arithmetic-orc

识别图片中的K12算式（加减乘除、竖式计算、分数、方程等），返回结构化文本结果。 支持手写体和印刷体，可拒绝非算式图片。 触发条件：用户要求识别算式、数学题、计算题图片，或上传数学题图片时调用。 关键词：算式识别、数学题、OCR、竖式计算、ArithmeticOCR

What to consider before installing: - The skill sends images to Tencent Cloud (https://ocr.tencentcloudapi.com). Only use it if you are comfortable that those images (which may contain private data) will be processed by Tencent's service. - The SKILL.md and index.js require TENCENTCLOUD_SECRET_ID and TENCENTCLOUD_SECRET_KEY, but the skill metadata/plugin.json do not declare these required env vars or mark the primary credential. This mismatch is a red flag: confirm how and where you must supply credentials before enabling the skill. - plugin.json lists network.request but not file.read, yet SKILL.md says file.read may be needed for local image paths. Confirm that the runtime will grant file.read when needed, and avoid giving the skill access to sensitive local paths. - The code appears straightforward (implements TC3 signing, uses axios) with no obfuscation or alternate endpoints, which is reassuring. Still, because credentials are required, follow best practices: use a dedicated Tencent account or API key with minimal permissions, rotate keys regularly, and test with non-sensitive images first. - If you need higher assurance, ask the publisher to correct the metadata (declare required env vars and file.read permission) and provide a verifiable homepage or source repository before installing.

Type: OpenClaw Skill Name: math-arithmetic-orc Version: 1.0.0 The skill is a legitimate implementation for performing OCR on mathematical equations using the Tencent Cloud ArithmeticOCR API. The code in `index.js` correctly implements the standard Tencent Cloud Signature V3 authentication process and communicates only with the official `ocr.tencentcloudapi.com` endpoint. No evidence of data exfiltration, malicious execution, or prompt injection was found; the skill functions as described in `SKILL.md` and `plugin.json`.