- Home
- Skills
- Finance & Accounting
- agent statement reconciler
More screenshots
Works with the AI Agents you already use
agent statement reconciler
Automate agent and intercompany statement reconciliation with deterministic matching and balance proofing.
Free
agent statement reconciler
Example session with this skill installed
Reconcile the October statement for our Hamburg agent against the ledger export ledger_hamburg_oct.csv and the statement export hamburg_st_oct.txt. Account is in EUR.
- Read your context and instructions
- Compiled the agent statement reconciler
- Generated the data export
I have completed the reconciliation between the ledger export and the Hamburg agent statement. The balance proof closed with 0.00 unexplained.
Ledger Closing Balance: 2,995.00 EUR
Statement Closing Balance: -380.00 EUR
Total Difference: 3,375.00 EUR
Reconciliation Summary
| ID | Type | Effect | Reason | Action |
|---|---|---|---|---|
| E01 | Missing Document | +1,200.00 | Agent invoice #HB-99 not in ledger | Propose Accrual |
| E02 | FX Difference | +15.50 | Period-end revaluation (1.08 -> 1.09) | Propose Journal |
| E03 | Timing | -2,159.50 | Payment P-888 sent but not on statement | Await Statement |
Proposed Journals
[
{
"type": "Accrual",
"amount": 1200.00,
"currency": "EUR",
"description": "Accrue for missing agent invoice HB-99",
"status": "proposed – needs approval"
},
{
"type": "FX_Revaluation",
"amount": 15.50,
"currency": "EUR",
"description": "Unrealized FX gain for October",
"status": "proposed – needs approval"
}
]
Next steps
- Review the generated
report.htmlfor a detailed line-by-line audit. - Approve the proposed journals in
proposed-journals.csvfor posting to your ERP. - Copy the draft email from
agent-email.txtto request the missing invoice HB-99 from the agent.
report.html
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Manual reconciliation of overseas agent and intercompany statements is a slow, error-prone process that stalls month-end closing. This skill automates the heavy lifting of matching ledger exports against third-party statements, identifying every discrepancy with deterministic precision. It eliminates the guesswork of foreign exchange revaluations and missing documents by proving the balance difference down to zero.
What it does
- Deterministic matching uses hard-coded rules to pair ledger entries with statement lines, recording the specific rule used for every match.
- Exception analysis categorizes every unmatched item by type, effect, and required action, such as missing invoices or timing differences.
- Balance proofing calculates the delta between ledger and statement closing balances, ensuring every cent is accounted for before finalizing.
- Journal proposal generates correcting entries for bank charges, accruals, and FX revaluations, marked for human approval.
- Communication drafting creates a ready-to-send email for counterparties, highlighting only the items they need to address.
How it works
- Input validation checks ledger CSVs and statement text files for date format consistency and sign conventions.
- Reconciliation execution runs the engine to match lines, calculate FX effects, and generate a structured audit trail.
- Difference verification ensures the "unexplained" balance is exactly zero by verifying opening balances and period boundaries.
- Result reporting delivers a full suite of outputs including a visual HTML report, exceptions CSV, and proposed journals.
Frameworks & tools
This skill utilizes a Node.js 20+ engine, Anthropic API for optional OCR/synthesis, and an MCP server for tool-based execution. It handles CSV, TXT (via pdftotext), and JSON formats.
Why this beats prompting it yourself
Standard LLM prompting struggles with large-scale arithmetic and multi-line matching, often leading to hallucinated balances. This skill uses deterministic code for the math and matching, only using AI for unstructured text parsing and natural language drafting, ensuring 100% numerical accuracy.
Use cases
- Reconciling foreign agent statements for freight forwarders or logistics firms.
- Performing month-end intercompany balance matching across different ERP systems.
- Resolving supplier statement discrepancies where bank charges or FX fees were deducted.
- Audit preparation by maintaining a
jsonlaudit log of every matching decision.
Known limitations
Requires Node.js 20 or later for the local engine. Statements in PDF format must be converted to layout-preserved text first. Ledger exports must be provided in CSV format.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Finance & Accounting.
Requires Node.js 20+. PDFs require layout-preserved text extraction. Does not support direct ledger posting or automated email sending. Balance proof must reach zero to be considered final.
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
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed today
- Free with an Agensi account
- Software
Git
Node.js 20+Ledger export as CSV
- Optional
Anthropic
First run clones the open-source engine from GitHub and runs npm install. Optional: set ANTHROPIC_API_KEY to read printed/PDF statements; without it a built-in parser is used and no data leaves your machine. Nothing is posted to a ledger or emailed.
Permissions
File Scopes
Terminal: runs node/npm for the reconciliation and the input checker. Read/Write files: reads your ledger and statement, writes reports to the output folder. Network: only git clone and npm install on first setup, and the Anthropic API if you set ANTHROPIC_API_KEY. Env vars: reads ANTHROPIC_API_KEY (optional). Nothing is posted to a ledger or emailed.