Explore Code (RigorPilot)
Bounded candidate code changes on isolated branches — module transplants, backbone adaptation, LoRA insertion — with rollback-aware records, never touching the trusted baseline
- What
- Bounded candidate code changes on isolated branches — module transplants, backbone adaptation, LoRA insertion — with rollback-aware records, never touching the trusted baseline
- Cost
- Free
- Needs
- a deep-learning repo you control (git branching), and explicit researcher authorization for exploratory code changes at use time — this skill only modifies code on isolated branches/worktrees.
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — the RigorPilot "Rigor Improve" implementation leaf skill for auditable candidate implementation in deep-learning research repositories. It applies when the researcher explicitly authorizes exploratory code changes on an isolated branch or worktree: source-anchored module transplants, backbone adaptation, LoRA or adapter layer insertion, head replacement, or stitching together meaningful low-risk migration ideas. The governing principle is isolation — exploratory code modifications stay quarantined from the trusted baseline, favoring source-anchored copying and minimal adaptation over freeform rewrites, with a clear record of why a candidate change is meaningful, how to roll it back, and why it remains a candidate rather than a verified contribution. Execution may hand off to `minimal-run-and-audit` or `run-train`; the skill itself owns the code-change side and writes `explore_outputs/` records (`CHANGESET.md`, `SCIENTIFIC_CHANGELOG.md`, `COMPARABILITY_REPORT.md`, `TOP_RUNS.md`, `status.json`). Narrow fits stay narrow: end-to-end exploration orchestration belongs to `ai-research-explore`, trusted baseline work and conservative debugging belong elsewhere. By @lllllllama, listed here with credit to its creator. Honest caveats: exploratory changes only after explicit researcher authorization — never a default for unapproved repo edits; everything produced stays a candidate, never a verified contribution, until evidence says otherwise; it's a code-change leaf, so it relies on sibling skills (or your own runs) for execution evidence. Skill Harbor never reviews the code, review it yourself before use.
Version:
Install
Prerequisites: a deep-learning repo you control (git branching), and explicit researcher authorization for exploratory code changes at use time — this skill only modifies code on isolated branches/worktrees. Install "Explore Code (RigorPilot)" for me. The RigorPilot implementation leaf skill: bounded candidate code changes (module transplants, backbone adaptation, LoRA insertion) on isolated branches, with rollback-aware records, summary-level logging, and strict separation from the trusted baseline. Repository: https://github.com/lllllllama/rigorpilot-skills/blob/main/skills/explore-code/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-explore-code". 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.