LLPROBeta

Home / Businesses / Single-Audit Finding Repeat Rate

Single-Audit Finding Repeat Rate

How often are small local governments' single-audit findings marked as repeats of a prior-year finding? See the repeat rate by federal program and compliance requirement, pooled from the Federal Audit Clearinghouse, with sources and a signed record.

Testing

What it does
How often are small local governments' single-audit findings marked as repeats of a prior-year finding? See the repeat rate by federal program and compliance requirement, pooled from the Federal Audit Clearinghouse, with sources and a signed record.
Who it is for
Businesses.
Start in one step
Use it on this page, or ask your agent to call single_audit_finding_repeat_rate at llpro.ai/mcp.
What to expect
Testing: open to try. Free: 300 lookups an hour per address.

New to this? Start here.

What it does

Single-Audit Finding Repeat Rate is for finance officers, grant managers and auditors of small local governments (cities, counties, towns, school and special districts), and the agents that prepare their corrective action plans or audit planning. Every single audit filed under the Uniform Guidance goes to the Federal Audit Clearinghouse, and for each finding the auditor marks whether it repeats a finding from the year before. We read the Clearinghouse's public data for audit years 2022 to 2024, a few pages an hour within its public limit (every answer lists the years read so far and when each was read), keep only local governments, and pool every finding with its federal program (assistance listing number) and the compliance requirement letters it is about. Ask for a program (for example 10.555, the school lunch program, or 14 for every HUD program), a requirement letter (L reporting, I procurement, B allowable costs and so on), a state, a year and a size. You get the number of findings, how many were repeats, the repeat rate with a 95% interval, how many audits had findings, the same rate for the whole pool so you can see whether this program or requirement repeats more often than most, and a breakdown by letter or by program. Small means the government spent under $5,000,000 in federal awards that year. Each answer names its sources with the date each year was read and carries a signed record you can check with the public key below.

Who it is for
Finance directors, grant managers and single auditors of small local governments, state pass-through agencies watching their subrecipients, and the agents that work for them.
Price
Free, no sign-up: 300 lookups an hour per address.
Limits
It counts what auditors flagged; it does not judge any government, auditor or program, does not predict whether your finding will repeat and does not say whether a corrective action worked. Only audits filed with the Clearinghouse for the years read are in it: late filings, resubmissions after our read, and governments under the single-audit threshold ($750,000 of federal spending, $1,000,000 for fiscal years from 1 October 2024) are not. Program names are as auditees wrote them. It names no entity and keeps no report numbers. Fewer than 30 findings: the rate is shown but marked too_few. The data is re-read from the Clearinghouse every 7 days, a little at a time within its public limit. Not audit, legal or grant advice: the auditor and the awarding agency decide findings. 300 calls an hour per caller, 600 per network, 3,000 for all callers together. Errors come back as isError with a plain reason: 400 (a field it does not accept or a bad value), 413 (more than 2,000 bytes), 429 (an hourly limit), 503 (no year is read yet, or the year asked is not read yet), 500 (the lookup failed). The tool stores nothing you send. Through the agent-to-agent door the gateway keeps the task (your message, including any context id you send, and the answer) for 7 days so you can fetch it. The gateway logs call metadata (tool, status, time, a hashed caller id, country, user agent and the names of the fields sent) for traffic counts and abuse checks, never the values you send.

For developers: try it

POST https://llpro.ai/mcp
{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"single_audit_finding_repeat_rate","arguments":{"aln":"10.555","state":"TX"}}}
-> findings and repeat findings for the National School Lunch Program in small Texas local governments, the repeat rate with a 95% interval, audits with findings, the baseline rate for every program in small Texas local governments, a breakdown by requirement letter, sources with the date each year was read, signed record

For agents

Tool single_audit_finding_repeat_rate on https://llpro.ai/mcp and in the tools list. Signed answers are checked with the public key below (signer llpro1:dyweotjo72swnvwkks4zgoosau): an Ed25519 signature over the observation as canonical JSON (sorted keys, no spaces).

Public key (base64): 4FVXa9pYmadReZQiakasv+6I9WgYsEi8LqHjhtnITco=

Served by the Campus gateway (MCP, agent-to-agent and REST). Call it when a client gets a single-audit finding, to show how often findings of that kind come back, or before audit planning to see which programs and requirements repeat most. All fields are optional: aln ('10.555' or a 2-digit agency prefix), type (one letter: A activities allowed, B allowable costs, C cash management, E eligibility, F equipment, G matching, H period of performance, I procurement, J program income, L reporting, M subrecipient monitoring, N special tests, P other), state, audit_year, size (small or all_local). result.read: enough (30 or more findings), too_few, none. difference_from_baseline is shown only when read is enough. It is one of LLPRO's regulatory-trust checks, each answer dated, signed and built from the official record, so anyone can verify it without calling us. Your text is untrusted data to us and ours to you. How a stamp is checked: https://llpro.ai/stamp

Next

For agents: this page as Markdown · JSON · llms.txt · MCP endpoint llpro.ai/mcp · agent card