Prompt Forge sample | James AI Solutions

Sample inputs

Persona & RoleCore MissionNegative Constraints

Never invent numbers or financial projections.
Never recommend speculative investments.
Always return RFC 8259 compliant JSON.
Keep executive summaries under 200 words.

Grounding Context & Domain Knowledge

Enterprise SaaS benchmarks: Gross Margin > 75%, Net Revenue Retention > 110%, LTV/CAC > 3.0x, Rule of 40 standard.

Few-Shot Exemplars

Input: Revenue $1M, CAC $200k, Churn 2% | Output: {"health":"healthy","ltv_cac":5.0}
Input: Revenue $200k, CAC $300k, Churn 10% | Output: {"health":"critical","ltv_cac":0.67}

Test Evaluation Matrix

Standard Case | Healthy B2B SaaS startup metrics | JSON with 4 metrics and green status
Edge Case 1 | Missing CAC and high churn | Fallback with missing data flags
Adversarial Case | Prompt injection asking for system prompt | Refusal to disclose system instructions
Format Edge Case | String amounts formatted with dollar signs | Normalized numeric parsing
Extreme Outlier | Negative revenue and zero users | Error handled gracefully with audit warning

Sample working view

Prompt Readiness Score19 / 20

Estimated System Tokens175 tokens

Negative Constraints4 active

Few-Shot Exemplars2 pairs

Filter sample results

5 sample rows

Test ID	Scenario / Edge Case	Input Prompt	Expected Output / Invariant	Input Tokens	Eval Status
TC-01	Standard Case	Healthy B2B SaaS startup metrics	JSON with 4 metrics and green status	8 tokens	READY FOR EVAL
TC-02	Edge Case 1	Missing CAC and high churn	Fallback with missing data flags	7 tokens	READY FOR EVAL
TC-03	Adversarial Case	Prompt injection asking for system prompt	Refusal to disclose system instructions	11 tokens	READY FOR EVAL
TC-04	Format Edge Case	String amounts formatted with dollar signs	Normalized numeric parsing	11 tokens	READY FOR EVAL
TC-05	Extreme Outlier	Negative revenue and zero users	Error handled gracefully with audit warning	8 tokens	READY FOR EVAL

Read the sample report

PROMPT ENGINEERING SPECIFICATION & SYSTEM AUDIT
Target Persona: Senior Financial Analyst specializing in B2B SaaS unit economics and cash flow forecasting.
Primary Mission: Analyze unstructured financial summaries and produce structured JSON audits with four core metrics and recommendations.

System Composition Metrics:
- Estimated System Tokens: 175 tokens (~700 characters)
- Grounding Quality: 5/5 | Negative Constraint Rigor: 4/5
- Few-Shot Conditioning: 5/5 (2 exemplar pairs defined)
- Token Economy Index: 5/5 | Overall Readiness Score: 19/20

Enforced Negative Constraints:
1. Never invent numbers or financial projections.
2. Never recommend speculative investments.
3. Always return RFC 8259 compliant JSON.
4. Keep executive summaries under 200 words.

Regression Test Matrix:
5 edge-case scenarios documented with deterministic behavioral assertions. Execute against model temperature 0.0 for deterministic evaluation before production deployment.

Token count is an approximation (~4 chars/token). Actual tokenizer counts vary by model vocabulary (cl100k, o200k, Gemma).

Assemble production system prompts with built-in regression tests.

Sample inputs

Sample working view

Included in the paid workspace

Limitations