AI video editor for creators. Transform raw footage into polished vlogs, talking-head videos, or social media content (TikTok/Shorts/Reels). Control the work...
Security Analysis
high confidenceThe skill's requirements and runtime instructions are consistent with an AI video-editing CLI that uses a Sparki API key and a helper tool runner; nothing requests unrelated credentials or system-wide privileges.
Name/description (AI video editor) align with what the skill does: it runs a sparki CLI, uploads local or Telegram-mini-app videos, polls for tasks, and downloads results. The declared primaryEnv (SPARKI_API_KEY), required binary (uv), and declared FS/network permissions match expected needs for invoking a remote video-editing service and storing outputs.
SKILL.md instructs the agent to run sparki CLI commands (doctor, setup, upload-tg, assets list, edit, status, download) and to check CLI version before running commands. It also instructs reading local files ($CWD) and writing to agent config and its workspace for video outputs. Those actions are reasonable for local-file upload and CLI configuration, but they do grant the skill access to the working directory and to writing into $HOME/.openclaw; users should be aware that videos and related metadata will be transmitted to Sparki's service.
This is instruction-only but includes an installation step using the 'uv' tool: 'uv tool install --upgrade sparki-cli'. That will install the sparki-cli tool (likely from PyPI) into uv's tool bin. Installing a third-party CLI at runtime is expected for this skill but is a moderate-risk action because it executes code obtained from external package repositories; this is proportional to the skill's functionality but worth noting.
Only one credential is declared: SPARKI_API_KEY (the primary credential). The SKILL.md explicitly guides obtaining the key from the Sparki Telegram bot and using 'sparki setup --api-key <KEY>'. No unrelated secrets or high-privilege env variables are requested. The skill will store config under $HOME/.openclaw, which is expected for CLI auth storage.
always is false and the skill does not request permanent platform-wide presence. It writes to its own agent config/workspace paths (permitted) and does not modify other skills or system-wide settings. Autonomous invocation is allowed by default (not flagged by itself).
Guidance
This skill appears to do what it claims, but before installing or using it consider: 1) You will need to provide a SPARKI_API_KEY and the skill will store it in the agent's config — do not paste that key into chat messages. 2) The skill will install a third-party CLI ('sparki-cli') via the 'uv' tool — verify you trust that package source (review sparki-cli on PyPI or the vendor). 3) Videos and metadata will be uploaded to Sparki's service (network access declared for agent-api.sparki.io) and outputs are written to $HOME/.openclaw/workspace/sparki/videos; if your videos contain sensitive information, confirm Sparki's privacy/storage policy first. 4) The SKILL.md requires you to run sparki doctor and to check version consistency — follow that step to avoid running commands from a stale skill description. If you want extra assurance, request the sparki-cli package provenance (PyPI link or GitHub repo) and confirm the network domain and data retention policy before proceeding.
Latest Release
v1.1.0
sparki-cli now installs from PyPI. New commands: sparki doctor, sparki assets delete. upload / edit / run accept positional files / object-keys, plus --dir for directory scans. Style catalog restructured to 4 categories (Vlog / Clips / Narrative / Tools). Error handling information updated.
Popular Skills
Published by @sparki-io on ClawHub
