harness-quality-gate

Runs measured quality gates and risk routing to produce binary pass/fail verdicts per change.

Updated Jul 29, 2026
One-click install
npx skills add https://github.com/MaiconGambini/opencode-harness-guide --skill harness-quality-gate-maicongambini
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harness-quality-gate
Source: https://github.com/MaiconGambini/opencode-harness-guide/tree/main/skills/harness-quality-gate
Command: npx skills add https://github.com/MaiconGambini/opencode-harness-guide --skill harness-quality-gate-maicongambini

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Teams need an objective, numbers-based way to decide whether a code change is safe to commit or return from a work lane, without relying on subjective review depth or unmeasured thresholds. ## Core Features & Use Cases - Measured quality gate: Runs scripts/harness-quality-gate.mjs in local mode (inside an implementation lane) or full mode (scheduler, before the single commit) and reacts to exit codes 0, 1, or 2. - Risk routing: Pairs the gate verdict with the harness-risk-router tier so review depth is decided by evidence, not guesswork. - Controlled write-back: Proposes at most one improvement per run — threshold ratchets after three consecutive beats, or escaped-defect entries in docs/review.md — while never editing thresholds or rulebooks directly. - Use Case: Before committing a scheduler batch, run the gate in full mode; if a blocking metric is red, fix the metric rather than loosening the threshold, and route the change to the appropriate review tier. ## Quick Start Run the harness quality gate in local mode for my current lane and report the verdict, tier, and metric table.

Frequently Asked Questions about harness-quality-gate

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run the harness quality gate before a commit?

Run the gate with --mode full --label scheduler before the single commit, using the absolute invocation from docs/harness/measured-gates.md. Inside an implementation lane, use --mode local --label <lane-id> instead.

What do the quality gate exit codes 0, 1, and 2 mean?

Exit 0 means pass or breaches only in observe-mode metrics. Exit 1 means a blocking metric is red or a threshold was loosened without a recorded reason. Exit 2 means the gate itself could not run, which is a harness blocker, not failing code.

What happens if quality-thresholds.json is missing?

Propose bootstrapping it at Phase A, where everything except regression_suite is set to observe. Never block a project that has never measured itself, but never call it green either — missing reports route to the full review tier.

Can the quality gate change its own thresholds automatically?

No. The gate never edits thresholds directly. If a metric beats its threshold on three consecutive runs, it proposes raising the threshold to the lowest of those three, but the change requires the recorded process in quality-decisions.md.

When should I use harness-quality-gate versus harness-quality-snapshot?

Use the gate for per-change, binary pass/fail decisions before commits or lane returns. Use the snapshot for per-milestone, qualitative scorecards; its evidence rows cite the latest gate report rather than re-deriving metrics.