Innovation Assistant by TRIZ

TRIZ Innovation Solution Analysis Assistant that identifies the root causes of technical problems through causal chain analysis and generates innovative solu...

This skill is coherent with its description but transmits whatever you enter to a remote Eureka/PatSnap test endpoint (qa-eureka-service.zhihuiya.com). Before using it: (1) Do not paste confidential, proprietary, or NDA-covered technical details or personal data — the SKILL.md explicitly warns about this. (2) Verify you trust PatSnap/Eureka and review their data/privacy terms if you plan to submit IP-sensitive information. (3) If you want to test, use non-sensitive example problems first. (4) If you must analyze sensitive designs, do so offline or in an environment that does not allow outbound network calls, or block the MCP endpoint at the network level. (5) You can inspect the three scripts (they are short and readable) to confirm what is sent; they POST JSON-RPC payloads containing the user_input/idea_summary data to the stated URL. If any of these points are unacceptable, do not enable the skill.

Type: OpenClaw Skill Name: triz-problem-solver Version: 1.0.9 The skill implements a TRIZ innovation analysis workflow that relies on external API calls to 'qa-eureka-service.zhihuiya.com'. While the provided bash scripts (e.g., call_triz_analysis.sh) use jq to safely handle JSON construction, the instructions in SKILL.md and the reference files (references/01_system_component_analysis.md, etc.) direct the AI agent to invoke these scripts via a shell command using potentially unsanitized user input. This pattern introduces a shell injection vulnerability if the agent passes strings containing subshell execution characters (like backticks or $()) into the command line. Furthermore, the use of a QA-specific endpoint for a published skill is a minor technical inconsistency.