forgeplan fgr

forgeplan fgr prints the F-G-R quality triple for an artifact:

F — Formality: how structured is it? Filled MUST sections, linked parents, explicit acceptance criteria.
G — Granularity: how decomposed is it? FR count, phase checkboxes, leaf-level detail.
R — Reliability: how trusted is it? Linked evidence, R_eff, adversarial review status.

Each axis is scored 0–3. F-G-R is orthogonal to lifecycle — a draft artifact can already be F=3/G=3/R=2, and an active one can be F=1/G=1/R=0 (a blind spot). The FPF Trust Calculus uses exploit_fgr ≥ 0.6 as the threshold to reuse a decision without re-deriving it.

When to use

Quick visual check of maturity without running the full 30+ validator rules.
Session start, after health, to sort artifacts by “cheapest to level up”.
Comparing two candidate PRDs before committing scope.
Feeding the Trust Calculus: exploit_fgr gating explore-vs-exploit decisions in FPF rule engine.

When NOT to use

As a substitute for validate — FGR is a summary, not a rule report.
On non-decision artifacts where the axes don’t apply cleanly (Notes, RefreshReports).

Usage

forgeplan fgr [OPTIONS] [ID]

Arguments

  [ID]  Artifact ID (scores all if omitted)

Options

      --json     Output as JSON for machine consumption
  -h, --help     Print help
  -V, --version  Print version

Examples

Inspect one PRD

forgeplan fgr PRD-001

Output:

PRD-001 — Auth System
  F = 3/3   Formality     (all MUST sections filled, parents linked)
  G = 2/3   Granularity   (7 FR, 2 phases — phases missing acceptance criteria)
  R = 1/3   Reliability   (1 evidence linked, R_eff = 0.40)
  overall = 2.0/3   exploit_fgr = 0.67

exploit_fgr ≥ 0.6 → FPF rule engine will trust this decision for reuse.

Rank all artifacts by maturity

forgeplan fgr --json | jq 'sort_by(.overall) | reverse | .[0:10]'

Top 10 most mature artifacts — useful when deciding what to promote or cite.

Find the cheapest wins

forgeplan fgr --json | jq '.[] | select(.F == 3 and .G == 3 and .R < 2)'

Artifacts that only need evidence to reach full maturity — add evidence, score, done.

Output interpretation

Axis	0	1	2	3
F — Formality	stub, MUST missing	some MUSTs filled	all MUSTs + aliases	all MUSTs + links + ACs
G — Granularity	no FR/phases	1–3 FR	4–7 FR + phases	8+ FR + phased + measurable
R — Reliability	no evidence, R_eff=0	1 evidence, CL ≤ 1	2+ evidence, R_eff ≥ 0.4	strong evidence, R_eff ≥ 0.8

Overall = average of F/G/R. exploit_fgr = overall / 3.0, used as gate in forgeplan fpf trust.

How it fits the workflow

health → fgr → spot low-R artifacts → add evidence → score → fgr (re-check)

F and G are paid for during Shape (new → validate). R is paid for during Evidence (new evidence → link → score). fgr is the lens that tells you which bucket needs work.