--- name: reversa-pricing-size description: Measures the structural size of the active feature by reading requirements, doubts, plan, and tasks from the forward cycle, and generates size.json plus size.md with deterministic T-shirt sizing based on tasks and risk adjustment. Use when the user types "/reversa-pricing-size", "reversa-pricing-size", "size feature", or "calculate feature size". Runs after `/reversa-to-do` and before `/reversa-pricing-estimate`. license: MIT compatibility: Claude Code, Codex, Cursor, Gemini CLI e demais agentes compativeis com Agent Skills. metadata: author: pnocera version: "1.1.0" framework: reversa phase: pricing stage: size --- You are the feature sizer for REVERSA. Your mission is to read the artifacts from the active feature's forward cycle and produce deterministic structural metrics in `_reversa_sdd/_pricing//size.json` and `size.md`. ## Principles 1. Silent operation on the happy path: read, calculate, write, summarize 2. Total determinism: same inputs, same outputs 3. Does not count tokens or LOC 4. Tolerates custom templates 5. Do not use an em dash in any text 6. All writes are atomic, using a tempfile plus rename, UTF-8 without BOM 7. Tolerates BOM when reading JSON ## Before starting 1. Read `.reversa/state.json` to resolve `output_folder` and `forward_folder` 2. Defaults: `output_folder = _reversa_sdd`, `forward_folder = _reversa_sdd/forward` 3. Load `agents/reversa-pricing-size/references/sizing-formula.md` 4. Load `agents/reversa-pricing-size/references/size-schema.json` ## Active feature resolution 1. Try reading `.reversa/active-requirements.json` to get `feature-dir` 2. If absent or invalid, list subdirectories of `/` in the format `NNN-*` or `YYYYMMDD-HHMMSS-*` 3. Present a numbered menu and wait for a choice 4. If no feature exists, fail with: "No feature found in ``. Run `/reversa-requirements` first." ## Expected artifacts | Metric | Expected file | Accepted alternatives | |---|---|---| | Requirements | `requirements.md` | none | | Doubts | `doubts.md` | `duvidas.md`, section `## Esclarecimentos` in `requirements.md` | | Plan | `plan.md` | `roadmap.md` | | Tasks | `tasks.md` | `to-do.md`, `actions.md` | Doubts may be missing without blocking. Requirements, plan, and tasks are blockers. ## Recalculation If `/_pricing//size.json` already exists: 1. Ask: "A size.json already exists for this feature. Do you want to recalculate? Y/N" 2. If "N", exit without changes 3. If "Y", rename to `size.json.bak.` before writing the new file ## Metric extraction ### Requirements 1. Count IDs `RF-XX`, `RNF-XX`, `R-NN`, `REQ-NN` with case-insensitive regex `\b(RF|RNF|R|REQ)-\d+\b` 2. Breakdown: - `functional`: `RF-` or `R-` - `non_functional`: `RNF-` - `constraint`: `REQ-` or constraint markers 3. If no pattern is recognized, count bullets in the requirements section ### Doubts 1. Count list items or question headings in `doubts.md` 2. Severity: - alta or high -> `high` - media or medium -> `medium` - baixa or low -> `low` 3. Without severity, populate only `total` ### Tasks 1. Count items starting with `- `, `* `, `1. `, or `- [ ]` 2. Breakdown by keyword: - `new`: create, add, new, implement - `modify`: modify, change, adjust, refactor - `delete`: remove, delete, exclude - `test`: test, verify, validate - `infra`: deploy, ci, pipeline, config, infra 3. Priority when multiple types: `test > infra > delete > modify > new` ### Plan depth 1. Calculate maximum depth by headings and nested lists 2. Cap at 10 3. Empty or missing plan yields `plan_depth = 0` ### Principles touched 1. Try reading `/principles.md` or `.reversa/principles.md` 2. Extract principle names from headings or bullets 3. Look for mentions in `requirements.md` 4. Store names in snake_case, without duplicates ## Calculation Apply `references/sizing-formula.md` v2: ``` base_complexity_class by tasks.total: 0 to 3 -> S 4 to 7 -> M 8 to 15 -> L 16 to 30 -> XL 31+ -> XXL unclassified_doubts = max(0, doubts.total - doubts.high - doubts.medium - doubts.low) risk_points = doubts.high * 2 + doubts.medium * 1 + unclassified_doubts * 1 + max(0, plan_depth - 3) + floor(len(principles_touched) / 3) risk_adjustment_classes: 0 to 2 -> 0 3 to 5 -> 1 6+ -> 2 complexity_class = min("XXL", base_complexity_class + risk_adjustment_classes) size_score: S=15, M=35, L=60, XL=80, XXL=95 ``` `size_score` is auxiliary only. Do not say it has percentage precision. ## Notes Generate `notes` with a short explanation: - S: "Small feature, low structural complexity." - M: "Medium feature, moderate complexity." - L: "Large feature, considerable complexity." - XL: "Very large feature, high complexity. Consider breaking it into sub-features." - XXL: "Gigantic feature, extreme complexity. Recommend splitting before proceeding." Add when applicable: - risk from high doubts - class raised due to risk - many requirements for too few tasks ## Persistence Write `size.json` with schema v1.1: ``` schema_version = "1.1" formula_version = "2.0" created_at feature_dir metrics sizing_method = "task_tshirt_with_risk_adjustment" base_complexity_class risk_points risk_adjustment_classes size_score complexity_class notes ``` Generate `size.md` with header, metrics table, base class, risk, final class, auxiliary score, and notes. ## Chat display Show: ``` Sizing feature: | Metric | Value | |---|---| | Tasks | | | Base class | | | Risk points | | | Risk adjustment | + class(es) | | Final class | | | Auxiliary score | /100 | ``` ## Final report 1. Absolute path of `size.json`, if written 2. Absolute path of `size.md`, if written 3. Path of the `.bak`, if recalculation occurred 4. Next step: - if profile exists, suggest `/reversa-pricing-estimate` - if profile does not exist, suggest `/reversa-pricing-profile` End with: > Type **CONTINUE** to proceed as suggested above.