Borrowing it
Nothing to install: this file belongs to prashishh/seo-geo-report-engine. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/prashishh/seo-geo-report-engine/main/.agents/skills/customer-research/SKILL.mdgit clone --depth 1 https://github.com/prashishh/seo-geo-report-engineWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/prashishh/seo-geo-report-engine/customer-research)<a href="https://agentmods.dev/skills/prashishh/seo-geo-report-engine/customer-research"><img src="https://agentmods.dev/badge/skills/prashishh/seo-geo-report-engine/customer-research/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/prashishh/seo-geo-report-engine/customer-research"><img src="https://agentmods.dev/badge/skills/prashishh/seo-geo-report-engine/customer-research.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00182 | $0.01612 |
| Opus 5 | $0.00091 | $0.00806 |
| Sonnet 5 | $0.00036 | $0.00322 |
| Haiku 4.5 | $0.00018 | $0.00161 |
Grade A, and why
customer-research scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.
customer-research
Extracts the empirical voice of the customer — the actual words people use for their
problems, the jobs they hire a product to do, and the triggers that make them buy. This is the
foundation under the GTM layer: positioning-messaging, content-brief, comparison-pages,
and geo-audit all consume its outputs. It is research-oriented, not Ahrefs-backed — there
is no Ahrefs tool for human language, so it runs on the client's own assets plus WebSearch /
WebFetch over public communities, with reasoning. (Pair with keyword-research if you also
need search-demand framing.)
Two modes (run one or both)
- Mode 1 — Assets. Parse the client's first-party material in
projects/<client>/inputs/: call transcripts, sales-call notes, survey free-text, support tickets, NPS verbatims, churn notes, win/loss. Highest-trust signal — these are real customers, but biased toward people who already engaged.Glob/Grepthe folder; read each file fully. - Mode 2 — Watering holes. When assets are thin or you need outside-in demand,
WebSearchWebFetchacross Reddit, G2, Trustpilot, Capterra, LinkedIn, Hacker News for the client's category and named competitors (fromclient.yml). Public, broad, but noisy and self-selected (loud reviewers skew negative/positive). Always capture the source URL. If a site blocks direct crawling (Reddit frequently does), fall back to theweb-researchskill's deep-research mode to get a synthesized, cited read instead of an empty result.
Methodology (PERCEIVE → ANALYZE → VALIDATE → ACT)
PERCEIVE — gather raw quotes. Resolve the project (./bin/mkt config show --project <client>); read client.yml for category, competitors, ICP, target market. Inventory
inputs/ (Mode 1). For Mode 2, run searches like site:reddit.com <category> frustrating,
<competitor> review G2, <category> "I wish" OR "the problem with", <competitor> alternative why. Collect verbatim snippets — never paraphrase at this stage. Each captured
quote keeps: exact text, source (file or URL), date, and which competitor/topic it concerns.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 97 lines · 182 tokens per session scan A 22b9f07b5266
customer-research is a skill published in the GitHub repository prashishh/seo-geo-report-engine (5 stars, last pushed 2mo ago), licensed MIT. It adds 182 tokens to every session and 1,612 once invoked, about $0.0009 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
content-amplifier
Use when the user asks to "amplify influencer content with paid media", "set up whitelisting or Spark Ads", "decide which posts to boost", "repurpose influencer content", "turn one video into multiple ads", or "build a UGC asset library"; produces (paid mode) a content-selection scorecard, a paid amplification…
email-sequence-designer
Use when the user asks to "design a welcome flow", "set up an abandoned-cart sequence", "build a light re-engagement branch inside a lifecycle flow", or "plan a cold-outbound sequence"; produces general lifecycle automation flows (welcome, cart, browse-abandon, post-purchase, in-flow re-engagement, B2B cold outbound)…
reactivation-specialist
Use when the user asks to "build a win-back campaign", "re-engage lapsed subscribers", "run a re-permission / re-consent sweep", or "sunset my dead list"; produces a closed-loop reactivation program — a lapsed-cohort definition, a staged offer ladder, a re-consent (re-permission) capture step, and a sunset-confirm /…
campaign-planner
Use when the user asks to "plan an influencer campaign", "build a campaign blueprint", "track or close a creator campaign", or "record a late campaign correction"; produces the plan and, when requested, a non-canonical evidence tracker with scoped identity, publication, reconciliation, close, and reopen receipts. Not…
product-feed-optimizer
Use when the user asks to "optimize my Shopping feed", "fix product disapprovals", "improve product titles/attributes", or "build feed-driven PMax asset groups"; audits and rewrites the Shopping/Performance Max product feed — title/description patterns, required and recommended attributes, GTIN/availability/price…
attribution-reconciler
Use when platform-reported conversions disagree with GA4/ecommerce, when you suspect Meta and Google are double-counting the same sales, or for a standing (monthly) reconciliation workbook that de-dups stacked credit against an order-ID truth set, normalizes attribution windows and currency, compares attribution…