The all-in-one platform for prompt engineering teams
Write, version, test, and deploy prompts — with skill files, plugins, playbooks, batch evaluations, and environment management. Built for teams that ship AI products.
Prompt Editor
Version Control
Testing & Evaluation
Team Collaboration
Plugins & APIs
Compatible with the Agent Skills standard — works with 25+ platforms
Claude Code
OpenAI Codex
Gemini
GitHub Copilot
OpenRouter
Anthropic
OpenAI
DeepSeek
Hugging Face
Ollama
Perplexity
Grok
Replit
Lovable
Windsurf
VS CodeClaude CodeOpenAI CodexGeminiGitHub CopilotOpenRouterAnthropicOpenAIDeepSeekHugging FaceOllamaPerplexityGrokReplitLovableWindsurfVS CodePrompts
Manage your team's prompt library
Generate empathetic, professional support responses
Analyze PRs and provide structured feedback
Create SEO-optimized outlines from a topic + keywords
Craft personalized outreach from prospect data
Classify and prioritize incoming bug reports
Transform meeting transcripts into actionable notes
Everything you need to ship AI products
From rich editing to observability, dependency graphs, and real-time collaboration. Every feature built, polished, and ready to ship.
Version control designed for prompts
Track every change with a visual diff engine, deploy to environments, and rollback with confidence.
- Version timeline with change summaries
- Side-by-side JSON diff viewer (added/removed/modified)
- Auto-generated change summaries
- One-click rollback to any version
- Environments: dev, staging, production + custom
- Deploy any version to any environment
- Deployment history with rollback reasons
Test prompts like you test code
Compare models side-by-side, score outputs with LLM-as-Judge, and run batch evaluations across entire datasets.
- Side-by-side comparison of 2–4 models
- LLM-as-Judge automated quality scoring (1–5)
- Batch evaluations with CSV datasets
- Golden examples with star ratings
- Test run history with results matrix
- Regression comparison against baselines
- Usage analytics with cost tracking
| Example | GPT-4o | Claude 3.5 | Gemini Pro | Llama 3.1 |
|---|---|---|---|---|
| Support ticket #1 | 4.5 | 4.8 | 3.9 | 4.2 |
| Support ticket #2 | 4.2 | 4.6 | 4.1 | 3.8 |
| Refund request | 3.8 | 4.9 | 4.0 | 4.5 |
| Bug report | 4.7 | 4.7 | 4.3 | 4.0 |
Built for teams, not individuals
Role-based access, threaded comments with text anchoring, activity feeds, and a master library that scales across workspaces.
- RBAC: admin, editor, viewer per workspace
- Threaded comments with resolve/unresolve
- Anchor comments — select text, leave feedback
- Activity feed with notification bell
- Token-based workspace invites (7-day expiry)
- Master library with fork/sync across workspaces
- Super admin dashboard for platform oversight
Should we add a fallback for when the API returns empty?
Good point — I added a null check in v4. Can you review?
The tone in paragraph 3 is too formal for our brand voice.
Skills, plugins, playbooks, and APIs
Go beyond prompts with an ecosystem of skill files, plugin packages, step-by-step playbooks, embeddable forms, and a full REST API.
- Skill files compatible with 25+ agent platforms
- Plugin packages with Claude Code export
- Playbooks — step-by-step guides linking resources
- Embeddable forms — turn prompts into public web forms
- REST API with key auth and rate limiting
- Webhook export with HMAC-SHA256 signing
- CSV and Markdown import
Production observability, built in
Track every LLM call across dashboard, API, and SDK. Real-time metrics, request logs with full I/O, latency percentiles, and cost breakdowns — so you catch regressions before your users do.
- Real-time metrics: throughput, error rate, avg latency, cost
- Time-series charts with p50/p95/p99 latency percentiles
- Full request log with expandable input/output previews
- Filter by model, status, source (dashboard/API/SDK/form)
- Per-prompt performance sparklines on detail pages
- Configurable alerts for error spikes and latency thresholds
- Environment-aware tracking (dev, staging, production)
Snippets — composable prompt building blocks
Create reusable content blocks and embed them across prompts with a slash command. Change a snippet once, update everywhere. Think design tokens, but for prompts.
- Rich-text snippet editor with the same Tiptap tooling
- Insert via /snippet slash command with fuzzy search
- Visual snippet blocks with colored border in the editor
- Auto-slug generation from snippet name
- Tag and search snippets across the workspace
- Usage tracking — see every prompt using a snippet
- Recursive resolution with cycle detection (max depth 5)
See how everything connects
An interactive dependency graph that maps every relationship in your workspace — prompts, skills, plugins, playbooks, snippets, and collections. Know the blast radius before you change anything.
- Interactive React Flow canvas with zoom, pan, and minimap
- Color-coded nodes by entity type (prompt, skill, plugin, etc.)
- Impact analysis — see what breaks before you edit
- Warning banners on edit pages: "Used in 3 playbooks"
- AI-powered prompt analysis with 1–100 quality scoring
- Findings by category: clarity, safety, structure, token usage
- One-click suggestions to improve prompt quality
Deploy with confidence, not anxiety
CI/CD guardrails scan for secrets, injection attacks, and regressions before any deployment. Approval workflows ensure the right people review changes before they go live.
- Secret scanning — detect API keys, tokens, PII in prompt content
- Prompt injection detection — flag bypass and jailbreak patterns
- Evaluation gates — block deploy if quality score drops below threshold
- Manual approval workflows with multi-step chains
- Environment-scoped rules (production-only, staging+prod, etc.)
- Admin override with reason logging for audit trail
- Approval queue with approve/reject and threaded comments
Track what matters — business outcomes
Every prompt platform measures tokens and latency. PromptCask measures what actually matters: did the user accept the suggestion? Did the email convert? Connect executions to real-world results.
- Binary outcomes — accepted/rejected, resolved/unresolved
- Rating outcomes — 1–5 star feedback from end users
- Label outcomes — categorize results (accepted, edited, rejected)
- Numeric outcomes — conversion rate, NPS score, revenue impact
- REST API for logging outcomes from any system
- Per-prompt effectiveness dashboard with trend charts
- Outcome-based version comparison — which version performs best
Chain prompts into executable workflows
Turn playbooks from documentation into runnable pipelines. Execute step-by-step with output chaining, or build visual DAGs with conditional branching and transforms.
- One-click playbook execution with step-by-step progress
- Output chaining — step 1 output becomes step 2 input
- Visual DAG builder with React Flow canvas
- Condition nodes — branch based on LLM output
- Transform nodes — reshape data between steps
- Execution history with cost, tokens, and duration per step
- Topological sort ensures correct execution order
A prompt editor built for power users
Slash commands, template variables, syntax highlighting, tables — all in a distraction-free writing experience.
Plus dozens of thoughtful details
Every feature is polished and production-ready.
Streaming Execution
Real-time token streaming via SSE with abort support
Model Configuration
Temperature, max tokens, top P — saved per prompt
API Key Encryption
AES-256-GCM encryption at rest for all API keys
Avatar Upload
Profile pictures with instant preview and storage
Analytics Dashboard
Cost tracking by model, department, user, and time
Multi-format Import
CSV bulk import and Markdown files
Embeddable Forms
Turn any prompt into a public web form with branding
Rate Limiting
Per-workspace daily limits based on plan tier
Command Palette
Cmd+K to navigate anywhere in your workspace instantly
Dark & Light Mode
System-aware theme with OKLCH color tokens
Variable Types
Text, textarea, select dropdown, and number inputs
Notifications
Bell icon with unread count and mark-as-read
Snippets
Reusable content blocks embedded across prompts
AI Optimizer
LLM-powered prompt analysis with scoring
Observability
Real-time monitoring, latency, error rate, cost
Dependency Graph
Visual map of prompt/skill/plugin relationships
SDK API
REST API for render, execute, log, and query
Real-time Presence
See who's editing what, live
CI/CD Guardrails
Secret scanning, injection detection, eval gates before deploy
Approval Workflows
Multi-step approval chains before environment promotion
Outcome Tracking
Connect executions to real-world business results via API
Workflow Builder
Visual DAG editor for multi-step prompt chains
Semantic Search
pgvector-powered similarity search across all content
See how PromptCask compares
Other tools manage prompts. PromptCask manages your entire prompt engineering workflow — prompts, skills, plugins, playbooks, and more.
| Feature | PromptCask | PromptHub | PromptLayer |
|---|---|---|---|
| Core Editor | |||
| Rich text editor (WYSIWYG) | |||
| Template variables | |||
| Slash commandsUnique | |||
| System prompt support | |||
| Code syntax highlighting | |||
| Version Control | |||
| Version history | |||
| Visual side-by-side diffUnique | |||
| Auto-generated change summariesUnique | |||
| One-click rollback | |||
| Git-style branching & merge requestsComing soon | |||
| Testing & Evaluation | |||
| Side-by-side model comparison | |||
| LLM-as-Judge scoring | |||
| Batch evaluations with datasets | |||
| Regression comparisonUnique | |||
| Golden examples with ratingsUnique | |||
| Multi-turn conversation evaluationComing soon | |||
| Collaboration | |||
| Role-based access control | |||
| Threaded commentsUnique | |||
| Text-anchor commentsUnique | |||
| Activity feed & notificationsUnique | |||
| Workspace invites | |||
| Community prompt sharingComing soon | |||
| Ecosystem | |||
| Skill files (Agent Skills standard)Unique | |||
| Plugin systemUnique | |||
| Playbooks (step-by-step guides)Unique | |||
| Embeddable formsUnique | |||
| Environments & deployment | |||
| Master library with fork/syncUnique | |||
| Snippets (reusable components)Unique | |||
| Dependency graphUnique | |||
| CI/CD guardrails (secret scan, injection detect) | |||
| Approval workflowsUnique | |||
| Visual workflow builder | |||
| Intelligence & Monitoring | |||
| AI prompt analysis & optimization | |||
| Production observability | |||
| Real-time presenceUnique | |||
| Outcome tracking (business metrics)Unique | |||
| Semantic search (pgvector)Unique | |||
| API & Integration | |||
| TypeScript SDK | |||
| REST API | |||
| Webhook exports | |||
| CSV / Markdown import | |||
| Analytics dashboard | |||
Fair pricing for every team size
Start free. Upgrade when your team grows. No surprise bills.
Free
For individuals exploring prompt management
- 3 team members
- 50 prompts, 10 skill files
- 10 versions per item
- 1 workspace
- 50 executions / day
- Basic editor features
- Community support
- Model comparison
- Batch evaluations
- REST API
Pro
For teams that need testing and collaboration
- Unlimited members
- Unlimited prompts & skill files
- Full version history
- 5 workspaces
- 500 executions / day
- Model comparison (2 models)
- LLM-as-Judge testing
- Collections & tags
- Email support
- Batch evaluations
Team
For organizations shipping AI products at scale
- Everything in Pro
- Unlimited workspaces
- Unlimited executions
- Model comparison (4 models)
- Batch evaluations & datasets
- Environments (dev / staging / prod)
- Embeddable forms
- REST API & webhooks
- Master library with fork/sync
- Analytics dashboard
- Priority support
PromptLayer Team costs $500/mo. PromptHub Team is $20/user/mo. PromptCask gives you more — prompts, skills, plugins, playbooks, evaluations, environments, and forms — starting at $15/user/mo.
Frequently asked questions
Everything you need to know about PromptCask. Can't find what you're looking for? Reach out to our team.