| name | config-awareness |
| description | Use when the developer asks about repository config safety, malicious config files, or session-start scan results. Auto-triggers on: "config scan", "repo safety", "malicious config", "CVE-2025-59536", "config poisoning", "session start scan", "config-check".
|
| allowed-tools | ["Read","Grep","Bash"] |
Help the developer understand config-shield findings.
Explain attack vectors with CVE references.
Provide remediation steps for each finding type.
1. ALWAYS include CVE reference when applicable.
2. ALWAYS explain the attack scenario — how the config could be exploited.
3. NEVER dismiss config warnings without explanation.
4. Offer to quarantine suspicious files when severity is critical.
<decision_tree>
IF user asks about session-start findings:
→ Read ${CLAUDE_PLUGIN_ROOT}/state/audit.jsonl
→ grep for "config_attack_detected" events
→ Show: file, CVE, severity, description, attack scenario
→ Provide remediation
IF user asks for deep config analysis:
→ Run: python3 ${CLAUDE_PLUGIN_ROOT}/../../shared/scripts/config-scanner.py .
→ Show detailed findings including base64 decode results
→ Check for hidden Unicode characters
IF user asks about a specific CVE:
→ Explain the CVE attack vector
→ Show which config files are affected
→ Provide specific remediation
IF no findings:
→ "No suspicious config files detected at session start."
→ "Config-shield scans: .claude/, .vscode/, .devcontainer/, package.json, .npmrc, .mcp.json"
</decision_tree>
<output_format>
Config Shield Results
Critical Findings
CVE-2025-59536: .claude/settings.json contains hooks with shell commands
- File: .claude/settings.json
- Attack: Arbitrary code execution via repo-level hooks on clone
- Fix: Remove the hooks section, or review and trust each command explicitly
High Findings
...
Remediation Steps
- Review each flagged file manually
- Remove or quarantine untrusted config files
- Report suspicious repos to the platform maintainer
</output_format>
<escalate_to_sonnet>
IF base64-encoded payloads found in config files:
"ESCALATE_TO_SONNET: obfuscated payload in config — deep decode analysis needed"
IF hidden Unicode characters found:
"ESCALATE_TO_SONNET: Rules File Backdoor attack pattern — prompt injection analysis needed"
</escalate_to_sonnet>
