QUALITY SCORE0
No analysis yetEngine readyv1.18.3
AI INSTRUCTION PRODUCTIVITY
Write better AI instructions. Learn what improves them.
PQA helps you get to a reliable result faster—before an instruction reaches your AI model. Analyze quality, remove ambiguity, learn from your history, and track measurable improvement over time.
Improveclarity & reliabilityLearnfrom recurring gapsTrackquality over time
RELIABILITY SCORE82
PRODUCTIVITY
Fix instruction problems before they become retries.
See the reliability gaps, recommended changes, and a stronger instruction in one focused workflow—so developers spend less time correcting avoidable AI output.
Start analyzing →LEARNING
Turn every prompt into a learning signal.
PQA identifies recurring weaknesses and shows how your quality scores evolve. Your learning report makes improvement visible instead of relying on intuition.
View learning experience →PROGRESS
Understand what is getting better—and what still needs work.
Track quality, risk, confidence, trends, and retained analysis history. Export polished reports when you need to share progress.
Explore progress →CONTROL
Use the right level of AI for the work.
Start with deterministic reliability checks, add hosted AI when available, and move into document-aware analysis, higher limits, and team controls as your needs grow.
Compare plans →ANALYZE
Improve your instruction
Your prompt is the center of the workflow. Write it, review it, improve it.
Analysis progressReady
0%
Prompt/context required for refresh recovery is stored locally in this browser. For platform reliability and abuse prevention, PQA records limited operational telemetry and a short prompt preview; uploaded project context is not stored in reliability telemetry. If Gemini is enabled, evaluation content is sent to the configured Gemini endpoint.
RELIABILITY ASSESSMENT
Instruction quality
0/100NOT ANALYZED
Hallucination risk —
Score range —
Confidence —
Evaluation mode —
Deterministic analysis active
PQA is using its fixed reliability rules without hosted AI review.
Deterministic vs hosted AI
Deterministic analysis applies PQA’s fixed scoring rules consistently and does not pretend to be machine-learning output. Hosted AI adds Gemini’s model-based interpretation of nuance, context, and instruction quality. Both modes are labeled explicitly.
Sign in or compare plans →Reason codes
machine-readableNo reason codes
Evidence & controls
positive safeguards- No evidence controls detected yet.
Findings
reliability gaps- Analyze an instruction to see findings.
Recommendations
next actions- Recommendations will appear here.
Project-context evidence
0 sourcesUpload Markdown/README files to enable evidence-backed project-context analysis.
REMEDIATION
Recommended instruction
HALLUCINATION RISK—
Awaiting evaluationCONFIDENCE—
Score confidenceLOCAL TREND—
Recent browser analysesRELIABILITY TREND
Recent prompt quality
Run analyses to build a local reliability trend. No synthetic data is shown.
PQA PLANS
Upgrade for more access to PQA
Choose more analysis capacity, deeper prompt improvement, document context, or team controls without changing how you work.
Cancel anytime. Monthly or annual billing. Prices and payment methods follow your selected country and currency.
Save more with annual billing
PQA LEARN
PQA Learn
Guided prompt learning and more hosted AI capacity.
—Loading price…
PQA PRO
PQA Pro Developer
Richer context and advanced prompt improvement tools.
—Loading price…
PQA TEAM
PQA Team
Shared capacity, controls, and centralized visibility for teams.
—Loading price…
CURRENT PLAN
Community Online
Hosted AI quota applies
AVAILABLE—Hosted AI allowance
✓ Prompt analysis stays at the center✓ Usage and billing stay visible✓ Upgrade without changing your workflow
ACCOUNT & BILLING
Your subscription, payments, and usage
Sign in to subscribe or manage a plan
Use your existing identity. Your payment access is linked securely to your PQA account.
ACCOUNT——
PLAN—
PROVIDER—
RENEWAL / TERM END—
AI USAGE—
POSTED PAYMENTS—
Payment history and AI usage details
PQA ADMINISTRATION
Owner control center
Manage plans, pricing, payment countries, AI runtime configuration, reliability, and diagnostics from one workspace.
PLANS & BILLINGPricing, countries & payment providers
Edit Learn, Pro, and Team prices; enable countries; configure Stripe, Mercado Pago, and Pix.
Open configuration → AI RUNTIMEModel & reliability controlsChange the active Gemini model, test connectivity, and review operational reliability telemetry.
Open AI controls → DIAGNOSTICSExecution evidence & logsInspect live execution logs and structured Gemini evidence when troubleshooting the platform.
Open diagnostics →USAGE & ENTITLEMENTSServer-enforced plan allowances
Monthly hosted-AI and history limits remain protected by the PQA_LIMIT_* and PQA_HISTORY_* server settings.
OWNER ONLY
Plan, country & payment configuration
Plan prices are edited here and saved to the Supabase billing catalog. Country and payment-provider changes take effect without a redeploy after live validation.
Plan pricesEdit monthly and annual Learn, Pro, and Team values below. Stored in Supabase
billing_prices.Usage limitsHosted-AI allowances are enforced server-side with the Netlify
PQA_LIMIT_* environment variables.PQA LEARN
Build better prompting habits from your own work
PQA turns repeated reliability gaps into targeted learning so improvement compounds across your real prompts—not generic examples.
Sign in with a Learn/Pro/Team/Enterprise entitlement to load lessons.
AI EFFICIENCY VALUE
Estimated savings and engineering productivity
AUDITABLE MODEL
Classification: ESTIMATED. Direct AI cost uses the selected pricing profile. Productivity value uses configurable rework assumptions. Use observed provider usage when available for financial reporting.
ENGINEv2.1.0
BENCHMARKv4.2 · 400 cases
RISK MODELEvidence-tiered v3
MODEPrompt + docs · repo optional
AUDITABILITY
Evaluation provenance
- PQA engine
- v2.1.0
- Rubric
- —
- Risk model
- evidence-tiered-v3
- Scoring profile
- —
- Judge
- —
- Judge latency
- —
- Request ID
- —
- Updated
- —
RISK EVIDENCE
Detected indicators
- No risk evidence available yet.
DIAGNOSTICS
Live execution log
Ready.
JUDGE EVIDENCE
Gemini structured response
No Gemini response yet.
RETAINED HISTORY
Prompt analysis history
History retention follows your account policy. Free accounts retain up to the configured maximum.
PROMPT LEARNING REPORT
Your reliability evolution
A report-card view of retained prompt quality, learning patterns, and improvement over time.
Deterministic insight
Complete analyses to establish a learning trajectory.
INTERPRETATIONComplete at least two analyses to establish a measurable trajectory.
AI evolution insight uses one hosted-AI allowance and is optional. The report itself is always generated deterministically.Complete analyses to build your prompt-learning transcript.
RELIABILITY ENGINEERING — OWNER ONLY
PQA Platform control room
Operational truth for requests, users, AI reliability, usage, geography and conversion.
ACTIVE GEMINI MODELLoading…runtime configuration
Traffic geography
coarse edge locationNo geography data yet.
Tier mix
request volumeNo tier data yet.
Recent analyses
prompt preview · user · location · outcome| Time | User | Prompt / use case | Location | Tier | Status | Score | Latency | Tokens |
|---|---|---|---|---|---|---|---|---|
| No analysis telemetry yet. | ||||||||
Users
recent accountsNo users loaded.
Platform events
diagnostics & securityNo events loaded.