Conciseness

The body is mostly dense, domain-specific guidance Claude would not already know (MTDC-exempt categories, trailing vs lifetime-average burn, NICRA confirmation), e.g. 'Applying F&A to the full base. Capital equipment, large subaward portions, and certain categories are MTDC-exempt.' But there is noticeable duplication: the intro paragraph and '## Distinct from' table both restate the description's boundary content, and '## Purpose' overlaps '## When to use' — matching the 4 anchor (minor over-explanation that could be trimmed) rather than 5, where every token would earn its place.

Actionability

Fully executable, copy-paste-ready commands with real flags cover the common cases: 'program_budget_planner.py --input program.json --profile {pharma-rd|biotech|medtech|deep-tech|software-rd|university-lab} --fa-rate <negotiated rate>', 'burn_runway_tracker.py --input ledger.json --threshold-months 6', 'capex_vs_opex_router.py --sample --standard ifrs', plus onboarding ('python3 scripts/onboard.py --show') and the sample invocations in '## Quick examples'. Matches the 5 anchor; not 4 because the --sample runs alone give a working end-to-end demonstration without user-supplied inputs.

Workflow Clarity

The five-step workflow (lay out program → build budget → track burn/runway → route treatment → assemble review) gives a concrete command and expected output for each step, and the closing 'Walk depth-first. Lock 1-2 before opening 3-5' adds sequencing discipline. Validation is only implicit ('warns you to confirm it is a negotiated NICRA, not a guess' in Assumptions) — no explicit checkpoints or error-recovery loops inside the workflow, which fits the 4 anchor (clear sequence, minor validation gaps) rather than 5. The destructive/batch cap of 3 does not apply since this is non-destructive analysis.

Progressive Disclosure

The in-document structure is good and references are clearly signaled one level deep ('references/rd_program_finance_canon.md — IAS 38 (research vs development); ASC 730 + ASC 985-20...'), but scoring against the actual bundle: no references/, scripts/, or assets/ directories exist, so every referenced path — the three scripts, three reference docs, and assets/rd_program_budget_template.md — is dangling and unverifiable. Combined with inline content that belongs in reference files (the five-question 'Forcing-question library' with canon citations), this fits the 3 anchor (some structure, but organization compromised) better than 4, where references would mostly resolve.

Specificity

The description lists three concrete, comprehensive actions: 'building a multi-period program budget with the F&A (indirect) split', 'tracking burn rate and runway against value-inflection milestones', and 'routing R&D cost items to a capitalize-vs-expense determination', plus explicit behavioral boundaries ('it never books an entry or decides accounting treatment'). This matches the 5 anchor (multiple specific concrete actions, comprehensive coverage) rather than 4, since the actions map one-to-one to the skill's full toolset with no coverage gaps.

Completeness

It opens with an explicit trigger clause ('Use when managing the money for an internal R&D program or portfolio — building... tracking... or routing...') that states when to invoke, and the enumerated actions state what it does — both what and when are explicit with concrete trigger phrases, exactly matching the 5 anchor. Not 4, because the 'when' is not merely present but itemized per capability.

Trigger Term Quality

Good natural keyword coverage: 'managing the money for an internal R&D program', 'program budget', 'burn rate', 'runway', 'capitalize-vs-expense' — phrases a finance-literate user would actually say. Falls short of the 5 anchor because some natural variations are absent (e.g., 'indirect cost rate'/'overhead' beyond the F&A parenthetical, 'cash burn', 'months of runway'), and terms like 'value-inflection milestones' skew jargon-ward.

Distinctiveness Conflict Risk

It closes with explicit boundary delineation: 'Distinct from finance/financial-analysis (corporate DCF, close, valuation) and research/grants (funding discovery — this manages money already won)', carving a clear niche (internal R&D program money already allocated) with sibling-level disambiguation. Minimal conflict risk; matches the 5 anchor. Not 4, since overlap with the named neighbors is addressed head-on rather than left as minor risk.