Works with the AI Agents you already use
Mote
A short, checked data language for agents: one-line table, file and reconciliation jobs with previewed changes and backups, instead of long pandas or SQL.
$14.99
Mote
Example session with this skill installed
Using the Mote skill, how many hours did each team spend on tickets? Then try adding hours to the team name and show me what happens.
- Read your context and instructions
- Compiled the mote
Summary
I ran two Mote programs on the sample tickets table (12 rows). The first returned the answer; the second was refused before running, with a message that says how to fix it.
Program 1
@tickets|G(team,S:hours)|O(v)
| team | hours |
|---|---|
| bill | 16.0 |
| sup | 32.0 |
Program 2 (a mistake on purpose)
@tickets|A(x=hours+team)|C
Stopped, nothing ran:
A: + - * / % ** needs numbers, got text. Use num(field) to convert text to a number
Why it matters
- One short line instead of a pandas or SQL block, so fewer tokens.
- Mistakes come back as plain fix-it messages the agent can act on, not stack traces.
- Changes only preview until you approve them, and every overwrite keeps a backup.
Next steps
- Point it at your own CSV folder or DuckDB file:
python scripts/mote_mcp.py --data <folder> - For bank-to-invoice matching see
reference/GUIDE_RECON.md.
Connects securely to your tools. The creator never sees your data.
About this skill
Agents burn tokens and make mistakes writing long pandas or SQL for routine data jobs. Mote is a tiny pipeline language: one short line, checked before it runs, with error messages that tell the agent how to fix it.
What your agent gets
- Tables, files, JSON, logs and web requests in one language
- Read-only by default; every change is previewed first; backups you can restore
- Writes limited to safe file types; web calls only to hosts you allow
- Built-in bank-to-invoice reconciliation (4 passes, refuses to guess)
- An MCP server included; works on CSV folders or a DuckDB database
Measured on our own tests About 15% fewer tokens than SQL on ordinary jobs and about 3.6x fewer than Python on bank reconciliation (rough token estimate, not a real tokenizer). A weaker model (Haiku) solved the test jobs correctly, working blind.
Honest limits Early version. No independent security review yet, and "safer than SQL" is not proven. No Postgres yet.
Licence Free for non-commercial use (PolyForm Noncommercial 1.0.0). Your purchase includes the commercial licence. Source and free non-commercial copy: github.com/ridhwaan-ai/mote-lang
The agent never needs your passwords. Web requests and file writes only happen when you approve them.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Data & Databases.
Requires Python 3.10+, pandas, and numpy. Network access restricted to hosts in a local whitelist. Only supports tabular data (CSV, DuckDB, JSON, logs).
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 17 days ago
- Software
Python
Permissions
Local by default. Terminal/Shell runs the Mote engine (python scripts/mote_mcp.py or scripts/mote_run.py). Read Files reads your CSV/JSON/log files. Write Files only happens when you approve a change (previewed first, backup kept, safe file types only, inside the data folder). Network Access: none by default. Only if a program contains a web step does it contact a host, and only hosts you list in scripts/allowed_hosts.txt (empty by default). No telemetry; nothing is sent anywhere else. The engine uses no eval/exec, no shell commands and no environment-variable reads.