agentleFS
christiankoertel-sketchchristiankoertel-sketch/adam-skills/business-readiness-assessor/SKILL.md

business-readiness-assessor

Skill0 starsChanged 8 months agoMITView on GitHub

Evaluate whether an AI agent, chatbot, or automation system can be successfully built for a specific business. Performs structured intake of business requirements and scores readiness across 5 dimensions: domain coverage, technical feasibility, data availability, risk profile, and complexity. Use this skill whenever someone wants to assess if AI is right for their business, evaluate a client for an AI project, scope an AI build, do a feasibility study for automation, or figure out what kind of AI agent a business needs. Also trigger when the user is onboarding a new client for AI services, building a proposal, or needs to determine if a business has enough data and infrastructure to support an AI deployment.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    How to evaluate: Check what relevant knowledge, experience, or reference material exists for this domain.

    line 101
  2. Distinctive:

    How to evaluate: Ask: "What's the worst thing that happens if the agent gives a wrong answer?" The answer determines the score.

    line 143
  3. Distinctive:

    Use the scores to determine the verdict:

    line 159
  4. Distinctive:

    No existing documentation or knowledge base

    line 238

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Business Readiness Assessor

Systematically evaluate whether an AI agent can be successfully built for a specific business. This skill encodes a structured intake process and 5-dimension scoring system proven on real client engagements.

When to Use This

This skill is for the critical "should we build this?" decision. Before investing time building an AI agent, chatbot, or automation system for a client (or yourself), run a readiness assessment. It catches showstoppers early — missing data, unrealistic expectations, high-risk domains — before you've burned hours on a doomed project.

The output is a clear READY / PARTIALLY_READY / NOT_READY verdict with specific scores, knowledge gaps, and risk flags.

Step 1: Business Intake

Gather the business information needed for assessment. Start by collecting the required fields, then probe for optional fields that improve assessment quality.

Required Fields (must have all 6)

FieldWhat to CollectExample
business_nameCompany or project name"CloudMetrics"
industrySector — determines which knowledge domains matter"saas", "ecommerce", "healthcare", "finance", "support", "trading", "marketing"
description1-3 sentences: what the business does and what the AI agent should do"B2B SaaS analytics platform. Agent should handle tier-1 support tickets and onboarding questions."
objectivesSpecific goals the agent should achieve (at least 1)["Reduce support ticket volume by 40%", "Automate onboarding FAQ responses"]
constraintsLimitations — must include budget (max_daily_spend_usd){"max_daily_spend_usd": 10, "no_pii_storage": true}
success_metricsMeasurable outcomes to evaluate the agent (at least 1)["First response time < 30 seconds", "Customer satisfaction > 4.0/5"]

Optional Fields (improve assessment quality)

FieldWhat It Tells YouWhy It Matters
data_sourcesWhat data the agent can accessNo data = no useful agent
toolsAPIs and platforms to integrate withIntegration complexity
brand_voiceHow the agent should communicateVoice matching difficulty
target_marketWho the agent servesAudience expectations
competitorsWho they compete withCompetitive intelligence needs
evaluation_scenariosTest cases the client cares aboutDefines "done" clearly
escalation_rulesWhen to hand off to a humanRisk management
operating_hoursWhen the agent should be active24/7 vs business hours
existing_toolsCurrent tech stackIntegration surface

Intake Interview Guide

Walk through these questions with the client or user. The goal is to fill in as many fields as possible before running the assessment.

Opening questions:

  1. "What does your business do, in a sentence or two?"
  2. "What problem would an AI agent solve for you? What task eats the most time?"
  3. "What does success look like? If the agent works perfectly in 3 months, what's different?"

Data and infrastructure: 4. "Where does your business knowledge live? (docs, website, CRM, Slack, etc.)" 5. "Do you have an existing knowledge base or FAQ? How large is it?" 6. "What tools and platforms do you use daily?"

Constraints and risks: 7. "What's your budget for AI API costs? (in dollars per day or month)" 8. "Are there things the agent absolutely must NOT do? (e.g., make financial commitments, access PII)" 9. "What industry regulations apply? (HIPAA, GDPR, financial compliance, etc.)"

Scope: 10. "Should this start small (FAQ bot) or handle complex workflows (multi-step processes)?" 11. "Who are the end users? Internal team, customers, or both?" 12. "When should the agent hand off to a human?"

Completeness Score

Calculate intake completeness as: (filled_required + filled_optional) / total_fields * 100

  • 80%+ = COMPLETE — ready for full assessment
  • 50-79% = PARTIAL — can assess but with caveats
  • <50% or missing required fields = INCOMPLETE — need more information before assessing

Step 2: Run the 5-Dimension Assessment

Score each dimension 1-5. The scores determine the overall readiness verdict.

Dimension 1: Domain Coverage (1-5)

Does the builder have enough knowledge about this industry and domain to build a useful agent?

ScoreMeaningExample
5Deep expertise, many relevant knowledge itemsSaaS support agent and you've built 10 SaaS tools
4Strong coverage, minor gaps fillableHealthcare FAQ bot, you have medical content but need compliance review
3Moderate — can support basic agent, gaps need fillingE-commerce returns bot, you understand the domain but need product-specific data
2Thin coverage, significant learning neededLegal contract analyzer, limited legal domain knowledge
1Near-zero coverage, would produce an uninformed agentNuclear safety compliance bot with no nuclear domain knowledge

How to evaluate: Check what relevant knowledge, experience, or reference material exists for this domain. If building for a client, check what content they can provide.

Dimension 2: Technical Feasibility (1-5)

Can the requested capabilities actually be built with available tools?

ScoreMeaningExample
5All capabilities achievable with standard tools (APIs, RAG, browser automation)FAQ chatbot using ChromaDB + Claude
4Mostly achievable, one capability needs custom workSupport bot + CRM integration requiring custom API connector
3Core features work, some need workaroundsMulti-language bot where translation layer adds complexity
2Significant capabilities require tools that don't exist yetReal-time video analysis agent
1Core capability is technically impossible or years awayAGI-level reasoning required for the task

How to evaluate: Map each requested capability to a concrete technical approach. If you can't describe how to build it, it's a feasibility risk.

Dimension 3: Data Availability (1-5)

Are the data sources the agent needs actually accessible?

ScoreMeaningExample
5All data specified, accessible, and in usable formatCompany wiki in markdown + API access to CRM
4Most data available, minor gaps fillableWebsite content + most docs, but some are in private Notion pages
3Some sources available, others need configuration or permissionCRM data exists but needs API setup, some docs in locked Drive folders
2Critical data exists but extraction is difficultKnowledge is in people's heads, minimal documentation
1Essential data doesn't exist or is completely inaccessibleNo written content, no API access, no documentation

How to evaluate: For each data source listed, verify: Does it exist? Can you access it? Is it in a format you can ingest?

Dimension 4: Risk Profile (1-5, where 5 = low risk)

How high-stakes are the agent's tasks? Higher stakes = more risk = lower score.

ScoreMeaningExample
5Low stakes — errors are easily corrected, no harm potentialInternal FAQ bot, content recommendation
4Minor stakes — errors cause inconvenience, not damageCustomer support triage (human reviews escalations)
3Moderate stakes — needs supervision, errors have costOrder processing automation, scheduling
2High stakes — errors could cause financial or reputational harmFinancial advice, legal document generation
1Critical stakes — errors could cause physical harm or legal liabilityMedical diagnosis, safety-critical systems

How to evaluate: Ask: "What's the worst thing that happens if the agent gives a wrong answer?" The answer determines the score.

Dimension 5: Complexity (1-5, where 5 = simple)

How complex is the agent to build and maintain?

ScoreMeaningExample
5Simple FAQ/lookup agent — one data source, one output type"Answer questions about our pricing page"
4Moderate — 2-3 data sources, some logic branching"Answer support questions from our docs and route complex ones to humans"
3Multi-tool workflow — needs integrations, conditional logic"Monitor CRM, draft responses, update tickets, escalate based on rules"
2Complex orchestration — multiple agents, real-time data, complex state"Manage multi-step sales pipeline with competitor tracking and automated follow-ups"
1Extremely complex — multi-agent system, real-time processing, custom ML"Autonomous trading system with regime detection and portfolio optimization"

Step 3: Calculate Overall Readiness

Use the scores to determine the verdict:

average_score = sum(all_5_scores) / 5
min_score = lowest individual score

if average_score >= 4.0 AND min_score >= 3:
    verdict = "READY"
elif average_score >= 2.5 AND min_score >= 2:
    verdict = "PARTIALLY_READY"
else:
    verdict = "NOT_READY"

Interpreting the verdict:

  • READY — Green light. The project has strong fundamentals across all dimensions. Proceed to build.
  • PARTIALLY_READY — Proceed with caution. Identify the weakest dimensions and create a mitigation plan before building. Common mitigations: fill knowledge gaps with targeted research, simplify scope to reduce complexity, add human-in-the-loop for high-risk areas.
  • NOT_READY — Don't build yet. One or more dimensions are critically weak. Address the blockers first: gather more data, reduce scope, or decline the project.

Step 4: Generate the Assessment Report

Structure the output as a clear, actionable report:

# Readiness Assessment: [Business Name]

## Verdict: [READY / PARTIALLY_READY / NOT_READY]
**Average Score:** [X.X/5]

## Dimension Scores
| Dimension | Score | Summary |
|---|---|---|
| Domain Coverage | X/5 | [One-line reasoning] |
| Technical Feasibility | X/5 | [One-line reasoning] |
| Data Availability | X/5 | [One-line reasoning] |
| Risk Profile | X/5 | [One-line reasoning] |
| Complexity | X/5 | [One-line reasoning] |

## Knowledge Gaps (if any)
- [Specific topic that needs to be learned or researched]

## Risk Flags (if any)
- [Specific concerns requiring human attention]

## Recommended Agent Type
[FAQ bot / Support assistant / Workflow agent / Multi-agent system]

## Recommended Next Steps
1. [First action based on the assessment]
2. [Second action]
3. [Third action]

Industry-Specific Knowledge Domains

Different industries require different knowledge coverage. Use this mapping to evaluate Domain Coverage:

IndustryKey Knowledge Domains
SaaSAgent architecture, API design, cost optimization
E-commerceAPI design, coding patterns, CRM integration
MarketingCRM, API design, cost optimization
HealthcareSecurity guardrails, API design, knowledge management
FinanceTrading/financial, security guardrails, API design
SupportCRM, knowledge management, agent architecture
TradingTrading domain, API design, cost optimization

Common Patterns and Anti-Patterns

Patterns that predict success:

  • Client has a large, organized knowledge base (docs, FAQs, courses)
  • Clear, measurable success metrics defined upfront
  • Low-to-moderate risk domain (errors don't cause harm)
  • Client willing to start with a simple scope and expand
  • Existing API access to required data sources

Red flags that predict failure:

  • "We want the AI to do everything" (undefined scope)
  • No existing documentation or knowledge base
  • High-stakes domain with no human review layer
  • Client expects 100% accuracy from day one
  • Critical data is in people's heads, not written down
  • Budget is insufficient for the required API usage

Adapting the Framework

This assessment framework works for any AI agent project, not just chatbots. Adjust the dimensions based on what you're building:

  • Chatbots: Weight Domain Coverage and Data Availability heavily — the bot is only as good as its knowledge base
  • Workflow automation: Weight Technical Feasibility and Complexity — integration complexity is the main risk
  • Decision support: Weight Risk Profile heavily — wrong recommendations can cause real damage
  • Content generation: Weight Domain Coverage and brand_voice — output quality depends on domain understanding

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.