- Home
- Skills
- Research & Analysis
- Decision Pressure Test
Works with the AI Agents you already use
Decision Pressure Test
Pressure-test a consequential architecture or product decision by exposing assumptions, alternatives, reversibility, blast radius, and blocking evidence before commitment.
$14
Decision Pressure Test
Example session with this skill installed
Pressure-test this proposal before we commit: move the primary database from Postgres to DynamoDB for scale. We believe writes will grow 50x, but the query patterns and rollback plan are not confirmed.
- Read your context and instructions
- Compiled the decision pressure test
Illustrative worked example. Synthetic data from the package; not an observed external execution.
DECISION: migrate primary store Postgres -> DynamoDB.
OBJECTIVE: handle projected 50x write growth.
ASSUMPTIONS: current bottleneck is DB writes (ASSUMED; no profiling shown); access patterns are key-based (UNKNOWN-BLOCKING).
ALTERNATIVES: read replicas plus partitioning on Postgres; managed Aurora; caching layer. None evaluated yet.
DECISION GAPS
- Are current queries relational (joins or ad hoc)? -> UNKNOWN-BLOCKING (DynamoDB punishes this).
- Is the 50x number validated? -> ASSUMED.
- Migration reversibility? -> UNKNOWN-BLOCKING once dual-write is removed.
BLOCKING EVIDENCE NEEDED: query-pattern audit; load profile confirming write-bound.
REVERSIBLE DEFAULTS: add read replicas plus a cache now while profiling.
RECOMMENDATION / NO-DECISION: NO-DECISION until query patterns and load profile are in. A relational-heavy workload on DynamoDB is a high-blast-radius, hard-to-reverse mistake.
Connects securely to your tools. The creator never sees your data.
About this skill
Before committing to a consequential architecture, product, or implementation direction, Decision Pressure Test frames the decision, objective, constraints, alternatives, and what would make the choice wrong. It tests the proposal against evidence instead of treating plausibility as proof.
What it does
- Surfaces assumptions and labels questions that could change the choice as KNOWN, ASSUMED, UNKNOWN-BLOCKING, or UNKNOWN-REVERSIBLE.
- Compares credible alternatives across reversibility, blast radius, operational burden, security or privacy impact, migration or rollback, and success evidence.
- Names blocking evidence and reversible defaults when the available record is not enough for a safe commitment.
- Returns a recommendation or NO-DECISION while keeping the approval boundary explicit.
How it works
- Frame the decision, objective, options, constraints, deadline, and what would make the decision wrong.
- Build a gap map and keep only questions whose answers could change the choice.
- Test at least one credible alternative through the decision lenses.
- Stop when remaining uncertainty is reversible and below the risk threshold, or pause on named blocking gaps.
Output
The pressure-test follows a fixed structure:
- DECISION
- OBJECTIVE
- ASSUMPTIONS
- ALTERNATIVES
- DECISION GAPS
- BLOCKING EVIDENCE NEEDED
- REVERSIBLE DEFAULTS
- RECOMMENDATION / NO-DECISION
Boundaries
This is a read-only decision pressure test, not implementation, project governance, an approval record, or a release verdict. It does not invent estimates or silently convert a recommendation into an approved decision. Dependency installation, commits, pushes, merges, deploys, and destructive migrations are not implied.
Why it is different
Generic brainstorming can produce plausible pros and cons. Decision Pressure Test makes uncertainty explicit, compares reversibility and blast radius, names the evidence that could change the call, and stops at a clear recommendation or no-decision boundary.
This is a read-only decision pressure test, not implementation, approval, project governance, or a replacement for domain experts. It requires a clearly framed consequential decision and enough evidence to identify what could change the choice. It does not invent estimates or verify unavailable runtime, cost, traffic, security, migration, or architecture evidence. Shell, Git, browser, network, MCP, and package-manager capabilities are not assumed; unavailable checks are marked UNVERIFIED or BLOCKED.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 17 days ago
Permissions
Read access is required for decision context, authoritative artifacts and relevant evidence. Terminal/Shell is optional for focused fact checks with repository-native or manual fallback. The workflow is read-only and does not require write access. Browser, network access, MCP services, external accounts and environment variables are not required by the package.