Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

goal-to-prototype · tonyfadel23 | DecimalAI

tonyfadel23/goal-to-prototype

SkillScore64Maintained

Transform a high-level goal into a validated prototype with market research, multi-angle analysis, and iterative refinement. Use when the user says "build me a prototype", "turn this idea into reality", "goal to prototype", "validate this concept", or "research and build".

researchby tonyfadel23

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Product Management pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/tonyfadel23-goal-to-prototype/SKILL.md

decimalai skills pull tonyfadel23-goal-to-prototype --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/tonyfadel23-goal-to-prototype/SKILL.md https://app.decimal.ai/s/tonyfadel23-goal-to-prototype/SKILL.md

Raw file: /s/tonyfadel23-goal-to-prototype/SKILL.md ↗ · machine-readable summary at /s/tonyfadel23-goal-to-prototype.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 54d ago

—+67% tokens0% turns64% of whole cases pass with skill (14/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-11	✗→✓	▲ Improved	44%	0%
case-13	✗→✓	▲ Improved	38%	0%
case-15	✗→✓	▲ Improved	0%	0%
case-18	✗→✓	▲ Improved	44%	0%
case-19	✗→✓	▲ Improved	80%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

tonyfadel23 / skill-inspector

examples/output/goal-to-prototype/SKILL.md

View on GitHub ↗

License MITImported 8/1/2026Version v1

Skill contents

SKILL.md

goal-to-prototype

Step 0 — Context Loading

Load user personas and company profile

Read and load the following context files:

Load references/personas.md
Load references/company-profile.md

Load market landscape and competitive intel

Read and load the following context files:

Load references/market-landscape.md

Also fetch data from:

Fetch https://api.example.com/market-data

Parse the user's goal statement into structured requirements: target audience, problem space, desired outcome, success metrics, and constraints

Step 1 — Data Acquisition

Launch parallel research agents to gather data simultaneously (parallel)

Run the following tasks simultaneously in parallel:

Web Research Agent

Search the web for recent articles, papers, and discussions related to the problem space. Fetch at least 10 high-quality sources. Use WebSearch and WebFetch tools.

Competitor Analysis Agent

Identify top 5-10 competitors in the space. For each, analyze: product offering, pricing, strengths, weaknesses, market share. Use WebSearch to find competitor data.

User Research Agent

Cross-reference the loaded personas against the problem space. Identify pain points, unmet needs, willingness to pay, and adoption barriers. Load references/personas.md for context.

Market Trend Agent

Analyze market trends, growth trajectories, regulatory changes, and technology shifts that could impact the opportunity. Use WebSearch for current data.

Step 2 — Signal Evaluation

Synthesize all research outputs into a unified intelligence brief

Read all outputs and synthesize them with weighted importance.

Compute signal quality metrics from the merged research:

market_signal: strength of market opportunity (0-1)
competitive_gap: size of unserved niche (0-1)
user_demand: evidence of user pull (0-1)
feasibility: technical buildability (0-1)
confidence: data quality and coverage (0-1)

Compute signal quality metrics from the merged research:

market_signal: strength of market opportunity (0-1)
competitive_gap: size of unserved niche (0-1)
user_demand: evidence of user pull (0-1)
feasibility: technical buildability (0-1)
confidence: data quality and coverage (0-1)

Verify signal strength before investing in deep analysis

Check whether the following criteria are met before proceeding:

confidence: >= 0.6
market_signal: >= 0.5
user_demand: >= 0.4

If any criterion fails, retry up to 3 times. On each retry, analyze what went wrong and adjust the approach.

Step 3 — Multi-Angle Analysis

Analyze the opportunity from multiple angles in parallel (parallel)

Run the following tasks simultaneously in parallel:

Technical Feasibility Assessment

Evaluate: tech stack options, build complexity, infrastructure needs, MVP scope, time-to-prototype. Score feasibility 0-1.

Market Viability Assessment

Evaluate: TAM/SAM/SOM, go-to-market strategy, pricing model, distribution channels, competitive moat. Score viability 0-1.

User Desirability Assessment

Evaluate: user journey, pain-point alignment, switching cost, value proposition clarity, retention hooks. Score desirability 0-1.

Step 4 — Synthesis

Converge all perspective assessments into a unified opportunity score

Read all outputs and synthesize them with weighted importance.

Gate on overall opportunity quality before prototyping

Check whether the following criteria are met before proceeding:

opportunity_score: >= 0.7
feasibility: >= 0.6

If criteria pass → proceed to next step. If criteria fail → "deep-research-loop" → route to remediation.

Generate a structured product brief with: problem statement, target user, value proposition, key features, success metrics, risks, and mitigations

Apply RALPH methodology to strengthen weak signals

Apply the RALPH methodology (up to 5 iterations):

Reflect on which signals are weak and why
Analyze gaps in the research — what data is missing?
Learn from competitor approaches — what are they doing right?
Plan targeted research to fill specific gaps
Hypothesize an improved positioning and re-score

Exit when all criteria are met:

opportunity_score: >= 0.7
confidence: >= 0.8

Validate the output against criteria after each iteration. Fix failures before proceeding.

Step 5 — Prototype Generation

Generate the Product Requirements Document

Save to the following output files:

Write to output/prd.md

Generate low-fidelity wireframes for the core user flows

Launch a general-purpose sub-agent to run in parallel with the following task:

> Generate low-fidelity wireframes for the core user flows

Tools available: Write

Load the following context before starting:

Read output/prd.md

Generate the implementation plan with architecture, tech stack, milestones, and sprint breakdown

Launch a general-purpose sub-agent to run in parallel with the following task:

> Generate the implementation plan with architecture, tech stack, milestones, and sprint breakdown

Tools available: Write, Read

Load the following context before starting:

Read output/prd.md

Step 6 — Final Review

Iteratively improve prototype quality

Apply the RALPH methodology (up to 3 iterations):

Reflect on prototype completeness — are all requirements covered?
Analyze coherence — do PRD, wireframes, and plan align?
Learn from quality gaps — what's missing or contradictory?
Plan specific fixes for each gap
Hypothesize the improved version and verify

Exit when all criteria are met:

completeness: >= 0.9
coherence: >= 0.9

Validate the output against criteria after each iteration. Fix failures before proceeding.

Present the complete prototype package to the user for review and approval

Ask the user to confirm before proceeding:

> Present the complete prototype package to the user for review and approval

Save the final deliverables

Save to the following output files:

Write to output/prd.md
Write to output/wireframes.md
Write to output/implementation-plan.md
Write to output/research-brief.md

Troubleshooting

If something goes wrong during execution, check these common issues:

Tool access issues: Ensure the following tools are available: Read, Write

Missing files: Ensure these files exist before running:

output/implementation-plan.md
output/prd.md
output/research-brief.md
output/wireframes.md
references/company-profile.md
references/market-landscape.md
references/personas.md

References

output/implementation-plan.md
output/prd.md
output/research-brief.md
output/wireframes.md
references/company-profile.md
references/market-landscape.md
references/personas.md

Illustrative Want to see what tonyfadel23/goal-to-prototype would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/4/2026, 09:39 UTC · 22 cases

Measured on 19 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+18%

64% vs 45%

Agent turns

1.0 vs 1.0

Tokens

+67%

2,909.63 vs 1,739.58

fail → pass · 5pass → pass · 9fail → fail · 7pass → fail · 1

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 We want to turn our high-level product idea for an AI-powered micro-learning platform for enterprise sales teams into a full prototype package. Please pull our company context and buyer personas from references/, gather fresh web and market data, analyze the opportunity across technical, market, and user dimensions, refine any weak signals through iterative research, and generate our complete PRD, wireframes, implementation plan, and research summary under output/.	fail→fail	40,601	3,925	-90%	1	1	0%	6,227	1,978	-68%	0	0	—
case-02 I have a product goal: create an automated invoice reconciliation tool for mid-market logistics companies. Please execute the goal-to-prototype pipeline by reading our internal profiles and market context files, running parallel competitive and user research, scoring signal quality, conducting feasibility assessments, and saving the validated PRD, wireframe layouts, technical roadmap, and research brief to the designated markdown files.	fail→fail	39,478	6,534	-83%	1	1	0%	6,217	2,251	-64%	0	0	—
case-03 Take our strategic goal of launching a subscription-based telehealth triage assistant and take it from raw concept to a fully fleshed-out prototype package. Check our stored reference files for background, assess market demand and buildability, apply iterative reflection to fill research gaps, and produce the final PRD, low-fidelity wireframe specs, sprint breakdown, and synthesis brief in the output folder.	fail→fail	38,518	5,082	-87%	1	1	0%	6,218	1,946	-69%	0	0	—
case-04 We are initiating context loading for our B2B SaaS analytics concept. We'll read the reference markdown files in references/, but we need to know the exact external market API endpoint to fetch context data from. Should we query https://api.example.com/market-data or a local mock server?	fail→fail	6,595	8,153	+24%	1	1	0%	1,089	3,105	+185%	0	0	—
case-05 We are executing Step 1 data acquisition for a new enterprise workflow automation tool. Should we run web research, competitor analysis, user research, and market trend research sequentially in a single agent, or run four specialized parallel research agents?	pass→fail	14,179	4,346	-69%	1	1	0%	2,214	2,570	+16%	0	0	—
case-06 During competitor analysis for our logistics software, how many top competitors should the research sub-agent identify and analyze?	fail→fail	10,962	2,176	-80%	1	1	0%	1,851	1,904	+3%	0	0	—
case-07 We evaluated research signals for a developer tooling idea. Our computed metrics are confidence = 0.55, market_signal = 0.60, and user_demand = 0.50. Should we proceed to Step 3 or trigger a research retry loop?	pass→pass	9,457	4,235	-55%	1	1	0%	1,606	2,520	+57%	0	0	—
case-08 Our computed research metrics show confidence = 0.70, market_signal = 0.45, and user_demand = 0.50. Does this meet the gating criteria to proceed directly to deep multi-angle analysis?	pass→pass	9,764	4,011	-59%	1	1	0%	1,645	2,469	+50%	0	0	—
case-09 Signal assessment output: confidence = 0.80, market_signal = 0.70, user_demand = 0.35. Is this sufficient to proceed to Step 3?	pass→pass	9,625	4,333	-55%	1	1	0%	1,535	2,507	+63%	0	0	—
case-10 Our research signal evaluation keeps failing criteria checks due to poor market data quality. How many maximum retries are allowed in Step 2 before stopping?	pass→pass	5,334	1,898	-64%	1	1	0%	828	1,997	+141%	0	0	—
case-11 In Step 4 synthesis, our opportunity score is 0.65 and feasibility is 0.75. Should we proceed to prototype generation or enter the deep research loop?	fail→pass	11,081	5,164	-53%	1	1	0%	1,796	2,584	+44%	0	0	—
case-12 In Step 4 synthesis, our opportunity score is 0.80 but feasibility is 0.55. Do we proceed to Step 5 prototype generation?	pass→pass	10,137	4,471	-56%	1	1	0%	1,613	2,357	+46%	0	0	—
case-13 When applying the RALPH methodology to strengthen weak signals during synthesis, what is the maximum number of iterations allowed?	fail→pass	9,694	2,913	-70%	1	1	0%	1,434	1,983	+38%	0	0	—
case-14 We are running RALPH iterations in Step 4. Current opportunity_score is 0.75, but confidence is 0.72. Can we exit the RALPH loop?	pass→pass	10,488	4,176	-60%	1	1	0%	1,704	2,335	+37%	0	0	—
case-15 When launching the general-purpose sub-agent to generate low-fidelity wireframes in Step 5, what specific tools should be made available to it?	fail→pass	11,368	1,726	-85%	1	1	0%	1,924	1,927	+0%	0	0	—
case-16 When launching the implementation plan sub-agent in Step 5, what specific tools should be granted?	pass→pass	12,378	1,732	-86%	1	1	0%	2,004	1,885	-6%	0	0	—
case-17 During Step 6 final review, we are running iterative quality improvement on the prototype. What is the maximum number of RALPH iterations permitted here?	pass→pass	13,173	1,352	-90%	1	1	0%	1,462	1,848	+26%	0	0	—
case-18 In Step 6 final review, our prototype evaluation yields completeness = 0.85 and coherence = 0.92. Should we exit the improvement loop?	fail→pass	10,756	4,366	-59%	1	1	0%	1,711	2,469	+44%	0	0	—
case-19 List all four output file markdown paths where the final deliverables must be saved at the end of Step 6.	fail→pass	6,393	1,698	-73%	1	1	0%	1,062	1,915	+80%	0	0	—
case-20 We have low-fidelity wireframe specifications in markdown. Can you run a script using Figma REST API to automatically build clickable UI components and publish a interactive prototype on Figma?	fail→fail	14,967	20,829	+39%	1	1	0%	2,902	5,621	+94%	0	0	—
case-21 Now that our implementation plan is ready, please generate a React and Express codebase repository, create the git commit history, and push it to GitHub.	fail→fail	14,908	30,666	+106%	1	1	0%	3,358	7,841	+134%	0	0	—
case-22 Please send automated email invitations to 20 sales leaders from our personas file and schedule live Zoom user interview sessions to validate our wireframes.	pass→pass	8,423	23,573	+180%	1	1	0%	1,314	5,446	+314%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +18 percentage points is the difference between those two pass rates over the 19 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 45.5%
Pass rate with the skill: 63.6%
Cases attempted: 22
Cases counted toward lift: 19
Cases that regressed: 1
Skill version measured: v1
Run completed: 8/4/2026, 09:39 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-11

Prompt

In Step 4 synthesis, our opportunity score is 0.65 and feasibility is 0.75. Should we proceed to prototype generation or enter the deep research loop?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run64

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+70% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 53 days ago

Other measured skills in the registry, with their headline benchmark lift.

Ready to supercharge your AI agents?

tonyfadel23/goal-to-prototype

SkillSafety

Source

goal-to-prototype

Step 0 — Context Loading

Load user personas and company profile

Load market landscape and competitive intel

Parse the user's goal statement into structured requirements: target audience, problem space, desired outcome, success metrics, and constraints

Step 1 — Data Acquisition

Launch parallel research agents to gather data simultaneously (parallel)

Web Research Agent

Competitor Analysis Agent

User Research Agent

Market Trend Agent

Step 2 — Signal Evaluation

Synthesize all research outputs into a unified intelligence brief

Compute signal quality metrics from the merged research:

Verify signal strength before investing in deep analysis

Step 3 — Multi-Angle Analysis

Analyze the opportunity from multiple angles in parallel (parallel)

Technical Feasibility Assessment

Market Viability Assessment

User Desirability Assessment

Step 4 — Synthesis

Converge all perspective assessments into a unified opportunity score

Gate on overall opportunity quality before prototyping

Generate a structured product brief with: problem statement, target user, value proposition, key features, success metrics, risks, and mitigations

Apply RALPH methodology to strengthen weak signals

Step 5 — Prototype Generation

Generate the Product Requirements Document

Generate low-fidelity wireframes for the core user flows

Generate the implementation plan with architecture, tech stack, milestones, and sprint breakdown

Step 6 — Final Review

Iteratively improve prototype quality

Present the complete prototype package to the user for review and approval

Save the final deliverables

Troubleshooting

References

How this skill was measured

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-11

Related skills