Skip to content
Back to skills

Block Wildcard Agent Permissions

ASecurity

Blocks agent grants that allow everything: a bare Bash wildcard, a `*` in an allow/alwaysAllow/tools array on the same line, or quoted defaultMode bypassPermissions. Any file is judged. Not caught: a multi-line array (`\"*\"` alone on a line, YAML `- \"*\"` list), unquoted defaultMode: bypassPermissions. Runs: commit, agent write, turn's end. Waiver: 'pragma: allowlist broad-agency' same line. Person's commit: honoured. Agent: only if already in HEAD; MCP gateway: never. Agent asks a person.

  • 3 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 13, 2026
ai-agentsbashgit

Works with

  • cli
  • mcp

Security analysis

A100/100

Scanned October 1, 2026

npx -y skills add open-coder-ai/chock-catalog --skill block-wildcard-agent-permissions --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Block Wildcard Agent Permissions?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Block Wildcard Agent Permissions
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/open-coder-ai-block-wildcard-agent-permissions-chock-catalog/badge)](https://www.skillsdirectory.com/skills/open-coder-ai-block-wildcard-agent-permissions-chock-catalog)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: block-wildcard-agent-permissions
description: "Blocks agent grants that allow everything: a bare Bash wildcard, a `*` in an allow/alwaysAllow/tools array on the same line, or quoted defaultMode bypassPermissions. Any file is judged. Not caught: a multi-line array (`\"*\"` alone on a line, YAML `- \"*\"` list), unquoted defaultMode: bypassPermissions. Runs: commit, agent write, turn's end. Waiver: 'pragma: allowlist broad-agency' same line. Person's commit: honoured. Agent: only if already in HEAD; MCP gateway: never. Agent asks a person."
metadata:
  chock.artifact: hook
  chock.enforcement: block
  chock.coverage_without_chock: advisory
---

# Block Wildcard Agent Permissions

Blocks agent grants that allow everything: a bare Bash wildcard, a `*` in an allow/alwaysAllow/tools array on the same line, or quoted defaultMode bypassPermissions. Any file is judged. Not caught: a multi-line array (`"*"` alone on a line, YAML `- "*"` list), unquoted defaultMode: bypassPermissions. Runs: commit, agent write, turn's end. Waiver: 'pragma: allowlist broad-agency' same line. Person's commit: honoured. Agent: only if already in HEAD; MCP gateway: never. Agent asks a person.

```
on(commit|tool_use): block(content_regex) scan=added_lines allowlist_pragma=pragma:\s*allowlist\s+broad-agency content_pattern(regex)
Wildcard agent permission grant detected. Scope the grant to specific tools or commands (e.g. Bash(git status:*), a named tool list); replace defaultMode bypassPermissions with a mode that asks (default or acceptEdits). Waiver: 'pragma: allowlist broad-agency' on the same line. A person's commit honours it; in the agent (write, the turn's end, an agent's commit) only a line already in HEAD counts, and the MCP gateway never does. An agent asks a person; it never writes the pragma.
```

This skill is advisory: the client reading it has no mechanism to enforce it. The same policy compiled by `chock` blocks at commit, on an agent's file writes and at turn end. See https://github.com/open-coder-ai/chock

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…