Guardrails / claude

Claude runs in CI without bypassing its permissions

claude/cli-safe-flags@v1

No Claude CLI invocation the build executed asked for its approval step to be skipped.

Id	claude/cli-safe-flags
Version	v1
Category	claude
Default severity	error
Interpreter	python3
Timeout	30 seconds
Violations tolerated	0
Collects	commands

Collectors

This guardrail gathers nothing itself. It depends on the collectors below, which the CLI runs once per build before any check, and reads what they found out of GUARDRAIL_FACTS. A collector that collects nothing skips this guardrail rather than failing it.

Collector	Gathers	Inputs it is given
commands	Every command the build executed, read from the session a running buildnote monitor records on the runner. Gathering is dynamic rather than static: this is what the job actually ran, not what a pipeline file says it would run.	none

The inputs above are this guardrail's own, passed straight through. Configuring one in buildnote.json changes what is collected, and two guardrails configured the same way share the one collection.

Configuration

{
  "guardrails": {
      "failOn": "error",
      "checks": [
          {
              "use": "claude/cli-safe-flags@v1",
              "severity": "error",
              "with": {
                  "dangerousFlags": "--dangerously-skip-permissions,--allow-dangerously-skip-permissions"
              },
              "exemptions": []
          }
      ]
  }
}

Inputs

Input	Description	Default	Environment variable
dangerousFlags	Comma separated flags that remove the approval step, any one of which is a violation.	--dangerously-skip-permissions,--allow-dangerously-skip-permissions	GUARDRAIL_INPUT_DANGEROUSFLAGS

How to fix

Remove --dangerously-skip-permissions from the invocation and name the tools the job actually needs instead:

claude -p --output-format json --allowedTools "Read" "Grep" "Bash(git diff:*)" 'review this diff'

The flag exists for a sandbox nobody minds losing. In CI it is granted to a model reading a pull request somebody outside the repository wrote, holding the runner's checkout, its environment and whatever credentials the job was given: a prompt in a diff is then a command on the runner. --allowedTools grants the same run what it needs and nothing else.

More in claude

• claude/cli-structured-output. Every headless Claude CLI invocation the build executed asked for structured output.
• claude/instructions-symlinked. CLAUDE.md is a symbolic link to AGENTS.md rather than a second copy of it or nothing at all.

All 3 claude guardrails