SkillHub

Skills lllllllama

rigorpilot-skills Skill collection

README-first research reproduction skills with bounded execution, auditable evidence, and byte-preserving README annotations.

Source on GitHubSort and filter them in the skills list

486 stars on GitHub, Nov 9, 2026 · 11 skills · 1 bundle

Its bundle

  1. RigorPilot

    Bundle of 11 skills
    Use it when:
    You want an agent to run a deep learning research repo from its README and record honestly what worked.
    It does:
    Finds the smallest documented run, such as one evaluation command, sets it up, runs it and audits the result, instead of a full training run.

    ai-research-reproduction, repo-intake-and-plan, paper-context-resolver, env-and-assets-bootstrap, analyze-project, minimal-run-and-audit, run-train, safe-debug, ai-research-explore, explore-code, explore-run

    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    493K
    #38
    Weekly Downloads
    7.3K
    #72

Its skills, most installed first

  1. paper-context-resolver

    Part of RigorPilot
    Use it when:
    You're reproducing a paper's code and the README leaves one specific question open, such as a dataset split or evaluation protocol.
    It does:
    Looks up that one detail in the paper's primary sources: the README's paper link, the project page, arXiv or OpenReview.
    type:workfloweffect:noneneeds:noneml-research
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    493K
    #38
    Weekly Downloads
    7.3K
    #76

  2. repo-intake-and-plan

    Part of RigorPilot
    Use it when:
    You're starting to reproduce a deep-learning repo and want it mapped before anything runs.
    It does:
    Reads the README and setup files, lists the documented commands and sorts them into inference, evaluation, training or other.
    type:workfloweffect:noneneeds:noneml-researchplanningstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    492K
    #39
    Weekly Downloads
    7.3K
    #80

  3. minimal-run-and-audit

    Only works as part of RigorPilot
    Use it when:
    Setup is done and you want a short documented run, such as a smoke test, inference or evaluation, with its evidence on record.
    It does:
    Runs that one command, or takes output already captured, and writes standard report files about it.
    type:workfloweffect:localneeds:noneml-researchverificationstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    492K
    #40
    Weekly Downloads
    7.3K
    #78

  4. env-and-assets-bootstrap

    Part of RigorPilot
    Use it when:
    A reproduction target is chosen and the environment, datasets and model checkpoints must be ready before the first run.
    It does:
    Plans a conda environment from the README and repo files, and works out where checkpoints, datasets and caches should go.
    type:setupeffect:localneeds:climl-researchstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    492K
    #41
    Weekly Downloads
    7.3K
    #82

  5. ai-research-explore

    Only works as part of RigorPilot
    Use it when:
    You have a working baseline and want to try new research ideas on top of it, with the benchmark and evaluation fixed in advance.
    It does:
    Ranks candidate ideas, tests them one at a time on an isolated branch, and compares each with the baseline.
    type:workfloweffect:localneeds:noneml-researchresearchstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    353K
    #56
    Weekly Downloads
    7.3K
    #74

  6. analyze-project

    Part of RigorPilot
    Use it when:
    You want to understand a deep-learning research repo before changing it, without running anything heavy.
    It does:
    Maps the model structure, the training, inference and evaluation entry points and how configs connect, and points out where changes could plug in.
    type:workfloweffect:localneeds:noneml-researcharchitecturecode-reviewstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    353K
    #57
    Weekly Downloads
    7.3K
    #72

  7. ai-research-reproduction

    Part of RigorPilot
    Use it when:
    You want an agent to run a deep-learning research repo from its README and record honestly what worked.
    It does:
    Lists the README's commands, picks the smallest trustworthy one, usually inference or evaluation, runs it through bundled scripts and verifies the evidence.
    type:workfloweffect:localneeds:noneml-researchverificationstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    353K
    #58
    Weekly Downloads
    7.3K
    #75

  8. safe-debug

    Part of RigorPilot
    Use it when:
    A deep-learning run fails with a traceback, CUDA out-of-memory error, checkpoint load failure, shape mismatch or NaN loss.
    It does:
    Reads the error, narrows down the likely cause with evidence, and proposes the smallest safe fix without editing code.
    type:workfloweffect:localneeds:noneml-researchdebuggingstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    352K
    #59
    Weekly Downloads
    7.3K
    #77

  9. explore-code

    Only works as part of RigorPilot
    Use it when:
    You've authorized trying a code change on a research model: transplanting a module, adapting a backbone, replacing a head, or adding LoRA or adapter layers.
    It does:
    Makes the change on an isolated branch or worktree, copying from the source repo rather than rewriting, and records why it matters and how to undo it.
    type:workfloweffect:localneeds:noneml-researchimplementationstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    352K
    #60
    Weekly Downloads
    7.3K
    #79

  10. run-train

    Only works as part of RigorPilot
    Use it when:
    A training command is already chosen and you want it started, checked over a short run, kicked off in full, or resumed.
    It does:
    Runs the command through RigorPilot's shared runtime and records the command, config, seed, logs, checkpoints and metrics.
    type:workfloweffect:localneeds:noneml-researchverificationstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    352K
    #61
    Weekly Downloads
    7.3K
    #81

  11. explore-run

    Only works as part of RigorPilot
    Use it when:
    You've authorized quick experiments on a research baseline: small-subset checks, short training probes, batch sweeps, idle-GPU searches or transfer-learning trials.
    It does:
    Plans candidate runs from the variants, subset sizes and step counts you set, ranks them by cost, chance of success and expected gain, and trims them to budget.
    type:workfloweffect:localneeds:noneml-researchstack:python
    Age
    5 weeks
    Stars
    486
    bottom 12%
    Total Downloads
    352K
    #62
    Weekly Downloads
    7.3K
    #83