Skip to content

cost-reasonableness-check

Slugcost-reasonableness-check
Version0.1.0
Statusexperimental
Last fully evaluatednone
Eval stateno validated eval cases
Categoryreview
Domainresearch-administration
Manifestationsprompt
Created2026-05-21
Updated2026-05-21

Tags: cost-allowability post-award reasonableness review federal-grants research-administration

Audience: post-award-staff, sponsored-programs-staff, grant-accountants

Manifestations in repo: prompt.md

Checks whether a single expense is necessary and reasonable under the prudent-person standard of 2 CFR 200.404. It is one of the single-requirement checks in the federal cost-allowability analysis workflow and emits a structured finding consumed by the final determination step.

Output contract: schema.json Contract scope: repo-local

Inputs

A normalized expense record, extracted award terms, and a regulated-activity classification from the Stage 1 components of the cost-allowability-analysis workflow. Any may be partial.

Outputs

A single structured finding — see schema.json — carrying a status of pass, issue, not_allowable, needs_info, or not_applicable, with rationale, cited evidence, follow-up actions, and a confidence level.

The check

Evaluates whether the cost is ordinary and necessary for the project and whether its amount is consistent with prudent, market-comparable spending. It judges necessity and amount only — allocability and budget conformance are separate checks. When the item, the amount basis, or the project necessity is unclear, the check returns needs_info rather than assuming reasonableness.

Contract scope

Repo-local. The finding object is the shared single-requirement finding contract used across the cost-allowability check components; check_id, check_name, and regulation_anchor are fixed for this check. It is not a shared AI4RA-UDM schema.

Triad integration

  • Evaluation datasets: none yet — repo-local synthetic coverage planned.
  • Harness notes: canonical manifestation is prompt.md; validation surface is schema.json. Invoked as a Step 2 task of the cost-allowability-analysis workflow.
  • Shared UDM relationship: aligned to sponsored-project cost-allowability semantics; does not define or depend on a shared UDM schema.

Manifestations

  • prompt.md — canonical, LLM-agnostic prompt

Evals

See evals/.

Provenance

Created 2026-05-21 as a single-requirement check in the federal cost-allowability analysis component set, anchored to 2 CFR 200.404.

Contract scope

  • Output format: json_object

  • Contract scope: repo_local_single_requirement_finding_contract

  • Validation surfaces: json_schema

  • Schema entrypoints: #

  • Notes: Repo-local single-requirement allowability finding for the cost reasonableness check (2 CFR 200.404). Emits the shared cost-allowability finding object.

  • Machine-readable catalog entry: component_catalog.json

Triad integration

  • UDM alignment: repo_local_single_requirement_finding_contract — Aligns to sponsored-project cost-allowability semantics; the finding object is a prompt-library contract shared across the cost-allowability check components, not a shared UDM schema.

  • Evaluation datasets: no shared evaluation-data-sets catalog entry recorded yet; current references are repo-local eval artifacts.

  • Harness notes: Validate JSON output against schema.json. Canonical invocation surface is prompt.md. Invoked as a Step 2 task of the workflows/cost-allowability-analysis Vandalizer workflow.

Prompt body

Source: prompt.md.

Show prompt

Cost Reasonableness Check

Purpose: Decide whether a single expense is necessary and reasonable under the prudent-person standard.

Expected input: A normalized expense record, extracted award terms, and a regulated-activity classification.

Expected output: One structured finding conforming to schema.json.

This component is a single-requirement check in the federal cost-allowability analysis workflow. It evaluates necessity and amount only.

Prompt

You are a federal cost-allowability reviewer performing one specific check: Cost Reasonableness under 2 CFR 200.404. Review one expense and return a single structured finding for this check only — do not evaluate any other allowability requirement.

Return only a single JSON object conforming to the finding contract. No prose, Markdown, comments, or code fences.

Input

You receive a normalized expense record, extracted award terms, and a regulated-activity classification, produced by the upstream steps of the cost-allowability analysis workflow. Any of these may be partial or missing.

What this check evaluates

Whether the cost is necessary and reasonable under 2 CFR 200.404 — the prudent-person standard. A cost is reasonable if, in its nature and amount, it does not exceed what a prudent person would incur under the circumstances prevailing when the decision to incur the cost was made. Consider:

  • whether the cost is of a type generally recognized as ordinary and necessary for the operation of the entity and the performance of the award;

  • sound business practices, arm's-length bargaining, and applicable federal, state, and other law;

  • market prices for comparable goods or services in the geographic area;

  • whether the individuals concerned acted with the prudence expected under the circumstances; and

  • whether the entity significantly deviated from its established practices and policies in a way that unjustifiably increased the cost.

This check evaluates necessity and amount only. It does not decide whether the cost benefits the award (allocability) or fits the approved budget (award-terms conformance).

Decision rule

Set status to exactly one of:

  • pass — the cost is ordinary and necessary for the project, and the amount is consistent with prudent, market-comparable spending on the evidence supplied.

  • issue — the amount appears high relative to the goods or services, or the necessity for the project is not clearly established, and a justification should be obtained.

  • not_allowable — the evidence shows the cost is plainly excessive or is not necessary for the performance of the award.

  • needs_info — there is not enough detail about the item, the basis for the amount, or the project necessity to judge reasonableness.

  • not_applicable — use only when the input is explicitly not a chargeable cost.

Use the most conservative status the evidence supports. Never record pass when the evidence needed to satisfy the requirement is absent — that is needs_info.

Evidence and non-fabrication

  • Ground every finding in the supplied evidence. Populate evidence with short source_label / detail pairs (e.g., expense:amount, expense:description, award_terms:approved_budget_categories).

  • Do not invent market prices, comparables, or necessity rationales. A missing rule input is needs_info, never pass.

Output

Emit the finding object:

  • check_id — "cost-reasonableness"

  • check_name — "Cost Reasonableness"

  • regulation_anchor — "2 CFR 200.404"

  • expense_id — the expense reference from the input, or null

  • status — one of the values above

  • summary — one sentence stating the outcome

  • rationale — the evidence-grounded reasoning

  • evidence — array of {source_label, detail}

  • follow_up_actions — concrete actions needed before the cost is approved or charged; empty array when none

  • confidence — "high", "medium", or "low", reflecting evidence completeness

Produce the finding now.

Output schema

Source: schema.json.

Show schema.json
{

  "$schema": "https://json-schema.org/draft/2020-12/schema",

  "$id": "https://github.com/AI4RA/prompt-library/components/cost-reasonableness-check/schema.json",

  "title": "Cost Reasonableness Check \u2014 Finding",

  "description": "Structured single-requirement allowability finding for the cost reasonableness check (2 CFR 200.404). Part of the federal cost-allowability analysis workflow.",

  "version": "0.1.0",

  "type": "object",

  "additionalProperties": false,

  "required": [

    "check_id",

    "check_name",

    "regulation_anchor",

    "status",

    "summary",

    "rationale",

    "evidence",

    "follow_up_actions",

    "confidence"

  ],

  "properties": {

    "check_id": {

      "const": "cost-reasonableness",

      "description": "Stable identifier for this check."

    },

    "check_name": {

      "const": "Cost Reasonableness",

      "description": "Human-readable check name."

    },

    "regulation_anchor": {

      "const": "2 CFR 200.404",

      "description": "Primary regulatory anchor for this check."

    },

    "expense_id": {

      "type": [

        "string",

        "null"

      ],

      "description": "Expense reference carried from the input."

    },

    "status": {

      "type": "string",

      "enum": [

        "pass",

        "issue",

        "not_allowable",

        "needs_info",

        "not_applicable"

      ],

      "description": "Outcome of this check."

    },

    "summary": {

      "type": "string",

      "minLength": 1,

      "description": "One-sentence statement of the outcome."

    },

    "rationale": {

      "type": "string",

      "minLength": 1,

      "description": "Evidence-grounded reasoning behind the status."

    },

    "evidence": {

      "type": "array",

      "description": "Evidence relied on. May be empty only when status is not_applicable.",

      "items": {

        "type": "object",

        "additionalProperties": false,

        "required": [

          "source_label",

          "detail"

        ],

        "properties": {

          "source_label": {

            "type": "string",

            "minLength": 1,

            "description": "Short source label, e.g. expense:amount."

          },

          "detail": {

            "type": "string",

            "minLength": 1,

            "description": "Concise summary of what the source shows."

          }

        }

      }

    },

    "follow_up_actions": {

      "type": "array",

      "description": "Concrete actions needed before the cost is approved or charged. Empty array when none.",

      "items": {

        "type": "string",

        "minLength": 1

      }

    },

    "confidence": {

      "type": "string",

      "enum": [

        "high",

        "medium",

        "low"

      ],

      "description": "Reviewer confidence in this finding."

    }

  }

}

Changelog

Source: CHANGELOG.md.

All notable changes to this component. Versions follow semver: MAJOR for output-contract breaks, MINOR for backward-compatible additions, PATCH for wording or clarity.

[0.1.0] — 2026-05-21

  • Initial experimental release.
  • Single-requirement allowability check anchored to 2 CFR 200.404.
  • Emits the shared cost-allowability finding object (status, summary, rationale, evidence, follow_up_actions, confidence) with check_id "cost-reasonableness".
  • No eval cases yet — status experimental until at least one golden finding is added under evals/cases/.