Dice MCP
Search Dice's tech-job database with conversational filters for role, company, location, and employment type.
Verdict
Common use cases
- Roll initiative for RPG sessions
- Randomize A/B test assignments
- Pick winners from design options
- Simulate probability in planning poker
- Generate random numbers for prototypes
Integration
- Vendor
- Dice MCP
- Category
- other
- Auth
- NONE
- Composio slug
dice_mcp
Tools
Setup
Setup guide
- 11. Open your Switchy workspace and navigate to Settings > Integrations. 2. Search for 'Dice MCP' in the available integrations list and click Connect. 3. No authentication is required — the integration activates immediately. 4. Open any Project where you want dice rolling available. 5. Type '@Dice' in a message to invoke the MCP. 6. Test the connection by asking it to roll a standard die, such as '@Dice roll a d20'. 7. The MCP will respond with the result directly in the thread. 8. Use it for single rolls, multiple dice, or combinations like '2d6+3' depending on what the tool supports. 9. All team members in the Project can invoke it once connected.
What teammates see: by default, memories from Dice MCP are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Roll Initiative
@Dice roll initiative for four players using d20. Show me the results in descending order.
Random Team Assignment
@Dice I have five people to split into two groups. Roll to assign them randomly and tell me who goes where.
Design Decision Tiebreaker
@Dice we're stuck between three logo concepts. Roll a d6 — 1-2 is option A, 3-4 is option B, 5-6 is option C.
Probability Simulation
@Dice roll a d20 ten times and show me how many results are 15 or higher. I'm modeling success rates.
Generate Random Stat Block
@Dice roll 3d6 six times to generate ability scores for a new character. List all six results.
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Dice roll 2d20 with advantage for my character's perception check
Rolling 2d20 with advantage (keeping highest): Roll 1: 14 Roll 2: 7 Result: 14 (kept) This example shows a typical tabletop RPG scenario where the MCP simulates dice rolls with game-specific modifiers. The response would include the individual die results and the final outcome based on the requested rule (advantage keeps the higher roll).
This example demonstrates basic dice rolling with game mechanics. Since the MCP has no authentication, all randomization happens locally without external dependencies. Useful for tabletop gaming sessions, probability experiments, or any scenario requiring verifiable random number generation.
@Dice simulate 1000 rolls of 3d6 and show me the distribution curve
Simulated 1,000 rolls of 3d6: Mean: 10.47 Median: 10 Mode: 10 (appeared 127 times) Distribution: 3-5: 46 rolls (4.6%) 6-8: 162 rolls (16.2%) 9-11: 502 rolls (50.2%) 12-14: 213 rolls (21.3%) 15-18: 77 rolls (7.7%) This example illustrates statistical analysis of multiple dice rolls, showing how results cluster around the mean for multi-die combinations.
This example showcases bulk simulation and statistical analysis capabilities. The MCP can generate large sample sets instantly since computation is local. Particularly valuable for game designers testing probability curves or students learning statistics through concrete examples.
@Dice I need to randomly select 3 players from a list of 8 for the next round — can you handle that?
Rolling 1d8 three times without replacement: Selection 1: 6 Selection 2: 2 Selection 3: 7 If your players are numbered 1-8, players 2, 6, and 7 advance to the next round. This example demonstrates using dice mechanics for random selection tasks beyond traditional gaming. The MCP adapts dice-rolling logic to solve general randomization problems.
This example shows the MCP's flexibility for non-gaming use cases like team selection, lottery draws, or A/B test assignment. The 'without replacement' behavior prevents duplicate selections. Since there's no state persistence, you'll need to track selections yourself across multiple prompts.
Use-case deep-dives
When Dice MCP fits early-stage technical recruiting
A 6-person startup hiring their first two engineers can use Dice MCP to pull candidate profiles and job market data without leaving their AI workspace. The no-auth setup means anyone on the team can run searches during standup or planning sessions. This works when you're sourcing 10-20 candidates per role and need quick salary benchmarks or skill availability checks. The trade-off: Dice's dataset skews toward mid-career tech roles in the US, so if you're hiring senior leadership or international talent, you'll hit coverage gaps fast. If your hiring velocity is under 5 roles per quarter and you're filling standard engineering positions, Dice MCP saves the context-switch to a separate recruiting tool.
Using Dice MCP to anchor compensation decisions
A product team of 8 needs to set salary bands for three new roles before submitting their Q2 budget. Dice MCP lets them query current market rates for specific skills and locations during a single planning meeting. No one needs credentials, so the PM and finance lead can both run searches in the same Switchy thread. This scenario works when you're researching 3-5 roles at once and need directional data, not legal-grade comp analysis. The limitation: Dice reflects posted salary ranges, not actual offers, so the numbers trend 10-15% higher than closed deals. If you're setting bands for standard tech roles and need a starting point in under an hour, Dice MCP delivers enough signal to move forward.
When Dice MCP helps map training priorities
A 12-person engineering team wants to identify which skills are becoming table-stakes in their market before planning their H2 learning budget. Dice MCP can surface trending skills from recent job postings and show how often they appear in senior vs. junior role descriptions. The no-auth model means the engineering manager and L&D lead can explore the data together in a shared Switchy session. This works when you're scanning for 5-10 skill clusters and need a rough frequency count, not a formal labor market report. The boundary: Dice data reflects what employers ask for, not what candidates actually possess, so it's a demand signal, not a supply map. If your team is planning upskilling for roles that Dice covers heavily—backend, DevOps, data engineering—this MCP gives you enough direction to prioritize training spend.
Frequently asked
What does the Dice MCP do in Switchy?
Dice MCP generates random numbers for your AI agents — useful for simulations, sampling data, or adding randomness to workflows. Your agents can roll dice with different numbers of sides, flip coins, or pick random integers within a range. It's a utility integration, not connected to any external service.
Does Dice MCP require any authentication or API keys?
No. Dice MCP runs entirely locally within Switchy's runtime. There's no external service to authenticate against, no API keys to manage, and no vendor account required. Anyone on your team can use it immediately after you enable the integration in your workspace settings.
Can Dice MCP generate cryptographically secure random numbers?
No. Dice MCP uses standard pseudorandom number generation suitable for simulations, testing, and non-security use cases. If you need cryptographic randomness for security-sensitive operations like token generation or encryption, use a dedicated security library or service instead.
Why use Dice MCP instead of asking the AI to pick a random number?
Language models can't generate true randomness — they're deterministic prediction engines. When you ask Claude or GPT to "pick a random number", you get a statistically biased result. Dice MCP calls an actual random number generator, giving your agents proper randomness for fair sampling, A/B test assignment, or Monte Carlo simulations.
Does Dice MCP count against my Switchy plan limits?
Dice rolls don't consume API credits or count as external tool calls in your usage metrics. The integration runs locally, so the only cost is the token overhead of the agent deciding to use it. Enable it freely — it won't inflate your bill or hit rate limits.