Forensic Research Engine
Agent Skill for Deep Investigative Research
Agent Skills have no user interface — this is a representative cover. See the deployment guide below to run it in your framework.
Built With
About this agent skill
A portable agent skill that runs deep, wide-ranging investigative research for legal research, investigative journalism, and criminal or financial investigations. It is built for cases where the goal is to prove or disprove a hypothesis from a confluence of circumstantial evidence. The skill is strictly scientific and mathematically transparent: it frames explicit primary and competing hypotheses, gathers evidence across public, government, court, and financial records, normalizes everything into an auditable evidence ledger (source-reliability and chain-of-custody graded with the NATO Admiralty system), scores each item, and then estimates the probability of the hypothesis with Bayesian updating, likelihood ratios, and Monte Carlo simulation. It ends with a court- or newsroom-grade report that shows every prior, weight, and formula. Packaged as a framework-agnostic skill (a SKILL.md plus reference docs, Python scripts, and templates) so it drops into Claude Agent Skills, Manus, Cursor, the OpenAI Agents SDK, LangChain/LangGraph, or any agent that can read a system prompt and run Python.
Deployment: run this skill in your agent framework
Forensic Research Engine ships as a framework-agnostic Agent Skill: a single SKILL.md entry point plus reference docs, Python scripts, and templates. Any agent runtime that can load instructions into context and execute Python can run it. Clone the repo, then point your framework at the SKILL.md and make the scripts/templates available in the working directory.
What ships in the package
- SKILL.md — the workflow entry point your agent reads first.
- references/ — data-sources, scoring-framework, statistical-methods, and report/viz standards.
- scripts/ — score_evidence.py and probability_model.py (Bayesian updating + Monte Carlo).
- templates/ — evidence-ledger-schema.json and findings-report.md.
Framework setup (6)
Drop the folder into your Skills directory; Claude auto-discovers SKILL.md via its YAML frontmatter.
- Clone the repo into your project's skills folder (e.g. `.claude/skills/forensic-research-engine`).
- Confirm SKILL.md keeps its YAML frontmatter (name + description) — that is how the skill is discovered and triggered.
- Enable the Code Execution / Files capability so Claude can run scripts/*.py and read templates/*.
- Invoke by describing an investigative goal; Claude loads the skill when the description matches.
git clone https://github.com/mattantimatter/forensic-research-engine \
.claude/skills/forensic-research-engineLawful use only: rely exclusively on public and lawfully obtained records, respect API rate limits and privacy law (DPPA, GLBA, FCRA), and present probabilities as decision-support metrics — not legal verdicts.
Special Features
- Hypothesis-driven framing (primary, competing, and null hypotheses)
- Auditable evidence ledger with timestamps and chain-of-custody
- NATO Admiralty source-reliability and credibility grading
- Transparent scoring for reliability, relevance, and FRE admissibility
- Bayesian updating and likelihood-ratio probability modeling
- Monte Carlo uncertainty propagation — never an unexplained number
- Court/newsroom-grade report with full methodology and visualizations
- Lawful-use guardrails (public records only; DPPA / GLBA / FCRA aware)
Workflow Steps
- Frame the case (hypotheses + standard of proof)
- Source & collect (public, court, financial records)
- Ingest & chunk into the evidence ledger
- Score evidence (reliability, credibility, admissibility)
- Model probability (Bayesian + Monte Carlo)
- Synthesize & visualize the findings report
Tags
License
MIT License — free to use, modify, and distribute. Use lawfully obtained public records only; present probabilities as decision support, not legal verdicts.
Antimatter AI
antimatterai.com ↗