Case Study Writing

B2B case study writing with STAR framework, data visualization, and research. Covers structure, customer quotes, metrics presentation, and distribution forma...

Security Scan

The skill's instructions are broadly coherent with a case-study writing purpose, but it directs the agent to download and run an external CLI (via curl | sh) and to perform a login to a third-party service without declaring credentials — this install/login step is the main unexpected risk.

Purpose & Capability

The name/description (B2B case study writing, STAR framework, visuals, research) matches the runtime instructions: it uses inference.sh to run search assistants and a python executor to produce charts and text. Minor mismatch: the SKILL.md instructs 'infsh login' (an authentication step) but the skill declares no required credentials or primaryEnv in its metadata.

Instruction Scope

Instructions stay within the stated task (research, writing, chart generation, saving a PNG). They do direct the agent to install and run a third-party CLI and to run remote commands. The instructions do not ask the agent to read unrelated host files or to exfiltrate arbitrary local data, but they do instruct saving files to disk (e.g., results-chart.png) and running code via the third-party python executor.

Install Mechanism

Although the skill bundle itself has no install spec, the SKILL.md recommends running curl -fsSL https://cli.inference.sh | sh — a remote install script that downloads binaries from dist.inference.sh. Even with claimed checksum verification, curl|sh is higher risk than instruction-only content because it results in arbitrary code being written/executed on the host. The SKILL.md does provide a checksum URL, which mitigates but does not remove the risk; the registry metadata did not declare this external dependency.

Credentials

The skill declares no required environment variables or credentials, but it explicitly calls 'infsh login' which implies authentication and creation/storage of tokens by the external CLI. Because the skill metadata does not list this, there's a transparency gap: users may be prompted to provide credentials or the CLI may create/stash tokens locally without the skill declaring that it needs them.

Persistence & Privilege

The skill itself does not request persistent inclusion (always:false) and has no declared privileges. However, following the SKILL.md will install an external CLI that may persist binaries and authentication tokens on the system. That external persistence is outside the registry metadata and should be considered when evaluating risk.

What to consider before installing

This skill appears to do what it claims (generate case studies and charts) but the SKILL.md instructs you to install and run a third-party CLI via curl | sh and to run 'infsh login'. Before installing or running this skill: (1) Inspect the install script at https://cli.inference.sh yourself rather than piping it blindly to sh; verify SHA-256 checksums from the listed checksums.txt; (2) Understand where 'infsh login' sends your credentials and what tokens it stores locally; prefer manual installation if you want to review binaries first; (3) Run the CLI in an isolated environment (container or VM) if you need to limit blast radius; (4) If you cannot or do not want to use the remote CLI, you can still use the guidance in the SKILL.md manually (write text and generate charts locally with your own tools). If you need higher assurance, ask the publisher for a reproducible install artifact and privacy/security documentation before installing.

Create compelling B2B case studies with research and visuals via inference.sh CLI.

Quick Start

curl -fsSL https://cli.inference.sh | sh && infsh login

# Research the customer's industry
infsh app run tavily/search-assistant --input '{
  "query": "SaaS customer onboarding challenges 2024 statistics"
}'

Install note: The install script only detects your OS/architecture, downloads the matching binary from dist.inference.sh, and verifies its SHA-256 checksum. No elevated permissions or background processes. Manual install & verification available.

The STAR Framework

Every case study follows: Situation -> Task -> Action -> Result

Section	Length	Content	Purpose
Situation	100-150 words	Who the customer is, their context	Set the scene
Task	100-150 words	The specific challenge they faced	Create empathy
Action	200-300 words	What solution was implemented, how	Show your product
Result	100-200 words	Measurable outcomes, before/after	Prove value

Total: 800-1200 words. Longer loses readers. Shorter lacks credibility.

Structure Template

1. Headline (Lead with the Result)

❌ "How Company X Uses Our Product"
❌ "Company X Case Study"

✅ "How Company X Reduced Onboarding Time by 60% with [Product]"
✅ "Company X Grew Revenue 340% in 6 Months Using [Product]"

The headline should be specific, quantified, and state the outcome.

2. Snapshot Box

Place at the top for skimmers:

┌─────────────────────────────────────┐
│ Company: Acme Corp                  │
│ Industry: E-commerce                │
│ Size: 200 employees                 │
│ Challenge: Manual order processing  │
│ Result: 60% faster fulfillment      │
│ Product: [Your Product]             │
└─────────────────────────────────────┘

3. Situation

• Who is the customer (industry, size, location)
• What relevant context existed before the problem
• 1-2 sentences of company background

4. Task / Challenge

• Quantify the pain: "spending 40 hours/week on manual data entry" not "had data problems"
• Show stakes: what would happen if unsolved (lost revenue, churn, missed deadlines)
• Include a customer quote about the frustration

5. Action / Solution

• What was implemented (your product/service)
• Timeline: "deployed in 2 weeks" / "3-month rollout"
• Key decisions or configurations
• Why they chose you over alternatives (briefly)
• 2-3 specific features that addressed the challenge

6. Results

• Before/after metrics — always quantified
• Timeframe — "within 3 months" / "in the first quarter"
• Unexpected benefits beyond the original goal
• Customer quote about the outcome

Metrics That Matter

How to Present Numbers

❌ "Improved efficiency"
❌ "Saved time"
❌ "Better results"

✅ "Reduced processing time from 4 hours to 45 minutes (81% decrease)"
✅ "Increased conversion rate from 2.1% to 5.8% (176% improvement)"
✅ "Saved $240,000 annually in operational costs"

Metric Categories

Category	Examples
Time	Hours saved, time-to-completion, deployment speed
Money	Revenue increase, cost reduction, ROI
Efficiency	Throughput, error rate, automation rate
Growth	Users gained, market expansion, feature adoption
Satisfaction	NPS change, retention rate, support tickets reduced

Data Visualization

# Generate a before/after comparison chart
infsh app run infsh/python-executor --input '{
  "code": "import matplotlib.pyplot as plt\nimport matplotlib\nmatplotlib.use(\"Agg\")\n\ncategories = [\"Processing Time\", \"Error Rate\", \"Cost per Order\"]\nbefore = [4, 12, 8.50]\nafter = [0.75, 1.5, 2.10]\n\nfig, ax = plt.subplots(figsize=(10, 6))\nx = range(len(categories))\nwidth = 0.35\nax.bar([i - width/2 for i in x], before, width, label=\"Before\", color=\"#ef4444\")\nax.bar([i + width/2 for i in x], after, width, label=\"After\", color=\"#22c55e\")\nax.set_ylabel(\"Value\")\nax.set_xticks(x)\nax.set_xticklabels(categories)\nax.legend()\nax.set_title(\"Impact of Implementation\")\nplt.tight_layout()\nplt.savefig(\"results-chart.png\", dpi=150)\nprint(\"Chart saved\")"
}'

Customer Quotes

What Makes a Good Quote

❌ "We love the product." (vague, could be about anything)
❌ "It's great." (meaningless)

✅ "We went from processing 50 orders a day to 200, without adding a single person to the team."
   — Sarah Chen, VP Operations, Acme Corp

✅ "Before [Product], our team dreaded Monday mornings because of the report backlog.
    Now it's automated and they can focus on actual analysis."
   — Marcus Rodriguez, Head of Analytics, DataCo

Quote Placement

• 1 quote in the Challenge section — about the frustration/pain
• 1-2 quotes in the Results section — about the outcome/transformation
• Always attribute: full name, title, company

Quote Formatting

> "We went from processing 50 orders a day to 200, without adding anyone to the team."
>
> — Sarah Chen, VP Operations, Acme Corp

Research Support

Finding Industry Context

# Industry benchmarks
infsh app run tavily/search-assistant --input '{
  "query": "average e-commerce order processing time industry benchmark 2024"
}'

# Competitor landscape
infsh app run exa/search --input '{
  "query": "order management automation solutions market overview"
}'

# Supporting statistics
infsh app run exa/answer --input '{
  "question": "What percentage of e-commerce businesses still use manual order processing?"
}'

Distribution Formats

Format	Where	Notes
Web page	/customers/ or /case-studies/	Full version, SEO-optimized
PDF	Sales team, email attachment	Designed, downloadable, gated optional
Slide deck	Sales calls, presentations	5-8 slides, visual-heavy
One-pager	Trade shows, quick reference	Snapshot + key metrics + quote
Social post	LinkedIn, Twitter	Key stat + quote + link to full
Video	Website, YouTube	Customer interview or animated

Social Media Snippet

Headline stat + brief context + customer quote + CTA

Example:
"60% faster order processing.

Acme Corp was drowning in manual fulfillment. 4 hours per batch. 12% error rate.

After implementing [Product]: 45 minutes per batch. 1.5% errors.

'We went from 50 orders a day to 200 without adding headcount.' — Sarah Chen, VP Ops

Read the full story → [link]"

Writing Checklist

• Headline leads with the quantified result
• Snapshot box with company, industry, challenge, result at top
• Challenge is quantified, not vague
• 2-3 specific customer quotes with attribution
• Before/after metrics with timeframe
• 800-1200 words total
• Skimmable (headers, bold, bullet points)
• Customer approved the final version
• Visual: at least one chart or before/after comparison

Common Mistakes

Mistake	Problem	Fix
No specific numbers	Reads like marketing fluff	Quantify everything
All about your product	Reads like a sales pitch	Story is about the CUSTOMER
Generic quotes	No credibility	Get specific, attributed quotes
Missing the "before"	No contrast to show impact	Always show the starting point
Too long	Loses reader attention	800-1200 words max
No customer approval	Legal/relationship risk	Always get sign-off

Related Skills

npx skills add inference-sh/skills@web-search
npx skills add inference-sh/skills@prompt-engineering

Browse all apps: infsh app list