Developer Documentation

Integrate the Recon Agent engine directly into your CI/CD pipelines, cron jobs, and custom applications using our official Python CLI or REST API.

Our backend provides a fully interactive Swagger UI for you to explore all available endpoints, required parameters, and response schemas.

You need a Personal Access Token (PAT) to authenticate API requests or use the CLI. Tokens inherit your tier limits and organizational roles.

Recon Agent CLI Reference

The CLI is published on the Python Package Index (PyPI). You can install it globally using standard Python tools. Requirements: Python ≥ 3.10.

Using standard Pip

pip install recon-agent

Using UV (Recommended)

uv tool install recon-agent

Before running commands, you must authenticate the CLI with your backend environment using a Personal Access Token.

# Interactive Login

$ recon login

API Base URL (https://recon-agent-i8mo.onrender.com): [ENTER]

Personal Access Token (ra_live_...): ra_live_xxxxx...

✓ Token saved securely in OS keychain.

Executes a reconciliation job between two files and immediately begins processing in the cloud.

$ recon reconcile source.csv target.csv --tolerance 0.05 --provider openai --mapping my_preset

Flag	Description
--tolerance, -t	Allowable float difference in amounts (default: 0.0)
--date-window, -d	Allowable integer difference in days (default: 0)
--provider	AI provider for fuzzy matching: `groq`, `openai`, `none`
--mapping, -m	ID of the Saved Mapping Preset to use for column extraction

List your recent reconciliation runs, or fetch the detailed exceptions of a specific run.

# List the 5 most recent runs

$ recon history --limit 5

# Fetch specific run details

$ recon history RUN_ID_12345

Triggers the AI CFO engine to analyze the exceptions of a completed run and generate a human-readable executive summary.

$ recon explain RUN_ID_12345

Downloads the detailed results of a reconciliation run to your local machine.

$ recon export RUN_ID_12345 --format csv