis section with candidate formulations and open questions.

Files the skill creates and how to resume

Answers are saved as plain files in the brand-identity directory. Each module is written to a modules folder, and a state.json file records the session name, output path, completed modules, the module in progress, the next module, participants and the last update time. This is how a later session picks up where the last one stopped.

When all modules are complete, the agent sets the master brandbook as complete in state.json. The skill's own example of the checkpoint format looks like this.

• Resume by opening a new session in the same project and asking to continue the brand interview.
• Multi-founder answers are written to founders/{participant}.md before reconciliation.
• Share 90_SYNTHESIS.md with designers or writers as the brief.

Safety, privacy and what you need

The skill needs an AI agent environment that can read and write files in the project directory. The files do not name any API keys, accounts or external software. The skill does not run scripts and does not use the network, according to the item's description.

Because answers are written as plain files in the project folder, anyone with access to that folder can read them. Avoid putting sensitive personal or client data into an interview unless the folder is appropriately protected. The skill also validates participant names and paths in multi-founder mode, accepting only letters, numbers and hyphens for participant names and rejecting path separators and '..' segments.

• Requires an agent that can read and write files in the project directory
• No API keys, accounts or external software are named
• Runs scripts: no. Uses network: no
• Store answers only where the folder access is appropriate

Tips for getting useful output

Give the interview room to run. The skill is built for multiple sessions, and rushing to the synthesis in one conversation produces shallow output. Finish each module before moving on, and answer with specific stories or numbers when the agent asks for them.

Do not skip multi-founder interviews. The skill asks that individual interviews finish before the group discusses the brand, to avoid anchoring bias. If you need to jump modules, say so; the agent will honour the request and note the skip in state.json.

• Answer with concrete examples, client stories or numbers when asked
• Use one session per module when time is short
• Keep the module files and state.json in the same folder between sessions