- Home
- Skills
- Research & Analysis
- FisGraf: Physics Error Propagation, Fits & LaTeX Plots
More screenshots
Works with the AI Agents you already use
FisGraf: Physics Error Propagation, Fits & LaTeX Plots
- **Curve fitting:** any model you write (`0.5*g*t^2 + v0*t`, `A*exp(-t/tau)`…), with errors in y or in x and y.
$10
FisGraf: Physics Error Propagation, Fits & LaTeX Plots
Example session with this skill installed
Use FisGraf to compute g from a pendulum: g = 4pi^2L/T^2, with L = 1.000 ± 0.001 m and T = 2.007 ± 0.005 s (independent standard uncertainties). Tell me which measurement dominates the error and compare the methods.
- Read your context and instructions
- Compiled the fisgraf
g = 9.80 ± 0.05 m/s² (relative error 0.51 %, 95 % interval 9.70 to 9.90).
The period dominates the uncertainty, so better timing is what would reduce the error.
Uncertainty budget
| Variable | Value | σ | Contribution |
|---|---|---|---|
| T | 2.007 s | 0.005 s | 96.1 % |
| L | 1.000 m | 0.001 m | 3.9 % |
Method comparison
| Method | Value | σ |
|---|---|---|
| Linear (GUM) | 9.8009 | 0.0498 |
| Second order | 9.8011 | 0.0498 |
| Monte Carlo | 9.8011 | 0.0498 |
| Kragten | 9.8009 | 0.0498 |
| Worst-case bound | 9.8009 | 0.0586 |
The methods agree, so the linear result is reliable here.
LaTeX for your report
$g = \left(9.80 \pm 0.05\right)\,\unit{m/s^2}$
\sigma_{g} = \sqrt{\left(\frac{4 \pi^{2}}{T^{2}}\right)^{2} \sigma_{L}^{2} + \left(-\frac{8 \pi^{2} L}{T^{3}}\right)^{2} \sigma_{T}^{2}}
Connects securely to your tools. The creator never sees your data.
About this skill
FisGraf turns your AI agent into a physics lab assistant. Instead of improvising calculations in Python, the agent uses fisgraf-cli, a Rust engine validated against NIST certified reference data, so you get the same numbers every time, rounded according to the GUM.
What it does
- Error propagation: value of f(x₁, …, xₙ) with its uncertainty from measurements, constants and correlations. Five methods (linear GUM, second order, Monte Carlo GUM-S1, Kragten, maximum bound), symbolic error function in LaTeX and an uncertainty budget showing which variable dominates.
- Curve fitting: any model you write (
0.5*g*t^2 + v0*t,A*exp(-t/tau)…), with errors in y or in x and y. Linear, Levenberg–Marquardt, Gauss–Newton, Nelder–Mead, effective variance and ODR. Returns parameters ± σ, R², χ²_red, p-value, AIC/BIC, correlations and residuals. - Scientific plots: points with error bars, fitted curves, reference functions, confidence bands, results box with equation and R², residual panel and decimal comma. Exports PDF, PNG, SVG, a full .tex or a tikzpicture ready to \input{} in your report.
- Reads data from Excel (.xlsx, .xls, .ods), CSV or directly from the conversation.
How it works
- Locate or install the
fisgraf-cli.exebinary in the local environment or download it from the official repository. - Prepare a JSON payload containing measurements, uncertainties, and the mathematical model or equation.
- Execute commands like
erroresfor propagation,ajustarfor curve fitting, orgraficofor visual output. - Review results including χ²_red, p-values, uncertainty budgets, and generated image files.
The agent checks the plot before delivering it, explains what χ²_red means for your data and gives you LaTeX-ready results for your lab report.
Includes
SKILL.md, a full reference of the project format and 8 example experiments (free fall, Ohm's law, capacitor charging, damped oscillator, Newton cooling, RLC resonance, photoelectric effect, Boyle's law).
Frameworks & tools
Built with Rust for performance and accuracy. Requires fisgraf-cli (Windows 10/11, no installation needed) and, for plots only, a LaTeX distribution like MiKTeX or TeX Live. Error propagation and fitting work without LaTeX. Documentation and tool messages are in Spanish; the agent answers in your language.
License
Please cite: Álvarez de Araya, M. (2026). FisGraf y FisErr (v1.0). https://github.com/malvarezdam/FisGraf-FisErr
Requires local LaTeX (MiKTeX/TeX Live) for PNG/PDF plots. Binary is Windows-based. Non-linear models need initial parameter guesses to avoid convergence issues.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 8 days ago
Requires
Permissions
Allowed Hosts
Terminal runs the local fisgraf-cli program. Files are read for input data and written for JSON inputs and plots. Network access is used only once, with the user's permission, to download fisgraf-cli from GitHub; its SHA-256 is verified before it runs. Calculations run locally.