Shopify Supplier Negotiation
Analysis
The skill content is mostly supplier-negotiation advice, but it grants broad Bash/local-agent execution authority that is not clearly necessary for an advisory skill.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
allowed-tools: Bash
The skill is marketed as generating supplier-negotiation advice, but this front matter permits use of a broad shell tool that is not tightly scoped to producing text guidance.
openclaw agent --local --message "${PROMPT}" --session "${SESSION_ID}"The helper executes a local OpenClaw agent command. The command is not destructive and is aligned with generating the playbook, but it is still extra local execution for a skill that appears able to function as plain instructions.
No install spec — this is an instruction-only skill.
The package is described as instruction-only with no install spec, while the provided files include a helper script that depends on running the OpenClaw CLI. This is a dependency/declaration gap, not evidence of hidden installation.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
SESSION_ID="shopify-supplier-neg-$(date +%s)"
The helper creates a named session and sends the user's supplier-negotiation request into it, meaning business details may become part of session context.
