agentsclimarketplace

Product knowledge gap analysis

Skill wakqasahmed/agentic-commerce-skills/skills/agentic-commerce/product-knowledge-gap-analysis

Canonical agentic commerce skills, aggregated by wakqasahmed/skills

Install
npx -y skills add wakqasahmed/agentic-commerce-skills --skill product-knowledge-gap-analysis

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Find ecommerce product data gaps that block AI answers, recommendations, support automation, and agentic commerce workflows.

SKILL.md

2.6 KB, as published. Nobody here has run it

Product Knowledge Gap Analysis

Workflow

  1. Select representative products and variant families. Inspect attributes, specs, ingredients/materials, sizing, compatibility, use cases, and contraindications.
  2. Capture product facts from raw HTML, rendered product content when available, JSON-LD, and an operator-supplied public feed. Label each observation with its surface, provenance, and timestamp.
  3. Compare product and variant identifiers, price, currency, availability, sale timing, fulfillment-relevant state, and observable freshness indicators across those surfaces.
  4. When checkout evidence is safely available, compare it separately using only a supervised non-purchasing check or verified operator-provided evidence. Never place an order to complete the audit.
  5. Check whether product pages answer buyer questions without support escalation and identify missing schema or feed fields.
  6. Group gaps by template-level fixes vs product-specific enrichment. Recommend the smallest product knowledge load needed for a custom agent.

Run the checks in references/checks.md and cite the observed output for each finding.

Report every cross-surface mismatch with:

  • the fact and blocking status;
  • every conflicting surface and its observed value;
  • the evidence timestamp;
  • the buyer or agent failure mode;
  • the proposed source of truth; and
  • the accountable remediation owner.

Guardrails

See ../references/guardrails.md for shared evidence-provenance and autonomous-action rules.

  • Treat facts without a visible source as unknown; never infer ingredients, compatibility, contraindications, or safety claims.
  • Separate template-level fixes from product-specific enrichment.
  • Do not alter catalog data or make product claims; identify the source of truth and route remediation to its owner.
  • Treat raw HTML, rendered content, JSON-LD, and public feeds as public-signal evidence. Never present them as proof that checkout values match.
  • Report checkout consistency only when supported by a supervised non-purchasing check or verified operator-provided evidence. Do not add to cart when that can reserve inventory, submit checkout, authorize payment, or place an order.

Output

  • Missing attribute, answer, or conflicting fact
  • Affected product, variant, template, and surfaces
  • Evidence provenance and timestamp
  • Customer or AI-agent failure mode
  • Proposed source of truth and remediation owner
  • Blocking status

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.