Check

Run the repository's configured quality gates command and return a concise, high-signal summary with log pointers.

This command is intended for humans to invoke directly. Agents should preferentially call @runner directly for execution/log-heavy tasks.

User invocation: /check [fast|slow|all|...] [--skip-autofix] [--dry-run]

Examples: /check # default (usually all) /check fast /check slow /check lint test

Determine which quality gates command to run:

1. Read AGENTS.md and look for an explicit quality gates runner instruction.

• If AGENTS.md includes a command like ./scripts/quality-gates.sh (preferred) or any referenced path/command for quality gates, use that.
• If multiple are present, prefer the most explicit "Run all quality gates" instruction.

1. Default fallback if no instruction found:

• ./scripts/quality-gates.sh

1. Pass through user-provided arguments (fast/slow/all/...) as-is to the resolved command.

1. Always run from repository root.

• Delegate actual execution to @runner (this command uses it as its agent).
• Ensure logs are saved under tmp/run-logs-runner// and that output includes:
• exact command
• exit code
• duration
• log path(s)
• top error snippets and tail excerpts
• If quality gates fail, prominently surface:
• which gate(s) failed
• pointers mentioned by quality-gates.sh (e.g., tmp/playwright-report, tmp/playwright-report/ai-failures.jsonl)

• Do not attempt fixes; this command is run-only.
• For fixing failures, use /check-fix (@fixer) or invoke @fixer directly.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: juliusz-cwiakalski
• Source: juliusz-cwiakalski/agentic-delivery-os
• License: MIT

Install and usage instructions live in the source repository linked above.