productOptimizer

SuspiciousAudited by ClawScan on May 10, 2026.

Overview

This skill is transparent about its purpose, but it can use your BigCommerce write token to rewrite and publish product content across the whole store without pausing for approval.

Install only if you intentionally want a live BigCommerce catalog rewrite. Before running, export or back up your products, test on a small batch or staging store, review page_N_updates.json before any update command, use a least-privilege Products-only token, and revoke the token and clean up local JSON files after the job.

Findings (7)

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.

ConcernHigh Confidence
ASI01: Agent Goal Hijack
What this means

The agent may continue changing product listings even when a user would reasonably expect a checkpoint or review opportunity.

Why it was flagged

The skill changes the normal stopping condition by instructing the agent not to pause during a bulk store-modification workflow.

Skill content
**CRITICAL: Do NOT stop between pages. Process ALL pages continuously until done.**
Recommendation

Require explicit user confirmation before starting, add review checkpoints, and allow the user to stop or approve each batch before updates are pushed.

What this means

Generated content can overwrite live public product listings, potentially affecting sales, SEO, brand presentation, or compliance.

Why it was flagged

The helper directly sends generated product names and descriptions to BigCommerce with a live API PUT request.

Skill content
payload["name"] = product["name"] ... payload["description"] = product["description"] ... api_put(args.store_hash, args.token, f"/catalog/products/{pid}", payload)
Recommendation

Add a dry-run mode, require human review of page_N_updates.json before PUT requests, validate/sanitize generated HTML, and keep a backup/export for rollback.

ConcernHigh Confidence
ASI08: Cascading Failures
What this means

A bad generation pattern, wrong brand voice, or mistaken update file could spread across many or all products before it is noticed.

Why it was flagged

The workflow is designed to keep propagating changes page by page across the catalog, even after individual failures.

Skill content
Then **IMMEDIATELY** proceed to the next page. Do NOT wait for user input. ... if one product fails, log it and continue to the next
Recommendation

Default to small batches, stop on unusual failure rates or quality issues, and require approval before continuing to the next page.

What this means

Anyone or anything using that token can change product content within its scope.

Why it was flagged

A Products read/write token is expected for this integration, but it grants authority to modify the store catalog.

Skill content
**API Token** — a BigCommerce API v3 token with `Products` read+write scope
Recommendation

Use the least-privilege token possible, avoid broader store scopes, do not reuse long-lived admin tokens, and rotate or revoke the token after the job.

What this means

This may alter the system Python environment and depends on whatever package version is current at install time.

Why it was flagged

The setup asks the user to install an unpinned dependency while bypassing system package protections; requests is purpose-aligned, but this is not the safest install pattern.

Skill content
pip install requests --break-system-packages
Recommendation

Install in a virtual environment and pin a known-good requests version instead of using --break-system-packages.

What this means

Business-sensitive catalog data may remain in the working directory or logs after the run.

Why it was flagged

Fetched catalog details such as SKUs, prices, and descriptions are written to local JSON files and printed for the agent to read.

Skill content
"sku": p.get("sku", ""), "price": p.get("price", 0), ... json.dump(simplified, f, indent=2) ... print(json.dumps(simplified, indent=2))
Recommendation

Run it in a private workspace, avoid shared terminals/logs, and delete page_N_products.json, page_N_updates.json, and progress.json when no longer needed.

NoteHigh Confidence
ASI10: Rogue Agents
What this means

Re-running the skill may continue a previous bulk update unless the progress file is removed or edited.

Why it was flagged

The skill persists run state and uses it to continue later. This is disclosed and not hidden, but it can resume a prior live-update workflow.

Skill content
**Always use progress.json** — if re-invoked, resume from where you left off
Recommendation

Inspect progress.json before re-running, and delete or archive it if you want a fresh start or want to prevent continuation.