oransim-causal-marketing-twin — Claude Code skill: how it works
Simulates marketing campaign outcomes using causal digital twins. It predicts ROI distributions and tests counterfactual budget or creator shifts before spending money.
npx skills add reason-machines/trending-skills --skill oransim-causal-marketing-twin -a claude-codeWhat oransim-causal-marketing-twin does
This skill connects agents to the Oransim causal engine. The system maps influence across a 64-node causal graph with 117 edges, tracing paths from budget allocation to purchasing decisions. Simulations generate quantile ROI estimates with median P50 and confidence bounds.
Interventions rely on the do() operator. You can simulate replacing a creator mid-flight on day 3 or test post-mortem allocations across channels like xhs and douyin. Synthetic buyer agents drive the underlying Hawkes process alongside TARNet counterfactual heads.
You must run the local Uvicorn backend on port 8001 for simulations to work. The skill fails without this local service. You can still test deterministic stubs without external API costs by setting LLM_MODE=mock.
What it helps with
- Forecast campaign ROI across creative and budget variations before launch
- Calculate the trajectory impact of swapping creators mid-campaign
- Evaluate post-mortem scenarios comparing alternative channel allocations
- Audit causal graph paths connecting ad spend to purchase intent
Example prompts for the agent
- Simulate ROI for a beauty launch across xhs and douyin with a 500k budget
- Run a counterfactual swap replacing kol_001 with kol_003 on day 3
- Inspect the causal graph to see all edges originating from budget_allocation
How to install oransim-causal-marketing-twin
1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill oransim-causal-marketing-twin -a claude-code2Manually: put the files in place
~/.claude/skills/oransim-causal-marketing-twin/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/oransim-causal-marketing-twin/
1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill oransim-causal-marketing-twin -a codex2Manually: put the files in place
.agents/skills/oransim-causal-marketing-twin/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill oransim-causal-marketing-twin -a cursor2Manually: put the files in place
.cursor/skills/oransim-causal-marketing-twin/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install reason-machines/trending-skills oransim-causal-marketing-twin2Manually: put the files in place
.github/skills/oransim-causal-marketing-twin/SKILL.md1Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill oransim-causal-marketing-twin -a gemini-cli2Platform CLI
gemini skills install https://github.com/reason-machines/trending-skills --path skills/oransim-causal-marketing-twinManually: put the files in place
.devin/skills/oransim-causal-marketing-twin/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add reason-machines/trending-skills --skill oransim-causal-marketing-twin -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about oransim-causal-marketing-twin
How do I install oransim-causal-marketing-twin in Claude Code?
Run npx skills add reason-machines/trending-skills --skill oransim-causal-marketing-twin -a claude-code. The skill definition is saved to ~/.claude/skills/oransim-causal-marketing-twin/SKILL.md.
What does oransim-causal-marketing-twin do?
It provides an interface to a causal simulation engine that predicts marketing ROI and analyzes counterfactual scenarios. Teams test budget reallocations or creator replacements before committing actual ad spend.
Is an LLM API key required to run simulations?
Real agent runs require an API key for providers like OpenAI or Anthropic, but you can set LLM_MODE=mock to run deterministic local tests without external API credentials.