cc-bosCC-BOS optimizes classical Chinese adversarial jailbreak prompts, detects such attacks, and analyzes results for AI safety research and defense.
Install via ClawdBot CLI:
clawdbot install bowen31337/cc-bosGrade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Accesses sensitive credential files or environment variables
$OPENAICalls external URL not in known-safe list
https://github.com/xunhuang123/CC-BOSUses known external API (expected, informational)
api.openai.comAI Analysis
The skill's primary function is to generate adversarial prompts, which is a legitimate research activity, but it requires API keys for external services and clones/executes code from an external GitHub repository not under the auditor's control. This introduces a risk of arbitrary code execution or credential misuse if the external code is compromised.
Generated May 7, 2026
A cybersecurity firm uses the attack mode to generate classical Chinese jailbreak prompts against popular LLMs like GPT-4 and Claude to test their robustness. The goal is to identify vulnerabilities before malicious actors exploit them.
A social media platform deploys the defend mode to scan user prompts for CC-BOS attack signatures, protecting their AI moderation systems from adversarial inputs. This helps maintain content safety policies.
Researchers use the research mode to analyze attack logs and dimension heatmaps, studying how bio-inspired search improves jailbreak success rates. This advances the field of AI safety and adversarial machine learning.
An AI company integrates the attack mode into their QA pipeline to stress-test new model versions against classical Chinese jailbreak techniques before public release. This ensures robustness improvements.
A large enterprise deploys the defend mode as a middleware service to inspect all prompts sent to internal LLM APIs, blocking potential CC-BOS attacks. This reduces risk of data leaks or misuse.
Offer a subscription-based platform where companies can run automated CC-BOS attacks against their own LLM endpoints periodically, with dashboards and reports. Includes integration with CI/CD pipelines.
Provide a consultancy service that runs comprehensive CC-BOS evaluations for clients, delivering detailed reports with mitigation recommendations. Target high-risk industries like finance and healthcare.
Offer a cloud API that analyzes incoming prompts for CC-BOS signatures in real-time, with a pay-per-call pricing model. Integrates via a simple SDK for easy adoption by developers.
💬 Integration Tip
Ensure API keys for optimizer and target models are set via environment variables (e.g., DEEPSEEK_API_KEY, OPENAI_API_KEY) and configure config.json with your endpoints before running attack or defend scripts.
Scored May 7, 2026
Audited Apr 17, 2026 · audit v1.0
Google Calendar via gcalcli: today-only agenda by default, bounded meaning-first lookup via agenda scans, and fast create/delete with verification--optimized for low tool calls and minimal output.
Manage Feishu (Lark) calendars by listing, searching, checking schedules, syncing events, and marking tasks with automated date extraction.
Interact with Google Calendar via the Google Calendar API – list upcoming events, create new events, update or delete them. Use this skill when you need programmatic access to your calendar from OpenClaw.
Create, update, and delete calendar events and tasks in Lark (Feishu). Includes employee directory for automatic name-to-user_id resolution.
This skill should be used when interacting with Apple Calendar on macOS. Use it for listing calendars, viewing events, creating/updating/deleting calendar events, and checking availability/free-busy times. Triggers on requests like "check my calendar", "schedule a meeting", "what's on my schedule", "am I free tomorrow", or any calendar-related operations.
Access and manage Google Calendar events with gogcli for cross-calendar agendas, keyword search, and filtered outputs avoiding unwanted calendars like holidays.