cli-anything-openrefine — Claude Code skill: what it does and setup
cli-anything-openrefine controls OpenRefine through an agent CLI to import messy datasets, apply JSON transformations, and export cleaned CSV or TSV files.
Apache-2.0 249 installs ★ 62,125 updated 2026-09-22
npx skills add HKUDS/CLI-Anything --skill cli-anything-openrefine -a claude-codeWhat cli-anything-openrefine does
This skill connects an agent to OpenRefine through the terminal. It handles project creation, row inspection via data rows, and exporting results to CSV or TSV. Transformations run through GREL expressions or existing JSON history files.
Setup requires two local pieces. First, install the harness in openrefine/agent-harness using pip. Next, launch the backend with openrefine -i 127.0.0.1 -p 3333. Set custom endpoints using OPENREFINE_URL or the --base-url flag.
Task state stays isolated with --session. Without an active OpenRefine server, backend commands will fail immediately. Piped commands through stdin convert Unicode to reversible escapes, returning a non-zero exit code on failure.
What it helps with
- import messy CSV and TSV tables into OpenRefine
- execute JSON operation histories across table columns
- inspect table rows using the limit flag
- undo or redo session steps during data processing
- export cleaned tables to CSV or TSV format
Example prompts for the agent
- Import
messy.csvinto OpenRefine and save the session torun/session.json. - Run a text transform on the Name column using expression
value.trim(). - Export the cleaned dataset from
run/session.jsontorun/clean.csv.
How to install cli-anything-openrefine
1Skills CLI (Vercel)
npx skills add HKUDS/CLI-Anything --skill cli-anything-openrefine -a claude-code2Manually: put the files in place
~/.claude/skills/cli-anything-openrefine/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/cli-anything-openrefine/
1Skills CLI (Vercel)
npx skills add HKUDS/CLI-Anything --skill cli-anything-openrefine -a codex2Manually: put the files in place
.agents/skills/cli-anything-openrefine/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add HKUDS/CLI-Anything --skill cli-anything-openrefine -a cursor2Manually: put the files in place
.cursor/skills/cli-anything-openrefine/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install HKUDS/CLI-Anything skills/cli-anything-openrefine2Manually: put the files in place
.github/skills/cli-anything-openrefine/SKILL.md1Skills CLI (Vercel)
npx skills add HKUDS/CLI-Anything --skill cli-anything-openrefine -a gemini-cli2Platform CLI
gemini skills install https://github.com/HKUDS/CLI-Anything --path skills/cli-anything-openrefineManually: put the files in place
.devin/skills/cli-anything-openrefine/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add HKUDS/CLI-Anything --skill cli-anything-openrefine -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about cli-anything-openrefine
How to install cli-anything-openrefine in Claude Code?
Run npx skills add HKUDS/CLI-Anything --skill cli-anything-openrefine -a claude-code. The skill file will be created at ~/.claude/skills/cli-anything-openrefine/SKILL.md.
What prerequisites are required before using this skill?
Install the harness via python -m pip install -e . in openrefine/agent-harness. Then start the local instance using openrefine -i 127.0.0.1 -p 3333.
How does the skill handle errors in automated scripts?
Adding --json makes errors output a JSON payload to stderr with ok: false. Piped stdin runs exit with a non-zero code on failure.
Source text
SKILL.md · Apache-2.0
--- name: "cli-anything-openrefine" description: "Use OpenRefine through an agent-native CLI for importing messy data, applying JSON operation histories, inspecting rows, exporting cleaned data, and managing session undo/redo." contributor: "CLI-Anything-Team" --- # CLI-Anything OpenRefine Use this skill when a task needs OpenRefine data cleaning, transformation, reusable operation histories, or CSV/TSV export from an automated agent workflow. ## Prerequisites Install the harness: ```bash cd openrefine/agent-harness python -m pip install -e . ``` Start OpenRefine before backend commands: ```bash openrefine -i 127.0.0.1 -p 3333 ``` Set a custom server with `OPENREFINE_URL=http://127.0.0.1:3333` or pass `--base-url`. ## Command Rules For Agents - Prefer `--json` on every one-shot command. - Use `--session <path>` for isolated task state. - Import or open a project before row, apply, export, undo, or redo commands. - Existing OpenRefine operation-history JSON can be passed directly to `data apply`. - Generated files are normal OpenRefine operation JSON and exported CSV/TSV data. ## Common Commands ```bash cli-anything-openrefine --json server ping cli-anything-openrefine --json project list cli-anything-openrefine --json --session run/session.json project import messy.csv --name cleanup cli-anything-openrefine --json --session run/session.json data rows --limit 10 cli-anything-openrefine --json ops text-transform run/trim.json --column Name --expression 'value.trim()' cli-anything-openrefine --json --session run/session.json data apply run/trim.json cli-anything-openrefine --json --session run/session.json data export run/clean.csv --format csv cli-anything-openrefine --json --session run/session.json session undo cli-anything-openrefine --json --session run/session.json session redo ``` ## REPL Run `cli-anything-openrefine` with no subcommand to enter the REPL. For automated user journeys, pipe newline-separated REPL commands through stdin. Redirected streams automatically use an ASCII-only input/output path while interactive terminals retain the Unicode banner, prompt history, and styling: ```bash printf 'help\nexit\n' | cli-anything-openrefine ``` Unicode payloads are rendered as reversible ASCII backslash escapes when the REPL is redirected, preventing legacy Windows output encodings from turning a successful command into a failed journey. Piped workflows return a nonzero exit status at `exit` or EOF if any command failed, allowing CI to detect an unsuccessful user journey. ## Error Handling When `--json` is set, command failures write a JSON object to stderr with `ok: false`.
© HKUDS · Apache-2.0 · HKUDS/CLI-Anything