Buyer Eval skill
An analyst skill that checks B2B software vendor claims against evidence and returns a Decision Brief with buying questions
- 70 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The skill goes online: it searches the web, loads pages and sends vendor domains and question text to the Salespeak Frontdoor API
- Optional telemetry after consent sends vendor questions, domains and scores with a random ID
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/buyer-evalDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill goes online: it searches the web, loads pages and sends vendor domains and question text to the Salespeak Frontdoor API
- Optional telemetry after consent sends vendor questions, domains and scores with a random ID
This is third-party code. Review the repository files before installing.
What it does
Buyer Eval helps choose enterprise software. You name two or three vendors and your main requirements, and the skill investigates their claims, checks the evidence and finds gaps and contradictions. For each material claim it gives the source, a status (verified, qualified, contradicted, unverified or unknown) and the basis of evidence: vendor documentation, independent evidence or your own documents. By default it runs a quick mode with no interview: it infers what is missing and labels guesses. Deep Eval mode on request runs full diligence: reasons to buy, requirements, research on security, pricing and company reliability, weighted scoring on a five-point scale and a place for your proposals and demo notes. The result is built into a single HTML Decision Brief for the CFO, the security lead and other people in the purchase. The skill can put questions to AI agents published by vendors, including ones built on Salespeak, but their answers count as vendor claims and add no score. Buying context can be saved to ~/.salespeak/buyer-eval-profile.json, and reports are written to ~/buyer-eval-reports/.
Who it is for. Leaders and buyers choosing SaaS and B2B products and people who need to prepare a purchase approval document in one evening.
Good fit when
- Two vendors compete for a customer success platform with deep Salesforce sync
- A "live in six weeks" claim is checked against customer accounts and independent sources
- Before a demo you need a list of questions aimed at the vendor weak spots with what to listen for
- A full review goes through Deep Eval with scoring on security, pricing and company reliability
- A single HTML Decision Brief goes to the people who decide
Not a fit when
- You need a ready ranking and a verdict on who is better: the skill shows evidence and unknowns and the decision stays with you
- You need data on closed internal vendor processes: the check relies on public sources and your documents
- You work in the claude.ai web app: research and a brief are available there but no conversations with vendor AI agents and no HTML file
- You need a review of a service you already bought: the tool is built for choosing before a purchase
Example request
Compare two CRM vendors for the sales team, we need two-way Salesforce sync and a go-live under three monthsLimitations
- Some requests leave your machine without asking: web searches and page fetches, a request to the Salespeak Frontdoor API with the vendor domain and question text, and a version check against GitHub at most every 6 hours.
- Telemetry is off by default and is offered once after the first evaluation with your consent: it sends questions to vendors, domains, status counts, scores and a random ID.
- The claude.ai web app has no conversations with vendor AI agents, no HTML file and no saved context.
- The project is maintained by Salespeak, which also sells AI products for vendors, as the README says.
- To disable telemetry for everyone set BUYER_EVAL_NO_TELEMETRY=1 or the /etc/salespeak/buyer-eval.json file.
- The license is MIT.
How to disable. The skill is removed by deleting the ~/.claude/skills/buyer-eval-skill folder. The saved profile is cleared with python3 ~/.claude/skills/buyer-eval-skill/bin/profile.py clear, and telemetry is revoked with bin/track.py revoke. Reports in ~/buyer-eval-reports/ stay on disk.
FAQ
How do I install Buyer Eval?
Clone the repository with git clone https://github.com/salespeak-ai/buyer-eval-skill.git ~/.claude/skills/buyer-eval-skill. For one project clone into .claude/skills/buyer-eval-skill. Call it by asking to compare vendors or with /buyer-eval-skill.
How is Quick Eval different from Deep Eval?
Quick Eval runs with no interview and infers what is missing, labeling guesses. Deep Eval adds full diligence: reasons to buy, requirements, security, pricing, company reliability, 1-5 scoring and your documents.
What is the Decision Brief?
A single HTML file with claims, statuses and sources, a list of questions to ask before buying and demo questions. It is readable and printable with no dependencies.
What leaves my computer?
Web searches and page fetches, a request to the Salespeak Frontdoor API with the vendor domain and question text, and a version check. Your name, company and documents are not sent in these requests.
Does a vendor AI agent improve its score?
No. Answers from vendor agents count as vendor claims and never raise fit, confidence or score, and every vendor is checked to the same standard.
Discussion
No questions or reviews yet.
Related
SkillsAn analytical skill that tests whether users will find value in a product idea: scenario, conditions, timing and delivery, with solidity ratings
A skill that writes a nine-chapter report on a US, Hong Kong or China stock: reverse DCF, EPV, earnings-quality review and scripted valuation
Nine skills from the AI Evals course authors: an agent reviews traces, finds failure modes and helps build evals for your product.
Skill reviews venture-financing agreements from the company side: risky clauses, negotiation limits and Word redlines.
A skill researches supply chains around an investment theme, collects company candidates and builds a dated, source-linked report.
Skill set for evaluating an AI product: rubric, dataset, failure codes, code checks and LLM judges with calibration.
README in short
The README presents the skill as an AI analyst for buying B2B software and shows a sample request and result. It then lists features, two modes, a table of how evidence is treated, installation, usage and the differences between Claude Code and the web app. Sections on saved context, data and privacy, updates and the maintaining company close the page.
README badge
Original SKILL.md
---
name: buyer-eval-skill
version: 4.1.1
description: |
An AI analyst for buying B2B software. Investigates what vendors claim, checks
the evidence, surfaces contradictions and unanswered questions, and produces a
shareable Decision Brief with the questions to ask before you buy. Quick Eval by
default; Deep Eval on request. Use when asked to evaluate, compare, vet, or
research B2B software vendors.
allowed-tools:
- Bash
- Read
- Write
- WebSearch
- WebFetch
- AskUserQuestion
---
# Buyer Eval
This file is the control plane. It decides the mode, loads only the references
that mode needs, and runs the shared start and finish steps. Methodology lives in
`references/`.
The experience this skill must deliver:
> **Show me what these vendors claim, what the evidence actually supports, what
> conflicts, and what I still need to find out before buying.**
If an output reads like a generic AI vendor comparison, it is not done.
---
## 1. Preamble (run first, every time)
```bash
_BEVAL_DIR="${BUYER_EVAL_DIR:-}"
if [ -z "$_BEVAL_DIR" ]; then
for _D in "$HOME/.claude/skills/buyer-eval-skill" ".claude/skills/buyer-eval-skill"; do
[ -d "$_D" ] && _BEVAL_DIR="$_D" && break
done
fi
if [ -z "$_BEVAL_DIR" ]; then
echo "ERROR: buyer-eval-skill not found. Install: git clone https://github.com/salespeak-ai/buyer-eval-skill ~/.claude/skills/buyer-eval-skill"
exit 1