Handbook

Estimation model bootstrap — prompt

Purpose: One-shot (or periodic) prompt to design, calibrate, and apply the team’s Forge estimation method in a markdown-canonical repo. Use after copying templates from blueprints/sdlc/templates/forge/estimation/ and…

Related: Estimation rules — Forge SDLC (markdown-canonical) · Estimates ledger — work items · ../../methodologies/forge/versona/catalog/discipline/governance/versona-estimation.mdc.template · Markdown-canonical workspace policy (optional repo profile)

Prompt body (copy from here)

Act as Estimation Versona designer and estimator for this Forge SDLC workspace.

Design and apply an estimation method for meta-requests, Forge Sparks, and defects using:

  • Fibonacci sizing
  • t-shirt sizes
  • estimated token consumption (bands)
  • Calibration against historical repo data in Markdown ledgers

Workspace: Markdown-only canonical artifacts (no CSV SoT).

Tasks

  1. Propose a mapping between: estimated work complexity · Fibonacci bucket · t-shirt size · estimated token range (use or tune Estimation rules — Forge SDLC (markdown-canonical) seed).

  2. Define estimation fields for YAML front matter and/or docs/requirements/ESTIMATES.md (e.g. est_fib, est_tshirt, est_tokens, actual_tokens, uncertainty, coordination_factor, risk_multiplier).

  3. Define how estimates roll up from Forge Spark → Ingot → Product Spark / milestone.

  4. Define defect handling: ambiguity multiplier, blast-radius multiplier, RCA uncertainty (combined risk_multiplier or explicit factors).

  5. Analyze historical items already in this repo (ESTIMATES.md, closed tasks in docs/requirements/**, WBS.md, IMPORT-LEDGER.md if present) and infer initial calibration heuristics; if no history, state defaults and first iteration collection plan.

  6. Create or update Markdown:

  7. forge/estimation/ESTIMATION-RULES.md (copy from blueprints/sdlc/templates/forge/estimation/ESTIMATION-RULES.template.md)
  8. docs/requirements/ESTIMATES.md (copy from blueprints/sdlc/templates/requirements/ESTIMATES.template.md)

  9. docs/requirements/WBS.md (estimation summary column or pointer to ESTIMATES.md)

  10. Recommend front matter on per-item files (est_fib, est_tshirt, est_tokens, actual_tokens, uncertainty, coordination_factor, risk_multiplier).

  11. If versona-estimation is not installed, copy versona-estimation.mdc.template to .cursor/rules/ per Cursor rules — quickstart (Forge + Versonas).

  12. Search for overlapping estimation guidance (STRUCTURE-PROPOSAL, WBS.md, PROJECT.md, PM docs). Choose one canonicalization action A–G from the markdown-canonical policy before replacing a standard.

Execution mode: Apply edits directly when confident; else output exact Markdown for each file.

Return: A. Estimation model · B. Token/Fibonacci/t-shirt correlation · C. Markdown files to add/update · D. Calibration approach · E. Draft Estimation Versona summary (or “use shipped versona-estimation”) · F. Canonicalization decision for estimation standards

Maintainer notes

  • Canonicalization: versona-pm owns commitments; versona-estimation owns method — prefer D (distinct, cross-link) vs merging into PM.
  • Register versona-estimation in versona-standards-matrix.yaml after install for standards profile parity.