research-finance

Build R&D program budgets, track burn and runway, and route capitalize-vs-expense decisions.

25.3k|3.6k|Updated Oct 19, 2025
One-click install
npx skills add https://github.com/alirezarezvani/claude-skills --skill research-finance
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: research-finance
Source: https://github.com/alirezarezvani/claude-skills/tree/main/research-ops/skills/research-finance
Command: npx skills add https://github.com/alirezarezvani/claude-skills --skill research-finance

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

Managing money for internal R&D programs requires defensible budgets with explicit F&A (indirect) splits, honest burn/runway reads against value-inflection milestones, and a first-pass routing of capitalize-vs-expense questions — all without crossing into corporate finance or making accounting determinations that belong to a controller.

Core Features & Use Cases

  • Program Budget Planning: Build multi-period budgets from work-package line items with an MTDC-style F&A base (excluding capital equipment and large subawards) and an explicit assumptions block on every number.
  • Burn & Runway Tracking: Compute trailing burn, runway in months, and whether each milestone is reachable before cash runs out, with flags for accelerating burn and below-threshold runway.
  • Capitalize-vs-Expense Routing: Score cost items against IAS 38 development-phase criteria or ASC 730 and route them to a named finance owner as CAPITALIZE-CANDIDATE, EXPENSE, or FINANCE-OWNER-REVIEW — never auto-deciding treatment.
  • Use Case: A biotech program controller prepares a portfolio review: run the budget planner with the negotiated NICRA rate, check runway against the first-in-human milestone, and route a pilot-line tooling cost to the finance owner for capitalization review.

Quick Start

Ask the agent to build a multi-period R&D program budget with F&A split and runway analysis using the research-finance skill's sample data.

Frequently Asked Questions about research-finance

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build an R&D program budget with F&A indirect costs?

Run program_budget_planner.py with a JSON file of work-package line items and your F&A rate. It applies the rate to an MTDC-style eligible base, excluding capital equipment and subaward portions over $25k, and outputs direct, F&A, and fully-loaded rollups per period with an assumptions block.

How do I calculate runway and burn rate for an R&D program?

Run burn_runway_tracker.py with a ledger JSON containing cash on hand, actual spend per period, and milestones. It uses trailing burn as the forward run-rate, computes runway in months, and flags milestones at risk and burn acceleration above 115% of average.

When can R&D costs be capitalized under IFRS vs US GAAP?

Under IAS 38, development costs must be capitalized once all six criteria are met, including technical feasibility; research costs are always expensed. Under US GAAP ASC 730, R&D is expensed as incurred, with software exceptions under ASC 985-20. The router scores items but a named finance owner makes the final call.

Does this skill decide accounting treatment automatically?

No. The capex_vs_opex_router is decision-support only: it routes items to CAPITALIZE-CANDIDATE, EXPENSE, or FINANCE-OWNER-REVIEW with a named owner. It never books an entry, and the controller plus auditor make the actual determination.

What is the difference between research-finance and financial-analysis skills?

research-finance manages money at the R&D program level: budgets, burn, runway, and capitalization routing. The finance/financial-analysis skill covers company-level work like DCF valuation, close, and SaaS metrics, while research/grants finds new funding rather than managing money already won.

What Python dependencies does the research-finance skill require?

None beyond the standard library. All scripts are stdlib-only, support --help, --sample, and --output human or json, and read optional onboarding config from ~/.config/research-ops/research-finance.json.