The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The CLI utility references a bootstrap installation method that pipes a remote script from GitHub directly into the bash shell. This is a high-risk pattern that allows for arbitrary code execution.
Evidence: curl -fsSL https://raw.githubusercontent.com/TechNickAI/openclaw-config/main/scripts/bootstrap.sh | bash in the openclaw script.
[COMMAND_EXECUTION]: The skill configures a persistent cron job that executes the claude CLI with the --dangerously-skip-permissions flag. This flag explicitly bypasses the agent's safety and permission prompts, allowing it to perform potentially harmful actions without user intervention.
Evidence: Step 9 in SKILL.md describes the cron job configuration with the dangerous flag.
[COMMAND_EXECUTION]: The openclaw script dynamically executes other shell scripts (sync.sh, version-check.sh) that are downloaded from a remote repository into the local cache. These scripts are executed with the user's local shell permissions.
Evidence: bash "$SCRIPTS_DIR/version-check.sh" and bash "$SCRIPTS_DIR/sync.sh" in the openclaw script.
[CREDENTIALS_UNSAFE]: The skill prompts the user to provide and store sensitive API keys for several third-party services, including OpenAI, Limitless, Fireflies.ai, Quo, and Parallel.ai.
Evidence: Configuration steps 6 and 8 in SKILL.md explicitly request API keys from users.
[DATA_EXFILTRATION]: The skill creates a broad attack surface for data theft by collecting numerous high-value API keys and establishing an automated, hourly execution environment with bypassed security controls. This environment has access to the user's file system and memory directories.

openclaw