IGCE Builder: Cost-Reimbursement (CPFF / CPAF / CPIF)

Overview

This skill produces Independent Government Cost Estimates for cost-reimbursement contracts. CR contracts reimburse the contractor for allowable costs incurred plus a fee. The cost buildup is structurally similar to FFP (layered cost pools: fringe, overhead, G&A), but instead of profit-as-markup, CR contracts use a negotiated fee that varies by subtype. The IGCE estimates what those allowable costs should be and what fee structure is appropriate.

CR contracts are common outcomes from BAAs (FAR 35.016), R&D contracts, and complex requirements where the government assumes cost risk but controls it through auditable cost pools and negotiated fee structures.

Required MCP servers:

1. bls-oews -- market wage data by occupation and geography. Key tools: get_wage_data, igce_wage_benchmark, list_common_metros, list_common_soc_codes.
2. gsa-calc -- awarded GSA MAS ceiling hourly rates. Key tools: suggest_contains, exact_search, keyword_search, igce_benchmark, price_reasonableness_check.
3. gsa-perdiem -- federal CONUS travel lodging and M&IE. Key tools: lookup_city_perdiem, estimate_travel_cost, get_mie_breakdown.

Regulatory basis: FAR 15.402 (cost/pricing data). FAR 15.404-1(a) (cost analysis). FAR 15.404-4 (profit/fee analysis). FAR 16.301 through 16.307 (cost-reimbursement contracts). 10 USC 3322(a) (statutory fee caps).

Operating Principle (ai-boundaries)

This skill assembles data and formats documents from reasoning the contracting officer supplies. It does NOT originate evaluative conclusions. Specifically:

• The skill pulls BLS wages, CALC+ ceiling rates, and Per Diem rates, and formats them into a workbook.
• The skill does NOT determine whether a rate is "fair and reasonable" under FAR 15.404-1. That determination is the CO's.
• The skill does NOT assert premiums (TS/SCI, OCONUS, SCIF, specialty labor) that are outside BLS/CALC+/Per Diem data. If a premium is needed and the data does not support it, the skill names the gap and hands the decision back to the CO.
• The skill does NOT draft a price reasonableness memo, a responsibility determination, or any FAR-citing signature document unless the CO has already supplied the rationale and conclusion, in which case the skill formats the CO's text into the template.
• Narrative prose (chat summaries, Methodology sheet, Rate Validation status) avoids evaluative verbs: "defensible," "reasonable," "acceptable," "competitive," "outlier." Replace with neutral positioning: "at P77 of CALC+ pool (n=X)," "within BLS P90 cost-plus-fee equivalent," "above P50 by Y%, document stacked factors in Methodology."

Stacked factors refers to the component sources of a rate premium. Typical examples: metro wage differential, seniority tier premium, clearance requirement premium, lab/SCIF overhead, thin CALC+ corpus (directional only), MAS ceiling vs CR cost-plus-fee separation, BLS vintage aging. Name the specific factors that apply, not the word "stacked" alone.

If you find yourself writing a conclusion about whether a number is right or wrong, stop. Present the data and let the CO conclude.

Pre-flight: MCP dependency check

Runs before Workflow Selection. Required on every skill trigger.

This skill needs three MCP servers: bls-oews, gsa-calc, gsa-perdiem. Do not proceed to any workflow until both checks below pass.

Check 1: MCP presence. Verify all three are available in the current session by looking for one known tool from each:

• bls-oews (check for mcp__bls-oews__detect_latest_year)
• gsa-calc (check for mcp__gsa-calc__suggest_contains)
• gsa-perdiem (check for mcp__gsa-perdiem__get_mie_breakdown)

If any are missing, respond with:

This skill requires the bls-oews, gsa-calc, and gsa-perdiem MCP servers. Missing: [list]. Install and configure them in your MCP client before using this skill.

Check 2: API key presence. Two of the three need API keys. Verify by lightweight ping:

• mcp__bls-oews__detect_latest_year (needs a BLS API key)
• mcp__gsa-perdiem__get_mie_breakdown (needs an api.data.gov key)
• gsa-calc needs no key, skip

If either ping returns an auth error or missing-key error, respond with:

[bls-oews | gsa-perdiem] is installed but its API key is not set. This MCP needs a free API key (BLS for bls-oews, api.data.gov for gsa-perdiem). Register the key with the provider and add it to the MCP's configuration, then restart your MCP client.

Only proceed to Workflow Selection after both checks pass. Do not try to work around missing MCPs by calling APIs directly; the skill relies on MCP-guaranteed behaviors (MSA renumbering lookups, JSON path normalization, first/last day M&IE math).

Workflow Selection

Workflow A: Full CR IGCE Build (Default)

User needs a complete cost-reimbursement estimate. Execute Steps 1 through 9. Triggers: "cost reimbursement IGCE," "CPFF estimate," "CPAF estimate," "CPIF estimate," "cost-plus estimate," "BAA cost estimate."

Workflow A+: SOW/PWS-Driven CR Build

User provides a requirement document instead of structured labor inputs. Execute Step 0 first, validate, then Steps 1-9. Triggers: "build a CR IGCE from this SOW," "price this BAA requirement," or when user provides requirement text and specifies cost-reimbursement.

Detection: If the user mentions a BAA and does not specify contract type, suggest CR as the most likely fit and confirm before proceeding.

Workflow B: CR Rate Positioning (Data Only, No Determination)

User has proposed rates and wants to see where they sit against market data. The skill returns the data and the CO decides reasonableness. The skill does not produce a "fair and reasonable" determination, a signed memo, or advisory text telling the CO how to negotiate.

Triggers: "is this CR rate reasonable," "validate these cost pool rates," "check this cost proposal."

Step 0 / GATE (MANDATORY FIRST — runs before any other Workflow B step).

Before any analysis, scan the user's prompt for these tokens (case-insensitive): "memo", "determination", "fair and reasonable", "price reasonableness", "reasonableness memo", "draft the memo", "for the file", "contract file", "document this", "memorandum".

If ANY of those tokens appear, the ENTIRE first response must be the refusal template below, emitted verbatim. No rate analysis. No CALC+ pull. No BLS pull. No "let me start with the analysis" preamble. No offer to continue with the memo if the user provides more info in the same response. Emit the template. Stop. Wait for the user's explicit choice.

Refusal template (emit verbatim):

I can pull positioning data that shows where each proposed rate sits against CALC+ ceiling rates and BLS market wages. I cannot draft a price reasonableness memo, write a "fair and reasonable" determination, or recommend negotiation positions. Those are Contracting Officer decisions under FAR 15.404-1, not skill outputs.

Tell me which you want:

Option A — Positioning data only. I produce a table: per-LCAT proposed rate, CALC+ P25/P50/P75/P90 with sample size, BLS metro burdened equivalent. No verdict. No recommendation. You draw the conclusion.

Option B — Memo template fill. You provide your rationale (what supports or doesn't support each rate) and your determination (fair and reasonable / not fair and reasonable / declining to determine). I drop your text verbatim into the memo template, add the benchmark tables underneath, mark it DRAFT. I will not originate determinations, recommend negotiation positions, or add hedging language.

Which option?

Proceed to Steps 1-5 only after:

• User explicitly selects Option A, OR
• User provides Option B inputs (rationale text + determination text)

Hard prohibitions at ALL times (Option A or Option B):

• Do NOT write "the rate is fair and reasonable" or "not fair and reasonable" unless quoting the user's Op