Use when a PM or product team needs a rapid feasibility signal on a feature idea before committing engineering time — assessing implementation risk, complexity, and make-vs-buy tradeoffs without full engineering involvement. Triggers on: "technical
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Use when a PM or product team needs a rapid feasibility signal on a feature idea before committing engineering time — assessing implementation risk, complexity, and make-vs-buy tradeoffs without full engineering involvement. Triggers on: "technical
license
MIT
metadata
{"author":"wondelai","version":"1.0.0"}
scenarios
["Product wants to build X — can you give me a quick technical feasibility read?","Help me assess whether this feature can be built in 2 sprints with our current stack","Run a feasibility check on this AI integration before we commit to the roadmap","이 기능이 우리 현재 스택으로 2스프린트 안에 가능한지 빠르게 평가해줘","로드맵 확정 전에 기술 타당성 검토를 도와줘"]
compatibility
{"recommended":["think-tool"],"optional":["sequential-thinking"],"remote_mcp_note":"think-tool이 있으면 5개 가능성 차원(데이터, 인프라, 통합, 팀 역량, 시간)에 대한 판단의 품질이 높아집니다. Claude 설정 → MCP Servers에서 remote SSE 엔드포인트를 추가하세요."}
Standing Mandates
ALWAYS assess all five feasibility dimensions: data, infrastructure, integration, team capability, and timeline.
ALWAYS surface the single highest-risk assumption explicitly.
NEVER give a binary feasible/not-feasible answer without a confidence level.
NEVER assess feasibility without knowing the team's current stack and capacity constraints.
Technical Feasibility Assessment Framework
When to Use / When Not to Use
Use when:
Deciding whether to include a feature in the roadmap before engineering is consulted
Running a reality check at a Shape Up betting table
Writing the Constraints / Technical Dependencies section of a PRD
Evaluating build vs. buy vs. partner for a new capability
Not for:
Engineering sprint estimation or story pointing
Detailed architecture design decisions
Contract negotiation with vendors
Process
Describe the feature — what it does, who it's for, any known constraints or deadlines
Score 5 dimensions — Data, Infrastructure, Integrations, Team Skill, Time (each: Low / Medium / High risk)
Identify red flags — any dimension flagging "escalate to engineering immediately"
Make vs. buy recommendation — based on differentiation, time-to-value, and team capability
Document risks — produce the PRD technical constraints section format
Output Template
Feasibility Signal: [High / Medium / Low / Unknown]
Dimension scores:
- Data: [Low / Medium / High risk] — [one sentence rationale]
- Infrastructure: [Low / Medium / High risk] — [one sentence rationale]
- Integrations: [Low / Medium / High risk] — [one sentence rationale]
- Team Skill: [Low / Medium / High risk] — [one sentence rationale]
- Time: [Low / Medium / High risk] — [one sentence rationale]
Make vs. Buy recommendation: [Build / Buy / Partner] — [reasoning]
Red flags requiring engineering consultation before any commitment:
- [if any]
Recommended next step: [what to investigate before treating this as feasible]
What Claude Does / What You Do
Claude
You
Scores 5 feasibility dimensions with rationale
Validates signals with engineering before committing
Identifies red flags that require engineering consultation
../shape-up/SKILL.md — use this before the betting table to set realistic appetite
../prd-development/SKILL.md — output feeds directly into PRD constraints section
../feature-prioritization/SKILL.md — effort signal informs prioritization scoring
Core Principle
A PM does not need to be an engineer to assess feasibility. They need to ask the right questions and know what the answers imply. This skill gives PMs a structured signal — not a spec, not an estimate — that surfaces technical risk early enough to change direction before sunk cost sets in.
What this is and isn't:
IS: a structured set of questions that surfaces complexity, risk, and uncertainty signals
IS: a decision framework for make vs. buy vs. partner
IS: guidance on what to escalate to engineering immediately vs. explore further
IS NOT: a substitute for engineering estimation or technical design
IS NOT: a way to commit to timelines without engineering sign-off
When to use it:
Before committing a bet at the Shape Up betting table (../shape-up/SKILL.md)
When writing the Constraints and Technical Dependencies section of a PRD (../prd-development/SKILL.md)
Before including a feature in a roadmap that promises delivery timelines
When evaluating build vs. buy tradeoffs for a new capability area
Agent Output
When a user describes a feature or capability and asks for a feasibility assessment, produce:
Feasibility signal — High / Medium / Low / Unknown (with one-sentence rationale per dimension)
Make vs. buy vs. partner recommendation with reasoning
Red flags requiring engineering consultation — if any, name them explicitly
Risk documentation — the right format for inclusion in the PRD constraints section
Recommended next step — what to investigate before treating this as "feasible"
The 5 Feasibility Dimensions
Run through all five before forming an overall signal. A single High-risk dimension can make an otherwise Simple initiative infeasible in the required timeline.
1. Data
What to assess: Does the data needed for this feature exist, is it accessible, and is it clean enough to use?
Signal questions:
Does this feature require data we currently collect? If not, how long to instrument and accumulate?
Does it require joining data across systems that don't currently communicate?
Does it require third-party data? Is there a data sharing agreement in place?
Does it require historical data at a scale we haven't validated?
Does it require user-provided data that users are unlikely to supply?
Risk signals:
"We'll need to collect new data for this" → add 1-3 months minimum before useful signal
"It relies on [other team]'s data pipeline" → coordination risk; not in your control
"We need X years of historical data" → migration risk; validate data quality before committing
Real-time data requirements → infrastructure and latency implications; escalate to engineering
Feasibility signal: High risk if data collection, quality, or access is unresolved. Low risk if the feature operates on data already well-instrumented in production.
2. Infrastructure and Platform
What to assess: Does the current infrastructure support this? What would need to change?
Signal questions:
Does this require new infrastructure (queues, caches, storage tiers, new services)?
Does it require scale characteristics the current system hasn't been tested for?
Does it need to work in environments (mobile, offline, specific regions) not currently served?
Does it depend on a platform or service that has known reliability, rate limit, or cost issues?
Does it require real-time processing where the current system is batch?
Risk signals:
"We'd need to move to a new architecture" → this is a multi-quarter project; don't scope it as a feature
"The system hasn't been load-tested at that scale" → escalate before committing to any public-facing rollout
"It requires [region] infrastructure we don't have" → compliance and latency implications; legal review may be required
Batch-to-real-time migrations are almost always underestimated
Feasibility signal: High risk if infrastructure changes are foundational. Low risk if the feature fits the current architecture.
3. Integrations
What to assess: Does this require connecting to external systems, APIs, or third-party services?
Signal questions:
Does this require a third-party API? Is it well-documented, stable, and affordable at scale?
Does it require integration with a system owned by another internal team?
Is there an existing integration that could be extended, or is this net new?
Does the integration have rate limits, SLA constraints, or data format quirks?
Is the third-party vendor reliable, or is there dependency risk?
Risk signals:
"It requires [ERP / legacy system] integration" → these are almost always 3-6x harder than estimated; require engineering review before any timeline commitment
Undocumented or unofficial APIs → fragility risk; changes without notice; avoid unless you own the contract
Internal team dependency with no agreed SLA → a coordination risk, not a technical risk, but it behaves the same
"We'll use [new vendor] for this" → vendor evaluation, contract, and onboarding time is a separate risk; don't include it in the build estimate
Feasibility signal: High risk if integrations involve legacy systems, internal team dependencies, or uncontracted third-party services.
4. Team Skill
What to assess: Does the engineering team have experience with the required technology, and if not, what does the learning cost?
Signal questions:
Does this require technology the team hasn't used in production before?
Does it require specialized skills (ML, cryptography, real-time systems, mobile native) that are not currently on the team?
Are there regulatory or security requirements that need specific expertise (HIPAA, PCI, GDPR)?
Is the knowledge concentrated in one person who is also a dependency for other work?
Risk signals:
"We'd need to hire for this" → it's not a feature timeline, it's a hiring + ramp timeline
"Only [one engineer] knows this part of the system" → bus factor risk; budget extra time for knowledge transfer
First-time ML / AI features are routinely 3-5x more complex than the PM imagines; data pipeline, model training, evaluation, and inference infrastructure are each substantial workstreams
Regulatory expertise gaps (security, privacy, accessibility compliance) → legal and compliance review required before committing
Feasibility signal: High risk if required skills are absent from the team. Medium risk if skills exist but are concentrated or the technology is new to the team.
5. Time
What to assess: Is there a fixed deadline driving this? Is the deadline real (contractual, regulatory, market window) or artificial (stakeholder preference)?
Signal questions:
Is there a hard external deadline? What is it anchored to?
What is the consequence of missing the deadline vs. the consequence of shipping something poor?
Has engineering been consulted on the timeline? What was their estimate?
Is the timeline compressing quality, scope, or both?
Risk signals:
"We need to ship by [date] for the customer contract" → escalate to engineering immediately; this is a committed deadline that changes the risk calculus
"We should try to get this out by Q[X]" → this is a soft deadline; negotiate scope, not quality
Engineering estimates that are 2x or more than stakeholder expectation → a conversation about scope or timeline is required before proceeding
Feasibility signal: Low risk if the timeline is flexible. High risk if deadline is fixed and engineering hasn't validated it.
Overall Feasibility Signal
After scoring all 5 dimensions:
Signal
Meaning
PM Action
Low risk across all 5
Standard feature; proceed with shaping
Include in roadmap; write PRD constraints section
Medium risk on 1-2 dimensions
Proceed with flagged risks documented
Include risks in PRD; discuss mitigations with engineering in next sync
High risk on 1 dimension
Significant uncertainty; needs engineering input before committing
Do not commit to timeline until engineering has reviewed; spike may be needed
High risk on 2+ dimensions
This is not a feature, it's a project
Rescope dramatically OR start a separate discovery/spike track; do not include as a standard roadmap item
