Skip to content

/gz-justify

Pre-execution reasoning walkthrough for GHIs, OBPIs, and drafts. Invoke before implementation when confidence is <90%, when a gz-adr-evaluate score lands below 3.0, or when scope boundaries feel ambiguous. The CLI renders an 8-section markdown scaffold pre-populated with anchor evidence; you fill each _[To be filled]_ block with grounded reasoning cited from the gathered evidence, then commit or attach the filled artifact to downstream governance.


Purpose

/gz-justify exposes the canonical gz-justify workflow for operator invocation. It is the operator-facing ritual for Prime Directive invariant 11: when self-reported confidence in a planned change is below 90%, the walkthrough turns "I should think about this more" into an artifact that can be cited, reviewed, and validated.

When to Use

Invoke this skill when your self-reported confidence in a planned implementation is <90%, after a gz-adr-evaluate run scores below 3.0 on an ADR with a tracking GHI or OBPI, before promoting a pool ADR into active work, or mid-pipeline when scope feels ambiguous. The gz-adr-evaluate skill's Low-Score Footer Guidance and the gz-obpi-pipeline skill's Stage 1→2 Confidence Gate both route operators here automatically.

What to Expect

The skill reads its canonical execution contract from .gzkit/skills/gz-justify/SKILL.md (mirrored into .claude/skills/, .agents/skills/, and .github/skills/). Follow the agent-facing instructions in that file for the exact execution protocol, scaffold filling rules, and round-trip validation procedure.

Invocation

Text Only
/gz-justify
Argument / Flag Required Description
<anchor> Yes (or --draft) GHI / OBPI identifier to reason about
--save No Persist to artifacts/justify/<slug>-<timestamp>.md
--output <path> No Persist to an explicit path
--draft <text> / --draft-slug <slug> No Literal draft passthrough
(see SKILL.md) Full flag set is defined by the canonical skill contract

Supporting Files

File Role Read/Write
.gzkit/skills/gz-justify/SKILL.md Canonical skill contract Read
.claude/skills/gz-justify/SKILL.md Claude mirror Read
.agents/skills/gz-justify/SKILL.md Codex mirror Read
.github/skills/gz-justify/SKILL.md Copilot mirror Read
Related Relationship
/gz-adr-evaluate Low-score footer routes to gz-justify
/gz-obpi-pipeline Stage 1→2 Confidence Gate routes to gz-justify
/gz-plan-audit Downstream consumer — filled walkthrough is plan-receipt evidence
skills index Browse the full skill catalog
governance runbook Workflow context