grant_finding v1.0.0

name: "grant_finding" description: "Find grants a small business or nonprofit actually qualifies for: screen public and foundation programs against the eligibility rules in their own documents, score fit before any writing starts, track deadlines and required documents, and keep a shortlist that only holds live programs, with ruled-out programs shown and the reason recorded. Use when someone asks to find grants, check if they qualify for a grant, build a grant shortlist or pipeline, or decide which grant to apply for first."

Grant Finding

Purpose

Most grant searching fails in the same way. Someone collects a long list of program names, starts writing the most exciting one, and learns three weeks in that their entity type was never eligible. This skill reverses the order. Screen for eligibility first, from the funder's own document. Score fit second, from evidence. Write only what survives both. The shortlist keeps ruled-out programs on it, with the reason, so the same dead ends do not get re-screened every cycle.

This skill covers finding, screening, scoring, and tracking. It does not write the application itself, and it does not promise that any program is open today. Every deadline, award amount, and eligibility rule must be read from the current Notice of Funding Opportunity (NOFO) or funder guidelines during the run that uses it.

Workflow

1. Build the applicant profile before searching for anything. Record entity type, tax status, location and service area, employees including affiliates, revenue or operating budget, years operating, population served, and registration state: Unique Entity Identifier (UEI) issued, SAM.gov registration active with its renewal date, Grants.gov registration done. Use the profile block at the top of assets/eligibility-screening-worksheet.md. A search run without this profile produces guesses, not screening.

1. Check registration readiness as a hard gate, early. Prime recipients of federal grants need a UEI and an active SAM.gov registration, renewed every 365 days, and registration plus entity validation can take days to weeks. If registration is missing or lapsed, the first deliverable is a registration task with a date, not a shortlist full of federal programs the applicant cannot yet apply to.

1. Search in this order, because each layer is a different kind of money. Federal first at Grants.gov Search: keyword, then filter by Eligibility set to the applicant's exact entity type, then by category. Then SBIR.gov if the applicant is a for-profit doing research and development. Then the state economic development agency and county or city programs. Then foundations, using Candid's prospect research method: subject, population served, and geography, including whether the funder has funded similar organizations nearby. The full source map is in references/source-map.md.

1. Apply the reality check before adding anything to the list. The US government does not offer free money for individuals, and federal grants are typically for states and organizations, per USA.gov. The SBA does not give general grants to start or expand a small business. For a for-profit, the real federal routes are narrow: research and development programs like SBIR and STTR, programs run through states such as the State Trade Expansion Program, and state or local programs. Say this plainly at the start when it applies, instead of building a hopeful list around programs that were never open to this applicant.

1. Screen every candidate with assets/eligibility-screening-worksheet.md, reading the eligibility section of the program's own NOFO or guidelines, never a summary. The standard NOFO eligibility block has three parts: who may apply, cost sharing or matching, and other requirements such as registration, one-proposal limits, and will-not-fund lists. Record a pass, fail, or unknown for each worksheet question, with the source line. Unknown is not a pass: name where it will be verified, which can be the full NOFO text, the funder's program contact, or a free SBDC or SCORE advisor.

1. Score only the programs that passed screening, using assets/fit-scoring-sheet.md. Seven lines, each scored 0 to 5 from evidence with a one line reason: mission 25, population 15, geography 15, evidence and track record 15, award and need fit 10, timeline 10, burden 10. Run the math with scripts/grant_score.py score shortlist.csv, which applies the hard gates first and then the weights. A program that failed a gate comes out ruled out no matter what fit numbers were typed in.

1. Sort the survivors into the four verdicts and act on them the same day. PURSUE (75 and up) goes on the deadline tracker with a named owner. PURSUE IF CAPACITY (60 to 74) goes after every PURSUE item and is cut first when the calendar fills. WATCH (45 to 59) stays only with a written strategic reason. Under 45 is ruled out for poor fit, with the score kept on record.

1. Track every PURSUE item in assets/deadline-tracker.csv: program, funder, source URL, stage, letter of intent deadline if the program has one, final deadline exactly as published, owner, documents needed, documents ready, next action, and the date it was last checked against the funder's page. Run scripts/grant_score.py check tracker.csv weekly to flag overdue items, deadlines inside 14 days, missing owners, and missing document lists.

1. Re-verify the shortlist on a fixed cadence, weekly while any deadline is inside 60 days. Programs close, deadlines move, and forecast rounds open. A shortlist entry whose status, deadline, or eligibility was not checked on its last checked date is stale, and stale entries get re-screened or removed, never carried forward on trust.

1. Close the loop at the end of each cycle. For every program applied to, record submitted, awarded, declined, or withdrawn. For every ruled-out program, record whether the ruling fact could change (for example gaining a fiscal sponsor, finishing SAM.gov registration, or a match source appearing) and the date to look again. That record is what makes the next cycle start from evidence instead of from a blank page.

Output Contract

A grant finding run delivers four things, in this order:

• The applicant profile used for screening, with registration status and renewal dates stated.
• The shortlist, as a CSV following assets/shortlist-template.csv plus a plain language table, holding every screened program: pursuits and ruled-out programs together, each ruled-out entry carrying its exact gate and reason, each scored entry carrying its fit score, verdict, and the scorer output.
• The deadline tracker for PURSUE items, following assets/deadline-tracker.csv, with an owner and a document list for every row and the checker output attached.
• A short notes section stating what was searched, which sources were read directly, what could not be verified, and the next re-check date.

A worked example of the full output, with a fictional applicant and three ruled-out programs shown with reasons, is in assets/worked-shortlist.md, and the CSV the scorer runs on is assets/worked-shortlist.csv.

Operating Rules

• Eligibility comes from the funder's own document or it does not count. If the NOFO or guidelines were not read, the program is unscreened, and the shortlist must say so.
• Never invent a deadline, an award amount, a match percent, or a success rate. If it is not in the program's own material, leave it out or label it unverified. Award ranges and phase amounts vary by agency and year, so read them live every time.
• Match wording matters. Required, encouraged, and not required are three different facts in a NOFO. Quote the verb the document uses, and never treat encouraged as required or required as optional.
• A gate failure beats a fit score, always. High mission fit does not soften an entity, geography, tax status, or registration rule.
• Do not create a program to chase a grant. Candid's guidance to nonprofits applies to businesses too: funders fund existing work that fits their priorities. If the project only exists because the grant exists, flag it before scoring.
• Foundations generally give to organizations with 501(c)(3) status. A for-profit or an informal group needs a program that explicitly allows its type, or a fiscal sponsor who holds the status, and the worksheet must name which one applies.
• SBIR and STTR have a specific shape: the applicant is always the small business, for-profit, US based, 500 or fewer employees including affiliates, at least 51 percent owned and controlled by US citizens or permanent residents. STTR also requires a formal partnership with a nonprofit research institution. A nonprofit or a university applying as prime fails the gate.
• Run the scam test on anything that arrives uninvited. Contact out of the blue claiming the applicant qualifies, money pitched for personal needs like bills or debts, a request for a Social Security number or bank account to check eligibility, or any up front fee, especially by gift card, wire, or crypto, means it is a scam per the FTC, and it goes on the shortlist as ruled out so the record exists.
• Never pay for a list of grants and never pay an up front fee to apply. The federal list is free at Grants.gov, and free help exists through SBDCs and SCORE. Paid grant writers are a separate decision about writing help, never a toll for access to the list.
• Keep personal identifiers out of the worksheets and tracker. Record that an EIN, UEI, or registration exists and its status or date, never the numbers themselves in shared files.
• When two programs compete for the same hours, the higher verdict wins, and inside a verdict the sooner deadline wins. State the trade in the notes instead of letting both drift.