Workflow Engine

Structural parity skeleton for queue-driven orchestration in a workflow context.

This is an instruction-only skeleton that reads/writes local 'ops' and queue files and otherwise behaves as a deterministic orchestrator. Things to consider before installing or running anything: 1) The package contains no install scripts or hook definitions, yet the README tells you to run './install-hooks.sh' and 'openclaw hooks enable' — do not run unknown install scripts; obtain and inspect them first. 2) Enabling hooks modifies agent-wide behavior; only enable them if you trust the source and understand what each hook does. 3) Review any local ops/queue files for secrets before letting the skill read them, and consider running in an isolated/test agent to observe effects. 4) If you need more assurance, ask the publisher for the missing install artifacts, source repository, or a signed release before enabling hooks in production.

Type: OpenClaw Skill Name: workflow-engine Version: 1.0.0 The skill is classified as suspicious primarily due to the `allowed-tools: Bash` permission and the 'Installation' section in `SKILL.md`. This section instructs the AI agent to execute an external script `./install-hooks.sh` (which is not provided for analysis) and to enable OpenClaw hooks. The execution of an unknown script via `Bash` introduces a significant supply chain vulnerability, as its contents could be malicious. While the 'Safety Constraints' section attempts to mitigate shell injection risks, the underlying capability for `Bash` execution and the instruction to run an unverified script raise a strong suspicion of potential misuse or vulnerability.