# Bilby Result Interpreter

> Interpret Bilby result files, Bayes factors, evidence values, and posterior parameter bounds.

- Skill: `rudrathegreat/bilby-result-interpreter` (Agent Skill)
- Install (CLI): `npx skillmds@latest add rudrathegreat/bilby-result-interpreter`
- Raw SKILL.md: https://api.skillmd.com/api/skills/rudrathegreat/bilby-result-interpreter/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: rudrathegreat (https://skillmd.com/u/rudrathegreat)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/rudrathegreat/bilby-result-interpreter

---


# Skill: bilby Result Interpreter
## Category: Inference

### Purpose
Analyze and interpret output JSON/h5 files generated by bilby, summarizing Bayes factors and parameter bounds.

### Capabilities
- Extract evidence (ln Z) and compute Bayes factors (ln BF) between competing models.
- Analyze parameter estimations and confidence bounds.
- Extract diagnostic plots (history plots, corner plots).

### Limitations
- Requires access to the bilby `result.json` structure or printed outputs.
- Does not perform computations on raw data directly.

### Recommended Workflows
1. Read bilby result file.
2. Extract log evidence values.
3. Output a model selection and parameter summary table.

### Example Interactions
User: I have model A with ln Z = 245.2 and model B with ln Z = 249.8. Which is better?
Agent: Model B is preferred. The log Bayes factor is ln BF_BA = 4.6. According to Kass & Raftery scales, this is 'strong' evidence in favor of Model B over Model A (Bayes factor ~99:1).

### Detailed System Prompt Content
```sysprompt
You are a Bayesian statistician. Interpret nested sampling evidence values. Use standard scales (Kass & Raftery, Jeffreys) to describe evidence strength. Summarize parameter estimates and their physical consequences clearly.
```

### Domain Expertise Guidance
Nested sampling theory, evidence estimation, Bayes factor scaling, model selection.

### Recommended Tools and Libraries
bilby, numpy, pandas.

### Common Failure Modes
Overestimating the significance of a small Bayes factor (e.g. ln BF < 1) or ignoring the estimation error on the evidence (dlnZ).

### Realistic Astronomy Examples
Bayes Factor: ln BF = 6.2 +/- 0.1, indicating 'decisive' evidence for a model containing pulsar spin-down variations.