Inventory Waste Optimizer

AdvisoryAudited by Static analysis on May 7, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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.

ConcernMedium Confidence
ASI02: Tool Misuse and Exploitation
What this means

The agent could potentially approve or create normal inventory orders without the user reviewing every purchase, which can affect business spending and stock levels.

Why it was flagged

This shows the skill contemplates purchase-order creation/control. Combined with the declared can-make-purchases capability, the artifacts do not clearly require explicit user approval for POs that fall within the listed thresholds.

Skill content
Price Jump Guard — Block the creation of POs where a vendor's unit price has increased >20% since the last cycle.
Recommendation

Require explicit human confirmation for every purchase or PO, set maximum spend/vendor/SKU limits, run in simulation first, and keep audit logs for all proposed and executed orders.

What this means

Granting broad spreadsheet access could expose sensitive sales and inventory data or allow unintended edits if the account permissions are too wide.

Why it was flagged

The skill needs access to primary business spreadsheets. This is purpose-aligned, but the registry declares no primary credential or required configuration paths, and the docs do not specify OAuth scopes or read/write boundaries.

Skill content
Connect the agent to your primary "Sales" and "Inventory" Google Sheets.
Recommendation

Use least-privilege sharing, preferably a dedicated service account or restricted copies of the sheets, and confirm whether the skill needs read-only or write access.

What this means

Running an unverified setup package could modify the local agent environment in ways not visible from this skill’s artifacts.

Why it was flagged

The setup relies on an external npm package run via npx, with no pinned version or reviewed package contents included in the artifacts. It is disclosed and purpose-related, but still a supply-chain dependency users should verify.

Skill content
Install ThumbGate: `npx thumbgate init --agent openclaw`
Recommendation

Inspect the ThumbGate package, pin a known version, verify its publisher, and run setup in a controlled environment before using it with real purchasing workflows.

What this means

Inventory, vendor, price, or order information could be sent to third-party alerting tools depending on configuration.

Why it was flagged

The optional alerting path sends events through external automation and messaging services. This is disclosed and purpose-aligned, but the artifacts do not specify what inventory/order data is shared.

Skill content
Use [Make.com](https://make.com) ... to route Slack or Email alerts when ThumbGate catches an over-order.
Recommendation

Limit alert payloads to necessary fields, avoid including confidential vendor/pricing details, and review Make.com, Slack, and email retention/access policies.

NoteLow Confidence
ASI10: Rogue Agents
What this means

If continuous monitoring is enabled elsewhere, it may keep processing business data beyond a single user-invoked task.

Why it was flagged

The artifact advertises continuous autonomous monitoring, but the provided files do not explain the runtime lifecycle, stop controls, or persistence mechanism.

Skill content
This autonomous agent monitors your inventory 24/7
Recommendation

Confirm how the agent is scheduled, how to disable it, and whether it has limits on when it may read sheets or initiate purchase workflows.