Skills

Skills

mq-agent ships with built-in skills for structured AI agent workflows and mq ecosystem integrations.

Skills live in skills/. Most map directly to CLI commands and agent classes; integration skills document how mq-agent should route work to neighboring mq tools.

Built-in skills

repo-audit

Read-only repository audit using the Planner→Executor→Verifier loop.

skills/repo-audit/SKILL.md

Command: mq-agent audit .

release-readiness

Full release validation: git state, tests, version, changelog, CI.

skills/release-readiness/SKILL.md

Command: mq-agent release-check

signal-assessment

repo-signal static scan + AI improvement plan with scored output.

skills/signal-assessment/SKILL.md

Command: mq-agent signal .

ci-diagnosis

Diagnose CI failures and generate actionable fix steps.

skills/ci-diagnosis/SKILL.md

Command: mq-agent fix-ci

visual-analysis

Analyze images, screenshots and visual diffs through mq-image-analyze.

skills/visual-analysis/SKILL.md

Command: mq-image analyze, mq-image analyze-ui, mq-image compare

mq-mcp-review-orchestration

Route review, risk, security and architecture workflows through mq-mcp without duplicating cognition logic in mq-agent.

skills/mq-mcp-review-orchestration/SKILL.md

Command: future mq-agent review / task-runner backed review workflows

Safety modes

All skills respect the safety gate:

Mode	Behaviour
read-only	Only read tools allowed
suggest	Plan shown, nothing executed
execute	Runs after safety check
dangerous	No restrictions

Run a skill

mq-agent audit .              # repo-audit
mq-agent signal .             # signal-assessment
mq-agent release-check        # release-readiness
mq-agent fix-ci               # ci-diagnosis
# future: mq-agent review .   # mq-mcp-review-orchestration

All commands support --dry-run and --json.