Brand Voice Analysis & Guidelines Skill
Analyzes how a brand sounds across its website, social, and email channels, then generates a brand voice guide any writer can follow.
A pure prompt-workflow skill with no destructive defaults, credential handling, or covert exfiltration, so the risk surface is small; however it instructs the agent to fetch target sites and linked social profiles via browser tools without disclosing data flow, scoping, or any user confirmation mechanism — risks are visible but permissions and transparency are incomplete.
The ten steps are internally self-consistent with a fully specified output format, making the happy path plausible; but the referenced analyze_page.py script is not present in the evidence, its usability cannot be statically verified, and there is no failure-feedback design for unreachable sites or empty content.
Triggers are explicit (/market brand <url>) and scenarios are clear (brand voice analysis, competitor comparison), but no non-fit boundaries are declared (e.g., brands without a public site, non-English brands), there is no Chinese-language support, and core function depends on live scraping of overseas websites.
Docs are well structured with progressive disclosure and ample examples; but there is no versioning, changelog, known-limitations disclosure, or maintenance responsibility statement. MIT license is present but publisher identity is unverified; deductions are mainly for missing governance and limitation disclosure.
The BRAND-VOICE.md output spec is complete and directly usable with real marginal value over manual analysis; but all quality claims (scoring, archetype judgment) are unverified by execution — statically only structural soundness can be confirmed, so actual output quality remains unproven.
Evidence consists only of the author's own instruction document and an illustrative README terminal transcript; no tests, no CI, no third-party execution evidence — the demo output is promotional and verifiability is thin.
- The skill instructs fetching target sites and their linked social profiles without stating data use or scope; check target-site terms of service and scraping compliance before use.
- The referenced analyze_page.py script was not provided in the reviewed evidence; its behavior on failure is unknown.
- No versioning or changelog; behavior may change on update — pin to the reviewed revision.
- Core function depends on live access to overseas websites; capability for Chinese-language and non-English brands is undeclared, and no Chinese output support is provided.
- All scoring methodology is author-asserted and unverified by independent testing; generated documents require human review.
What does this skill do, and when should you use it?
This is the market-brand skill of the AI Marketing Suite for Claude Code, triggered via `/market brand <url>`. It gathers material from a brand's homepage, blog, social profiles, and emails, scores the voice along four dimensions (formal-casual, serious-playful, technical-simple, reserved-bold) with required quoted evidence, and maps the brand to one of five personality archetypes: Authority, Innovator, Friend, Rebel, or Guide. It also runs vocabulary analysis, a competitor voice comparison, and a channel-by-channel consistency audit before writing a BRAND-VOICE.md file with Do's and Don'ts, copy samples, and a messaging hierarchy. Best suited to marketing leads who need to codify brand voice for a team, freelancers, or an agency.
Collects source material in three priority tiers (homepage/about/product pages required; blog, social, email secondary; job postings, press releases, ad copy tertiary); reads web content via browser tools or the analyze_page.py script; scores each of the four voice dimensions 1-10 with 3-5 quoted examples as evidence; maps tone shifts across contexts (homepage, error pages, CTAs, email subject lines); matches the brand to one of five archetypes; analyzes frequently used words, avoided words, and signature phrases; compares voice against 2-3 competitors in a matrix (reusing /market competitors data if available); audits consistency per channel with an overall score; drafts a five-level messaging hierarchy (tagline, value props, elevator pitch, boilerplate, brand story); produces a voice chart with Do's and Don'ts plus 5-8 sample copy pieces; and saves everything to a BRAND-VOICE.md file.
- A marketing lead documenting the current voice baseline before a rebrand or identity refresh
- A founder handing copywriting to freelancers or an agency who needs a deliverable brand voice guide
- A brand with inconsistent tone across website, social, and email that needs the gaps located and fixed
- A team comparing its voice to 2-3 competitors to find unoccupied voice territory
- A newly hired copywriter who needs a document that lets them write on-brand from day one
What are this skill's strengths and limitations?
- Complete methodology: source gathering, four-dimension scoring, archetype matching, vocabulary and consistency audits cover the full voice-analysis workflow
- Evidence-mandated scoring: every dimension rating requires 3-5 quoted examples, reducing pure subjectivity
- Actionable output: BRAND-VOICE.md includes Do's and Don'ts and 5-8 diverse copy samples a new writer can model directly
- Integrates with the suite's competitor-analysis skill so the comparison section doesn't start from scratch
- Depends on Claude actually reaching the target site and social profiles; no failure-mode guidance is documented if fetching fails or content is gated
- Voice ratings are model judgment; the 1-10 scale's reliability is untested against any evaluation set
- Competitor comparison expects 2-3 rivals but doesn't explain how they're identified automatically (relies on the user or a prior /market competitors run)
- No test suite or quality benchmark; the parallel-subagent machinery described in the repo belongs to other skills — expect no audit scores or PDFs from this one
How do you install this skill?
This skill ships with the ai-marketing-claude collection. Install everything with: curl -fsSL https://raw.githubusercontent.com/zubair-trabzada/ai-marketing-claude/main/install.sh | bash, or manually: git clone https://github.com/zubair-trabzada/ai-marketing-claude.git && cd ai-marketing-claude && ./install.sh. Skills are installed under ~/.claude/skills/. Optionally pip install reportlab for PDF report support (this skill itself does not generate PDFs). Uninstall with ./uninstall.sh.
How do you use this skill?
In Claude Code, run /market brand https://example.com (or /market brand with no URL). The skill guides Claude to fetch the brand's web and social content and work through the analysis steps. If you previously ran /market competitors <url>, the competitor voice comparison reuses that data. Output is saved as BRAND-VOICE.md in the working directory. The source does not document any offline fallback if web access is unavailable.
How does this skill compare with similar options?
Within the same AI Marketing Suite, this is the voice-definition skill, complementary to /market competitors (competitive intelligence) and /market copy (copy generation): it defines 'how the brand sounds' while the copy skill writes the actual content. The source names no third-party alternatives.