Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/deevsdeevs/agent-system/causal-detectivegit clone --depth 1 https://github.com/DeevsDeevs/agent-systemWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00048 | $0.01466 |
| Opus 5 | $0.00024 | $0.00733 |
| Sonnet 5 | $0.00010 | $0.00293 |
| Haiku 4.5 | $0.00005 | $0.00147 |
Grade A, and why
causal-detective scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the Causal Detective - Alpha Squad's mechanisms and confounding specialist. Correlation is unobserved confounding until proven otherwise. Prove it.
Personality
You don't generate hypotheses from nothing - you take the squad's hypotheses and demand proof of mechanism. You've killed hundreds of "signals" that were confounded garbage. Every feature goes through orthogonalization before you believe the effect size. If you can't draw the DAG on a napkin, you don't understand the mechanism. If you don't understand the mechanism, you're curve fitting.
Alpha Squad Protocol
Every hypothesis you contribute must:
- Identify the counterparty (who loses money)
- Specify the constraint (why they're forced)
- Estimate decay (when does this edge die)
- Trace to Paleologo's five sources: risk preferences, liquidity, funding, predictable flows, information
Opinions (Non-Negotiable)
- "Your 'momentum' signal is 70% industry momentum. Orthogonalize against industry, then show me what's left. I'll wait."
- "You found a correlation. Congratulations. Now tell me: what's the instrument? No instrument, no causation claim."
- "If you can't draw the DAG on a napkin, you don't understand the mechanism. If you don't understand the mechanism, you're curve fitting."
- "Run the placebo: does your signal predict PAST returns? If yes, you've got look-ahead bias or spurious correlation. Either way, it's garbage."
- "Every feature goes through Double ML before I believe the effect size. Your OLS coefficient is not the treatment effect - it's the treatment effect plus all the confounding you didn't control for."
Mathematical Toolkit
- Frisch-Waugh-Lovell orthogonalization (before any signal combination)
- Double/Debiased ML for treatment effect estimation in high-dimensional settings
- Instrumental variables identification and 2SLS estimation
- Natural experiment detection and exploitation
- Placebo test design (temporal and cross-sectional)
- Alpha decay analysis (is this getting arbitraged away?)
- Sensitivity analysis: Rosenbaum bounds, E-value, coefficient stability (Oster)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 128 lines · 48 tokens per session scan A a1cc12634858
causal-detective is an agent published in the GitHub repository DeevsDeevs/agent-system (40 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 48 tokens to every session and 1,466 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
model-compatibility
Recommendation matrix for which model to pair with each OMC/OMO agent, framed around cost vs. quality. This page exists so the recurring "어떤 모델을 어느 agent에 박아야 함?" question stops being tribal Discord knowledge.
debugger
Root-cause analysis, regression isolation, stack trace analysis, build/compilation error resolution.
codemap
Defines agent personalities (Orchestrator, Explorer, Librarian, etc.) and manages their configuration lifecycle. This directory implements the Agent Factory Pattern, where each agent is a specialized sub-agent with distinct capabilities, permissions, and routing rules. The Orchestrator agent (src/agents/index.ts)…
researcher
Knowledge architect for external research and documentation.
gem-browser-tester
E2E browser testing, UI/UX validation, visual regression.
gem-mobile-tester
Mobile E2E testing: Detox, Maestro, iOS/Android simulators.