AI Research Reproduction (RigorPilot)
README-first reproduction of deep-learning repos — smallest trustworthy target (documented inference → eval → training startup), conservative patch rules, standardized evidence bundle in repro_outputs/
- What
- README-first reproduction of deep-learning repos — smallest trustworthy target (documented inference → eval → training startup), conservative patch rules, standardized evidence bundle in repro_outputs/
- Cost
- Free
- Needs
- Python 3.11+ and git for the bundled orchestration scripts. Target repositories may require reviewed dependencies, network access, or accelerators.
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — the RigorPilot "Rigor Reproduce" skill for README-first deep-learning repository reproduction: an end-to-end, minimal-trustworthy flow that reads the repository first, selects the smallest documented inference or evaluation target, coordinates intake, setup, trusted execution, optional trusted training, optional analysis and paper-gap resolution — then records evidence, assumptions, deviations and human decision points in a standardized `repro_outputs/` bundle. The fast path is concrete: read the target README, run `scripts/orchestrate_repro.py` in plan mode to review command candidates, selection fingerprint and side-effect contract, run the selected candidate, then verify outputs — with explicit user timeout bounds preserved, never silently tightened. Trusted target selection prefers documented inference first, then documented evaluation, then training startup or partial verification, and full training only after explicit user confirmation; the README is primary reproduction intent — repo files clarify it but never silently replace it, and README/paper conflicts get recorded, not manufactured into success. The patch boundary is conservative: prefer CLI args, env vars and dependency fixes before code changes; reproduction fixes must be stated openly (what changed, why, whether scientific meaning or comparability shifted); never change model architecture, core inference semantics or loss functions; if repo files must change, branch `repro/YYYY-MM-DD-short-task` and keep patch commits sparse and recorded in `PATCHES.md`. Outputs are `SUMMARY.md`, `COMMANDS.md`, `LOG.md`, `SCIENTIFIC_CHANGELOG.md`, `COMPARABILITY_REPORT.md`, `status.json`, `ANNOTATED_README.md` (original README preserved byte-for-byte outside evidence blocks) and `PATCHES.md` only if patched. By @lllllllama, listed here with credit to its creator. Honest caveats: it executes real commands on real repos — always review the plan-mode side-effect contract before running; target repos may need reviewed dependencies, network access or accelerators; the bundled orchestration needs Python 3.11+ and git; "result-match" is awarded only against explicit expected metrics and tolerance, so unclear README expectations mean unclear verdicts. Skill Harbor never reviews the code, review it yourself before use.
Version:
Install
Prerequisites: Python 3.11+ and git for the bundled orchestration scripts. Target repositories may require reviewed dependencies, network access, or accelerators. Install "AI Research Reproduction (RigorPilot)" for me. The RigorPilot Rigor Reproduce skill: README-first deep-learning reproduction flow — reads the repo, selects the smallest documented inference/evaluation target, runs it conservatively with explicit patch boundaries, and records everything in a standardized repro_outputs/ evidence bundle. Repository: https://github.com/lllllllama/rigorpilot-skills/blob/main/skills/ai-research-reproduction/SKILL.md 1. Fetch the SKILL.md file (and any helper files) from the repository path into a temporary folder and summarize what it does in one or two sentences. 2. Safety check: review the SKILL.md and scripts for anything suspicious (unexpected network calls, shell commands, credential harvesting). This repo should contain zero secrets in code, credentials only via the secure vault. Verify that holds here; STOP on any red flag and tell me. 3. Install it as a skill: copy SKILL.md and its helper files into the agent's skills directory, in a folder named "rigorpilot-ai-research-reproduction". 4. Verify with no network calls: frontmatter valid, files in place. 5. Report what was installed, where, and what I still need to do myself. GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.
Questions
How do I install a build?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the build up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a build before installing it.