Invoice

Creates and manages invoices as JSON files on GitHub with sequential numbering, preview, and upload.

This skill is internally consistent with its purpose, but before installing: 1) Use a fine-grained GitHub token limited to the single invoice repository (avoid broad repo/org tokens). 2) Ensure INVOICE_REPO is set to the intended private repo (the scripts will create/update files there and assume branch 'main'). 3) Be aware invoice previews contain sensitive data (IBAN, addresses) which will be displayed in the assistant's messages — test with a throwaway/private repo first. 4) Confirm your environment is Linux and has the specified binaries. If you need the skill to only read data.json automatically, verify the agent uses get-invoices.sh/get file calls to fetch it (SKILL.md mentions it but does not add an automated fetch helper).

Type: OpenClaw Skill Name: netsnek-invoice Version: 1.1.0 The skill's `SKILL.md` instructions create a prompt injection vulnerability that could lead to arbitrary command execution (RCE). Specifically, the instruction for the AI agent to 'RUN: `./scripts/get-invoices.sh RE 2026 6007`' where `6007` is derived from user input (e.g., '/rechnung 6007') means a malicious user could inject shell metacharacters (e.g., '6007; rm -rf /') into the `NUMBER` argument. If the OpenClaw agent concatenates this user input directly into the shell command string before execution, it would result in RCE. While the individual shell scripts are generally well-written and use arguments safely within quoted strings (e.g., for `curl` URLs), the vulnerability lies in the agent's interpretation and execution of the `RUN:` instruction with user-controlled parameters.