codeclaw

Export Claude Code and Codex conversation history to Hugging Face as structured training data. Use when the user asks about exporting conversations, uploading to Hugging Face, configuring codeclaw, reviewing PII/secrets in exports, or managing their dataset.

9 stars

Best use case

codeclaw is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Export Claude Code and Codex conversation history to Hugging Face as structured training data. Use when the user asks about exporting conversations, uploading to Hugging Face, configuring codeclaw, reviewing PII/secrets in exports, or managing their dataset.

Teams using codeclaw should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/docs/SKILL.md --create-dirs "https://raw.githubusercontent.com/ychampion/codeclaw/main/docs/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/docs/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How codeclaw Compares

Feature / AgentcodeclawStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Export Claude Code and Codex conversation history to Hugging Face as structured training data. Use when the user asks about exporting conversations, uploading to Hugging Face, configuring codeclaw, reviewing PII/secrets in exports, or managing their dataset.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

Related Guides

SKILL.md Source

<!-- codeclaw-begin -->

# codeclaw Skill

## THE RULE

**Every `codeclaw` command outputs `next_steps`. FOLLOW THEM.**

Do not memorize the flow. Do not skip steps. Do not improvise.
Run the command -> read the output -> follow `next_steps`. That's it.

The CLI tracks your stage (1-4: auth -> configure -> review -> done).
`codeclaw export` (push) is **gated** -- you must run `codeclaw confirm` first or it will refuse.

## Getting Started

Run `codeclaw status` (or `codeclaw prep` for full details) and follow the `next_steps`.

## Output Format

- `codeclaw prep`, `codeclaw config`, `codeclaw status`, and `codeclaw confirm` output pure JSON
- `codeclaw export` outputs human-readable text followed by `---CODECLAW_JSON---` and a JSON block
- Always parse the JSON and act on `next_steps`

Key fields:
- `stage` / `stage_number` / `total_stages` -- where you are
- `next_steps` -- follow these in order
- `next_command` -- the single most important command to run next (null if user input needed first)

## PII Audit (Stage 3)

After `codeclaw export --no-push`, follow the `next_steps` in the JSON output. The flow is:

1. **Ask the user their full name** -- then grep the export for it
2. **Run the pii_commands** from the JSON output and review results with the user
3. **Ask the user what else to look for** -- company names, client names, private URLs, other people's names, custom domains
4. **Deep manual scan** -- sample ~20 sessions (beginning, middle, end) and look for anything sensitive the regex missed
5. **Fix and re-export** if anything found: `codeclaw config --redact "string"` then `codeclaw export --no-push`
6. **Run `codeclaw confirm` with text attestations** -- pass `--full-name`, `--attest-full-name`, `--attest-sensitive`, and `--attest-manual-scan`. It runs PII scan, verifies attestations, shows project breakdown, and unlocks pushing.
7. **Push only after explicit user confirmation**: `codeclaw export --publish-attestation "User explicitly approved publishing to Hugging Face."`

## Commands Reference

```bash
codeclaw status                            # Show current stage and next steps (JSON)
codeclaw prep                              # Discover projects, check HF auth (JSON)
codeclaw setup                             # Guided onboarding (HF, dataset, projects, MCP, watcher)
codeclaw prep --source both                # Claude + Codex sessions
codeclaw prep --source codex               # Only Codex sessions
codeclaw prep --source claude              # Only Claude Code sessions
codeclaw confirm --full-name "NAME" --attest-full-name "..." --attest-sensitive "..." --attest-manual-scan "..." # Scan PII, verify attestations, unlock pushing (JSON)
codeclaw confirm --file /path/to/file.jsonl --full-name "NAME" --attest-full-name "..." --attest-sensitive "..." --attest-manual-scan "..." # Confirm a specific export file
codeclaw list                              # List all projects with exclusion status
codeclaw projects                          # Show connected project scope
codeclaw projects --connect "proj1,proj2" # Connect specific projects
codeclaw projects --use-current            # Connect only current project
codeclaw diff                              # Show what redaction pipeline will remove before confirm
codeclaw stats --skill                     # Show growth-oriented trajectory metrics
codeclaw list --source both                # List Claude + Codex projects
codeclaw list --source codex               # List only Codex projects
codeclaw config                            # Show current config
codeclaw config --repo user/my-personal-codex-data  # Set HF repo
codeclaw config --source both              # REQUIRED source scope: claude|codex|both
codeclaw config --exclude "a,b"            # Add excluded projects (appends)
codeclaw config --redact "str1,str2"       # Add strings to redact (appends)
codeclaw config --redact-usernames "u1,u2" # Add usernames to anonymize (appends)
codeclaw config --confirm-projects         # Mark project selection as confirmed
codeclaw config --encryption status        # Check encryption-at-rest state
codeclaw export --publish-attestation "..." # Export and push (requires codeclaw confirm first)
codeclaw export --no-push                  # Export locally only
codeclaw export --source both --no-push    # Export Claude + Codex sessions
codeclaw export --source codex --no-push   # Export only Codex sessions
codeclaw export --source claude --no-push  # Export only Claude Code sessions
codeclaw export --dry-run                  # Preview export/publish plan without writing files
codeclaw export --all-projects             # Include everything (ignore exclusions)
codeclaw export --no-thinking              # Exclude extended thinking blocks
codeclaw export -o /path/to/file.jsonl     # Custom output path
codeclaw watch --status                    # Watcher status with transparency fields
codeclaw watch --logs --follow             # Stream daemon logs
codeclaw watch --monitor --follow          # Live watcher monitor (status + recent logs)
codeclaw watch --pause                     # Pause watcher without stopping process
codeclaw watch --resume                    # Resume watcher polling
codeclaw watch --switch-project "project"  # Scope watcher to one connected project
codeclaw console --source both             # Open interactive slash-command console
codeclaw                                  # Open full-screen TUI by default
codeclaw tui --source both                 # Open full-screen TUI mode
codeclaw update-skill claude               # Install/update the codeclaw skill for Claude Code
```

## Gotchas

- **Never run bare `huggingface-cli login`** -- it's interactive and will hang. Always use `--token`.
- **`--exclude`, `--redact`, `--redact-usernames` APPEND** -- they never overwrite. Safe to call repeatedly.
- **Source selection is REQUIRED before export** -- explicitly set `codeclaw config --source claude|codex|both` (or pass `--source ...` on export).
- **`codeclaw prep` outputs pure JSON** -- parse it directly.
- **Always export with `--no-push` first** -- review before publishing.
- **`codeclaw export` (push) requires `codeclaw confirm` first** -- it will refuse otherwise. Re-exporting with `--no-push` resets this.
- **PII audit is critical** -- automated redaction is not foolproof.
- **Large exports take time** -- 500+ sessions may take 1-3 minutes. Use a generous timeout.

## Prerequisite

`command -v codeclaw >/dev/null 2>&1 && echo "codeclaw: installed" || echo "NOT INSTALLED -- run: pip install codeclaw"`

<!-- codeclaw-end -->

Related Skills

swe-cli-skills

12
from SylphAI-Inc/skills

Senior engineer CLI expertise for AI agents — workflows, safety guardrails, gotchas, and anti-patterns across cloud, IaC, containers, databases, dev tools, and platforms

DevOps & Infrastructure

PicoClaw Fleet

11
from EricGrill/agents-skills-plugins

Orchestrate a fleet of remote PicoClaw workers over SSH for fast, ephemeral one-shot tasks.

DevOps & Infrastructure

VibeCollab — Setup Instructions for AI Assistants

9
from flashpoint493/VibeCollab

You are helping a user set up VibeCollab in their project.

Workflow & Productivity

raycast-extension-docs

9
from lemikeone/Codex-skill-raycast-extension

Guidance for building, debugging, and publishing Raycast extensions using the Raycast documentation set. Use when Codex needs to create or modify Raycast extensions (React/TypeScript/Node), consult Raycast API reference or UI components, build AI extensions, handle manifest/lifecycle/preferences, troubleshoot issues, or prepare/publish extensions to the Raycast Store or Teams.

Coding & Development

evomap

9
from hyz0906/paper

Connect to the EvoMap collaborative evolution marketplace. Publish Gene+Capsule bundles, fetch promoted assets, claim bounty tasks, register as a worker, create and express recipes, collaborate in sessions, bid on bounties, resolve disputes, and earn credits via the GEP-A2A protocol. Use when the user mentions EvoMap, evolution assets, A2A protocol, capsule publishing, agent marketplace, worker pool, recipe, organism, session collaboration, or service marketplace.

AI Agent Marketplace

maestro

8
from Viniciuscarvalho/maestro

Intelligent skill knowledge gateway. Routes tasks to the right knowledge without loading all skills into context. MUST be consulted before any coding task — call the search_skills MCP tool to retrieve relevant expertise from 100+ indexed skills covering Swift, SwiftUI, concurrency, testing, architecture, performance, and security.

Coding & Development

opentui

7
from LeonardoTrapani/better-skills

Comprehensive OpenTUI skill for building terminal user interfaces. Covers the core imperative API, React reconciler, and Solid reconciler. Use for any TUI development task including components, layout, keyboard handling, animations, and testing.

Coding & Development

calm-ui

7
from brijr/vibe

Apply a restrained, Swiss/Japanese/Scandinavian/German-influenced product design system when building or refining UI in React, Next.js, TypeScript, and shadcn/ui. Use when the user asks to build, refine, critique, redesign, or review a page, screen, component, form, table, dashboard, layout, or other frontend interface, especially in projects using shadcn/ui. Do not use for marketing sites, landing pages, non-UI work, or requests for bold, playful, maximalist, or otherwise expressive aesthetics.

Frontend Development

solid

7
from fellipeutaka/denji

Apply SOLID principles to write flexible, maintainable, and testable code. Use when designing classes, interfaces, and module boundaries. Covers Single Responsibility, Open/Closed, Liskov Substitution, Interface Segregation, and Dependency Inversion with practical TypeScript examples and detection heuristics.

netops-asset-manager

7
from Boos4721/netops-asset-manager-skill

Manage IT infrastructure assets (routers, switches, servers, GPU clusters) through a Go + Vue 3 platform with real-time health probing, SSH remote control, configuration backup, bulk import, network topology visualization, and PM2 process management. Supports H3C, Huawei, Cisco, MikroTik, Ruijie, DCN, and Linux. Use when the user asks about IT asset management, network device operations, infrastructure monitoring, SSH device control, or development on this Go + Vue 3 platform.

Goal: Build an LLM-based RAG App

6
from Harmeet10000/skills

Here is the MVP Implementation Plan.

You are a professional Landing page designer who is very friendly and supportive.

6
from Harmeet10000/skills

Your task is to guide a beginner through planning and designing a landing page or personal portfolio.

Workflow & Productivity