markdown-extract
v1.0.0Extract clean markdown from any URL using auto, AI, or browser methods via the markdown.new API with error handling and flexible extraction options.
⭐ 0· 561·2 current·2 all-time
by@aviclaw
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
high confidencePurpose & Capability
Name/description align with what the code and SKILL.md do: they call https://markdown.new to convert pages to markdown. However the Python implementation invokes the external 'curl' binary but the skill metadata declares no required binaries—this is an incoherence (the runtime actually needs 'curl' available).
Instruction Scope
SKILL.md and extract.py confine themselves to making HTTP requests to markdown.new (GET or POST) and returning the markdown or error. They do not read local files, environment variables, or unrelated system configuration. Error handling and Cloudflare-block detection are limited but present.
Install Mechanism
There is no install specification (instruction-only), which is low-risk. The package does include a runnable Python script but does not attempt to download or install external code at install time.
Credentials
The skill requests no credentials or env vars, which is appropriate; however it will send the target URL (and for POST flows likely the page content or a request to fetch it) to markdown.new. That can leak URLs or page content to a third party. Also, the code requires the 'curl' binary but does not declare it. Finally, because the skill fetches arbitrary URLs, it can be used (by the user or an autonomous agent) to probe internal endpoints or expose internal content (SSRF / data-leak risk) if run in an environment with access to internal network ranges.
Persistence & Privilege
The skill does not request persistent system privileges, does not set 'always: true', and does not modify other skills or system-wide configuration.
What to consider before installing
Before installing, note these points: (1) The Python script uses the system 'curl' command but the skill metadata does not declare it—ensure curl is available or update the manifest. (2) All requests (the requested URL and any fetched content) are sent to markdown.new; verify you trust that endpoint and avoid using the skill on sensitive or internal URLs. (3) Because the skill fetches arbitrary URLs, there is an SSRF/data-exposure risk if the agent runs in an environment with access to internal services. (4) The code uses subprocess.run with argument lists (not a shell string), which reduces shell-injection risk, but still review the script before granting execution. If you want to proceed, test with non-sensitive public pages and consider adding an explicit required-binaries entry for 'curl' and a manifest note about external data transmission.Like a lobster shell, security has layers — review code before you run it.
latestvk9738f973p16atkd07zvmgt181816f7v
License
MIT-0
Free to use, modify, and redistribute. No attribution required.