# AI use cases for Bootloops (BootLoops-ai/bootloops) Repo: [BootLoops-ai/bootloops](https://github.com/BootLoops-ai/bootloops) | Use cases: 4 | Page: https://dedupedai.com/use-cases/bootloops-ai-bootloops/ BootLoops is a Python harness of certified scientific computing tools and house engines that LLM agents can run for exact and high-precision physics and quantitative science. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Validate a fresh BootLoops clone before trusting it: https://dedupedai.com/use-cases/bootloops-ai-bootloops/validate-fresh-clone/ - Reproduce the one-loop box with the Landau Alphabet engine: https://dedupedai.com/use-cases/bootloops-ai-bootloops/reproduce-landau-one-loop-box/ - Have an agent map a precision problem to the right tools: https://dedupedai.com/use-cases/bootloops-ai-bootloops/agent-plan-for-precision-problem/ - Audit a draft's prose and bibliography with the skills: https://dedupedai.com/use-cases/bootloops-ai-bootloops/audit-prose-with-skills/ ## How to validate a fresh BootLoops clone before trusting it with Bootloops Repo: [BootLoops-ai/bootloops](https://github.com/BootLoops-ai/bootloops) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Run the validation command to confirm the toolkit packages pass their own batteries on your machine. **Who it is for:** A researcher or engineer who just cloned BootLoops and needs proof the install works before relying on any result. **The problem:** A scientific toolkit is only useful if its install reproduces its own checks. Without a clear status per package, a broken install can look like a wrong result. **The outcome:** A status line for each of the 50 packages under tools/, with the packages that need missing external data named. ### Steps 1. Clone the repository: `git clone https://github.com/BootLoops-ai/bootloops` (inferred, check the README) 2. Enter the directory: `cd bootloops` (inferred, check the README) 3. Install the core dependencies: `pip install mpmath sympy numpy python-flint pytest` (inferred, check the README) 4. Run the validation command: `python3 run_selftests.py --par 8` 5. Read the status line for each package and note the 3 packages that stop with a named error about missing data ### Prompt to paste into your agent ```text In the bootloops repository, install the core Python dependencies, then run `python3 run_selftests.py --par 8`. Report how many packages come back green, which 3 stop with a named error, and what each error says is missing. Verify that no failure is caused by a Python version below 3.12. ``` ### Caveat The README says the run takes a few minutes and that 3 packages stop because they need data the repository does not include, so a non-green result is not automatically a failed install. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/bootloops-ai-bootloops/validate-fresh-clone/ ## How to reproduce the one-loop box with the Landau Alphabet engine with Bootloops Repo: [BootLoops-ai/bootloops](https://github.com/BootLoops-ai/bootloops) | Difficulty: Advanced | Time: 30 minutes | Basis: From the README Run the Landau Alphabet test to reproduce the solver's reference results for a one-loop box integral. **Who it is for:** A physicist working on loop integrals who wants a first real computation to check the toolkit against known reference values. **The problem:** Loop integrals are easy to get subtly wrong. A first run against published reference results shows whether the solver is trustworthy for your setup. **The outcome:** The solver's reference results for the one-loop box, reproduced in about a minute and a half after setup. ### Steps 1. Clone the repository: `git clone https://github.com/BootLoops-ai/bootloops` (inferred, check the README) 2. Enter the directory: `cd bootloops` (inferred, check the README) 3. Run the Landau Alphabet test: `python3 tools/landau-alphabet/test_landau_alphabet.py` 4. Compare the printed results with the solver's reference results shown in the test output 5. If the test skips, read the message, which names the engine you must install ### Prompt to paste into your agent ```text In the bootloops repository, run `python3 tools/landau-alphabet/test_landau_alphabet.py`. Report whether it reproduces the solver's reference results for the one-loop box and how long it took. Verify that the reported values match the reference values in the test, not values you computed yourself. ``` ### Caveat This is a reproduction of reference results, so it confirms the engine against known outputs and does not by itself validate a new integral you have not checked. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/bootloops-ai-bootloops/reproduce-landau-one-loop-box/ ## How to have an agent map a precision problem to the right tools with Bootloops Repo: [BootLoops-ai/bootloops](https://github.com/BootLoops-ai/bootloops) | Difficulty: Beginner | Time: 15 minutes | Basis: From the README Point an agent at the tools index so it proposes a plan for your problem and waits for your approval. **Who it is for:** A scientist with an integral from a paper, a dataset, or a published number they want checked, and no sense yet of which instrument applies. **The problem:** Choosing among dozens of specialised tools is slow, and an agent that runs them without a plan can produce results nobody can check. **The outcome:** An agent-written plan naming which instruments apply, or stating that none do, which you approve before anything runs. ### Steps 1. Clone the repository and open your agent inside the Bootloops directory (inferred, check the README) 2. Name your object: an integral from a paper, a dataset, or a published number to check 3. Give the agent this prompt: Read `tools/README.md` and the tool guides it points to, then propose a plan for this problem for me to approve. 4. Review the plan, including which instruments apply and which acceptance tests each result must pass 5. Approve the plan only if it names the checks you require, then let the agent run it ### Prompt to paste into your agent ```text Read tools/README.md and the tool guides it points to in the bootloops repository. For the problem , propose a plan listing the instruments that apply, or state that none do, with the acceptance test each result must pass. Verify that every tool you name appears in tools/README.md. ``` ### Caveat The README says an agent may find that no instrument applies, and it does not guarantee that a tool fits your problem, so the approved plan still needs your scientific judgment. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/bootloops-ai-bootloops/agent-plan-for-precision-problem/ ## How to audit a draft's prose and bibliography with the skills with Bootloops Repo: [BootLoops-ai/bootloops](https://github.com/BootLoops-ai/bootloops) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README Install the skills repository so an agent can check claims, quoted numbers, and bibliography entries. **Who it is for:** A researcher preparing a paper, report, or result page who needs every quoted number traced and every citation verified before release. **The problem:** Unsupported claims, filler, and untraceable numbers slip into drafts. Manual checking is slow and easy to skip. **The outcome:** An agent-run review of your draft that flags unsupported claims, untraceable numbers, and bibliography problems, using the skills protocols. ### Steps 1. Clone the skills repository beside Bootloops: `git clone https://github.com/BootLoops-ai/skills ../skills` (inferred, check the README) 2. Install the skills into your agent using the routes described in the skills repository 3. Point your agent at the prose-editing skill and the bibliography-verification skill 4. Provide the draft and ask the agent to list each quoted number with its source 5. Check each flagged item yourself before accepting or removing it ### Prompt to paste into your agent ```text Using the BootLoops skills repository at ../skills, apply the prose-editing and bibliography-verification skills to my draft . Report each unsupported claim, each quoted number that cannot be traced to a source, and each bibliography entry that fails verification. Verify that you did not add any claim that is absent from the draft. ``` ### Caveat The skills define the protocols, but they do not check facts on their own, so every flagged item still needs a human check against the original source. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/bootloops-ai-bootloops/audit-prose-with-skills/