Skip to main content
affpapa.org
RU EN
skill Claude Code CX Codex Cursor GitHub Copilot

explore-code — Claude Code skill: auditable trial edits in research

A skill for cautious code experiments in deep learning repositories: it transplants modules and adapts architectures on an isolated branch, with rollback-aware records.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓ 
npx skills add lllllllama/RigorPilot-Skills --skill explore-code -a claude-code
Open on GitHub

What explore-code does

The skill handles trial code changes in a research repository. It copies a source module and adapts it minimally rather than rewriting from scratch: swap a backbone, insert a LoRA or adapter layer, replace a model head. Work happens on an isolated branch or worktree, so the trusted baseline stays untouched.

Output lands in explore_outputs/: CHANGESET.md, SCIENTIFIC_CHANGELOG.md, COMPARABILITY_REPORT.md, TOP_RUNS.md and status.json. The records state why a change matters, how to roll it back, and why it is still a candidate rather than a verified contribution. scripts/plan_code_changes.py and scripts/write_outputs.py handle planning and writing, and references/explore-policy.md holds the exploration policy.

The researcher has to explicitly authorize exploratory edits. Without that, the skill does not apply. It is not meant for trusted baseline reproduction, conservative debugging, environment setup or a broad refactor. When a task spans both current_research coordination and exploratory runs, use ai-research-explore instead; execution can be handed off to minimal-run-and-audit or run-train.

What it helps with

  • transplant a module from another repo with minimal adaptation
  • swap the backbone and see how training reacts
  • insert a LoRA or adapter layer into an existing model
  • replace the model head for a new task
  • record a trial change so it can be rolled back

Example prompts for the agent

  • Transplant this module into our repo on a separate branch and log the changes in explore_outputs/
  • Add a LoRA layer to the current model but leave the trusted baseline alone
  • Replace the model head and write the comparability report into explore_outputs/

How to install explore-code

1Skills CLI (Vercel)

npx skills add lllllllama/RigorPilot-Skills --skill explore-code -a claude-code

2Manually: put the files in place

~/.claude/skills/explore-code/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/explore-code/

From the same set9

The whole set →
repo-intake-and-plan lllllllama Scans a repository through its README and project files, then builds a small reproduction plan. Use it before running experiments. Claude Code CX Codex Cursor +1 474,500installs minimal-run-and-audit lllllllama Captures and normalizes execution evidence from deep learning smoke tests and inference commands into standardized repro_outputs/ files. Claude Code CX Codex Cursor +1 474,400installs env-and-assets-bootstrap lllllllama Prepares environment and checkpoint/dataset paths before the first run of a README-documented repo. Claude Code CX Codex Cursor +1 474,300installs ai-research-explore lllllllama Governs deep learning research exploration on top of an existing baseline. It isolates candidate experiments and generates structured comparison reports. Claude Code CX Codex Cursor +1 335,600installs analyze-project lllllllama Reads a deep learning repository and maps its structure, entrypoints and suspicious spots. Changes nothing, runs no heavy jobs. Claude Code CX Codex Cursor +1 335,500installs ai-research-reproduction lllllllama This skill guides a repository-grounded reproduction run from its README to an evidence bundle. Use it to test a documented model run without quietly changing the experiment. Claude Code CX Codex Cursor +1 335,300installs explore-run lllllllama Plans bounded exploratory machine learning runs with comparability caveats and saves summaries to explore_outputs. Claude Code CX Codex Cursor +1 334,700installs run-train lllllllama Runs a selected training command and organizes the resulting evidence. Use it for startup checks, short-run verification, full training kickoff, or resume handling. Claude Code CX Codex Cursor +1 334,700installs

Similar

eval-harness affaan-m A skill for eval-driven development: it sets pass/fail criteria before you write code and tracks agent reliability with pass@k and pass^k. Claude Code CX Codex Cursor +1 4,700installs skill-creator official anthropics Creates and refines skills, then tests them with prompts, evaluations, and performance checks. Claude Code Claude.ai 414,180installs writing-for-agents Matt Pocock Guidance for writing skills and agent-facing documents. Use it when editing AGENTS.md, CLAUDE.md, or reference files reached from them. Claude Code 466,719installs media-use heygen-com Finds or generates media for HyperFrames videos, including voiceovers, music, cutouts, and color grades. Stores files locally in assets. Claude Code CX Codex Cursor +1 1,100,000installs mcp-builder official anthropics This skill walks through building an MCP server in TypeScript or Python, from reading the protocol spec to writing tests and evaluation questions. Claude Code Claude.ai 130,949installs subagent-driven-development Jesse Vincent This skill runs an implementation plan through fresh subagents for independent tasks and reviews each result. Use it for multi-step work in one session. Claude Code CX Codex 237,746installs orchestration stablyai Coordinates supervised Orca workers with task DAGs, decision gates, and threaded messages. Fits workflows that split work across multiple agents. Claude Code CX Codex Cursor +1 349,200installs

Questions about explore-code

How do I install explore-code in Claude?

Run npx skills add lllllllama/RigorPilot-Skills --skill explore-code -a claude-code. The skill lands at ~/.claude/skills/explore-code/SKILL.md.

What does explore-code do?

It makes trial code changes in a deep learning repository on an isolated branch or worktree and records them in explore_outputs/ with a rollback plan.

When should I not use explore-code?

When exploratory edits are not explicitly authorized, and for trusted baseline reproduction, conservative debugging, environment setup or a broad refactor.

Source text

SKILL.md · MIT 
---
name: explore-code
description: Rigor Improve implementation leaf skill for auditable candidate implementation in deep learning research repositories. Use when the researcher explicitly authorizes exploratory work on an isolated branch or worktree to transplant modules, adapt a backbone, add LoRA or adapter layers, replace a head, or stitch together meaningful low-risk migration ideas with rollback-aware records in `explore_outputs/`. Do not use for end-to-end exploration orchestration on top of `current_research`, trusted baseline reproduction, conservative debugging, environment setup, verified contribution claims, or default repository analysis.
---

# explore-code

Use this as the Rigor Improve implementation leaf skill. The installed slug
remains `explore-code` for compatibility.

Use the shared operating principles in
`../ai-research-reproduction/references/agent-operating-principles.md`; this skill should guide
bounded candidate code work without over-prescribing implementation details.

## When to apply

- When the researcher explicitly authorizes exploratory code changes on an isolated branch or worktree.
- When the task is source-anchored module transplant, backbone adaptation, LoRA or adapter insertion, or low-risk module combination.
- When summary-level recording is sufficient and the result is a candidate, not a trusted conclusion.

## When not to apply

- When the request is for trusted baseline work, conservative debugging, or normal training execution.
- When the user did not explicitly authorize exploratory modifications.
- When the task is a broad refactor or a from-scratch idea implementation.

## Clear boundaries

- This skill owns exploratory code modifications only.
- It must keep work isolated from the trusted baseline.
- Use `ai-research-explore` instead when the task spans both current_research coordination and exploratory runs.
- It may hand off execution to `minimal-run-and-audit` or `run-train`.
- It should favor source-anchored copying and minimal adaptation over freeform rewrites.
- It should record why a candidate change is meaningful, how to roll it back,
  and why it remains a candidate rather than a verified contribution.

## Output expectations

- `explore_outputs/CHANGESET.md`
- `explore_outputs/SCIENTIFIC_CHANGELOG.md`
- `explore_outputs/COMPARABILITY_REPORT.md`
- `explore_outputs/TOP_RUNS.md`
- `explore_outputs/status.json`

## Notes

Use `references/explore-policy.md`, `../ai-research-reproduction/references/research-rigor-principles.md`, `scripts/plan_code_changes.py`, and `scripts/write_outputs.py`.

© lllllllama · MIT · lllllllama/RigorPilot-Skills

See also

Project sponsors

Companies that keep this analytics open
Gold sponsor
Nutra.Club

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor