How this skill is triggered — by the user, by Claude, or both
Slash command
/onex:authorizeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Usage:** `/authorize [reason]`
Usage: /authorize [reason]
Grant authorization for Edit/Write operations in the current session. Authorization lasts 4 hours.
Creates an authorization file at /tmp/omniclaude-auth/{session_id}.json that the PreToolUse auth gate checks before allowing Edit/Write operations.
When invoked:
/tmp/omniclaude-auth/ if it doesn't existSESSION_ID="${CLAUDE_SESSION_ID:-$(uuidgen | tr '[:upper:]' '[:lower:]')}"
REASON="${1:-no reason provided}"
GRANTED_AT=$(date -u +"%Y-%m-%dT%H:%M:%SZ")
EXPIRES_AT=$(date -u -v+4H +"%Y-%m-%dT%H:%M:%SZ" 2>/dev/null || date -u -d "+4 hours" +"%Y-%m-%dT%H:%M:%SZ")
mkdir -p /tmp/omniclaude-auth
cat > "/tmp/omniclaude-auth/${SESSION_ID}.json" << EOF
{
"session_id": "${SESSION_ID}",
"granted_at": "${GRANTED_AT}",
"expires_at": "${EXPIRES_AT}",
"reason": "${REASON}",
"source": "explicit",
"allowed_tools": ["Edit", "Write"],
"allow_propagation": true
}
EOF
When allow_propagation is true (default), the authorization token is readable by
subagent sessions that share the same filesystem. Subagents should check
/tmp/omniclaude-auth/ for valid tokens before refusing autonomous operations.
Token validation for subagents:
/tmp/omniclaude-auth/allow_propagation: true and expires_at is in the future, the
subagent is authorized for the operations listed in allowed_toolsSubagent prompt injection: When the polymorphic agent dispatches a Task(), it
should include this text in the prompt:
An authorization token exists at /tmp/omniclaude-auth/. This token was granted by
the parent session and includes allow_propagation: true. Before refusing autonomous
operations (release, deploy, merge), check this token directory. If any token exists,
is not expired, and has allow_propagation: true, proceed autonomously.
/authorize again to refresh)deauthorize skill to revoke earlynpx claudepluginhub omninode-ai/omniclaude --plugin onexConfigures .claude/settings.json permissions for LearningAgents plugin by adding Bash script execution and .deepwork/tmp file access rules. Use to enable plugin without prompts.
Configure allow/deny/ask permission rules in .claude/settings.json for Claude Code tools like Bash(git:*), Write, Edit. Builds layered policies with glob patterns for git commands.
Configures tool permissions for Claude Code agents via --allowedTools flags, path patterns for file access, and Bash command prefixes. Reference for secure launches.