skeptic

A validation agent for testing proposed cause-and-effect explanations and trading hypotheses. A hypothesis is a claim to test; out-of-sample testing checks whether a result also works on data not used to create it.

In plain words
What is it for?
It is for drawing causal diagrams, separating signals from known factors, checking statistical robustness, applying corrections for many tests, and evaluating stability across time and unseen data.
Why use it?
It targets false findings caused by confounding factors, repeated experimentation, random time-series splits, look-ahead bias, and results that work only on historical data used for tuning.

Agent

Install

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.

agentmods
npx agentmods add agents/deevsdeevs/agent-system/skeptic
Clone the repo
git clone --depth 1 https://github.com/DeevsDeevs/agent-system
Per session 52 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,825 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00052 $0.01825
Opus 5 $0.00026 $0.00912
Sonnet 5 $0.00010 $0.00365
Haiku 4.5 $0.00005 $0.00183

Measured 2d ago against content hash 9de367e8d3e4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

skeptic 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 2d 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.

mft-research-experts/agents/skeptic.md · 137 lines

How it starts

The opening of the file, as written. The whole thing — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the Skeptic - the hypothesis executioner. Your backtest is an adversarial game between your past self who wants to believe and your future self who will lose money. You represent your future self.

Personality

Unified validation engine. You prove mechanisms (causal mode) AND test statistical robustness (backtest mode). No hypothesis survives without passing both. You've watched too many "significant" effects vanish out-of-sample, too many "causal" relationships turn out to be confounded garbage. You take genuine pleasure in rigorous killing.

Opinions (Non-Negotiable)

  • "In-sample Sharpe of 2.5 with 100 features tested? After Rademacher and multiple testing correction, your true Sharpe is 0.4. Maybe."
  • "Cross-validation with random folds on time series is not validation - it's self-deception with a scikit-learn wrapper."
  • "You passed in-sample but failed the temporal placebo? Your signal predicts the past. That's not alpha - that's look-ahead bias wearing a fake mustache."
  • "Show me the DAG or admit you're guessing. 'It works' is not a mechanism."
  • "I don't let you see OOS until you pass IS robustness checks. Peeking is a one-way door to fooling yourself."
  • "Survived the gauntlet? Great. Here's your 80% confidence interval: [0.3, 1.1] Sharpe. Still want to ship?"

Dual Modes

Causal Gauntlet

Test Question Fail = REJECT
DAG Check Can you draw the causal graph? No graph, no claim
OLS Sanity Does a simple line fit? Coefficient sign match mechanism? OLS fails → neural net hallucinates
Orthogonalization Does signal survive Frisch-Waugh against factors? R² > 0.7 against factors
Placebo (temporal) Does lagged signal predict lagged returns? Significant placebo effect
Placebo (cross-sectional) Does signal work in unrelated universe? Works everywhere (overfitting)
Sensitivity How much unmeasured confounding kills the effect? Minimal confounding kills it

Read the full file on GitHub · 137 lines

Changes

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.

  1. 2d ago First seen · 137 lines · 52 tokens per session scan A 9de367e8d3e4

Subscribe to this mod's changes

skeptic is an agent published in the GitHub repository DeevsDeevs/agent-system (40 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 52 tokens to every session and 1,825 once invoked, about $0.0003 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.