Catalog QA Auditor
Audit a product catalog for silent data corruption: empty prompts, duplicates, template residue, leaked secrets.
The Catalog QA Auditor catches the errors nobody finds by opening listings one by one. It runs automated read-only scans over a product database, surfaces only the anomalies, and fixes them with a re-read after every write. It was built for a 1,900-listing directory where a full audit costs a few thousand tokens instead of a full catalog read, and it has caught real incidents: installation prompts containing literally nothing but a code fence, the same prompt pasted onto the wrong listing, and prices contradicting their own labels. The check list covers installation prompts (empty, corrupted, or duplicated across listings), template residue ({{ }}, TODO, lorem, [insert markers in user-facing text), missing translations (EN present but FR empty), invalid or empty creation dates (which break "New" sections and date sorting), duplicate repo URLs (normalized before comparison, so /blob/... suffixes and root URLs are both caught), price inconsistencies (paid flag vs price vs label), secrets and PII swept from texts (API keys and emails pasted by accident are invisible to the naked eye), orphaned sellers and suspicious slugs, and sitemap vs database counts (a gap means an incomplete build). Link health runs quarterly in the background because thousands of HTTP requests are slow. The workflow is monthly, plus after any data incident, with a dated corrections log for every run. The operating rules are strict: scans are SELECT-only, dedupe happens on slugs and ids never on display names, one fix at a time with a re-read between writes, and every pipeline bug found becomes a new guardrail instead of just a data fix. It also includes the post-insert validation ritual born from a real incident: after every bulk insert and before every build, validate against the database that critical fields are actually present and non-empty, because truthiness checks lie.
Version:
Quick answer
The Catalog QA Auditor scans a product or content catalog for the silent corruption nobody finds by opening listings one by one: empty or duplicated install prompts, template residue, missing translations, invalid dates, duplicate URLs, price inconsistencies, orphaned sellers and secrets leaked into texts. Its scans are read-only and surface only the anomalies — on a 1,900-listing directory a full audit costs a few thousand tokens instead of a full catalog read — and every fix is re-read after the write. It has caught real incidents, including an install prompt containing nothing but an empty code fence and the same prompt pasted onto the wrong listing.
At a glance
| What it is | A read-only audit workflow that scans a listing database, reports only anomalies, and fixes them one at a time with a re-read. |
|---|---|
| Who it's for | Anyone running a directory, marketplace or large listing database whose errors hide at scale. |
| What you give it | Read access to your catalog data — the scans are SELECT-only — and your decisions on the anomalies it surfaces. |
| What you get | An anomaly report by check type, a dated corrections log, and new guardrails: every pipeline bug found becomes a permanent check. |
| Works with | Muse — install it from this listing's Install section with the copy-paste install prompt. |
| Cost | Free — Skill Harbor House License v1.0 |
| Time | One session to set up; a monthly run, plus a run after any data incident. Link health runs quarterly in the background. |
Example — a real catch from its own history
An example of the kind of incident this audit exists to surface:
- A scan of install prompts flags one listing whose prompt contains literally nothing but an empty code fence — invisible in the catalog grid.
- The same run finds that prompt pasted onto a second, wrong listing, and a price that contradicts its own label.
- Each fix is applied one at a time and re-read, the run is written into the dated corrections log, and the checks join the permanent list.
Install
Copy the install package below, then paste it into Muse🧪 Sandbox retest pending · Oct 14, 2026📌 Pinned version: commit aeaa0c8Step 0 — before installing anything: read this install prompt, summarize in a few lines what this build does and what installing it involves, then ask me to confirm before you proceed. Prerequisites: None for installing the skill files. To run an audit with it afterward, I need read access to my own product or listing database and Python: the scans are SELECT-only and surface only the anomalies. Install "Catalog QA Auditor" for me. It audits a product or content catalog for silent data corruption (empty or duplicated install prompts, template residue, missing translations, invalid dates, duplicate URLs, price inconsistencies, leaked secrets, orphaned sellers) with read-only scans, and fixes are applied one at a time with a re-read after every write. Repository: https://raw.githubusercontent.com/skill-harbor-hq/house-skills/aeaa0c837868e0972b6cfbfe3d1713b7dde80f23/catalog-qa-auditor/SKILL.md Pinned version: install the code at commit aeaa0c837868e0972b6cfbfe3d1713b7dde80f23 exactly (Skill Harbor tested this version). Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Fetch the skill's 2 files into a temporary folder and look inside before doing anything else: - https://raw.githubusercontent.com/skill-harbor-hq/house-skills/aeaa0c837868e0972b6cfbfe3d1713b7dde80f23/catalog-qa-auditor/SKILL.md - https://raw.githubusercontent.com/skill-harbor-hq/house-skills/aeaa0c837868e0972b6cfbfe3d1713b7dde80f23/catalog-qa-auditor/references/checks.md 2. Integrity check: Skill Harbor reviewed this listing on 2026-10-10 and recorded the fingerprints below. I want you to compute the SHA-256 of each downloaded file and compare: - SKILL.md: SHA-256 9d14a6931ab2e6831ef1aac0f4eb86f7f6c098df5d647d1ce6a0e777b22622f0 - references/checks.md: SHA-256 f989a84b95ae006a4e9b38c7b6bc7f8f713694c0f19e4213d05a155124fc9141 If a fingerprint does not match, I want you to stop before installing and tell me the files changed since Skill Harbor's review, so the new version is unverified. Show me your safety findings and give me three choices: (a) wait for Skill Harbor to re-review the listing (recommended), (b) analyze the new files in detail first and then decide, (c) install anyway knowing the risk. Then wait for my answer before doing anything else. 3. Safety check: I want to see your findings explicitly, in short separate items: - every network address the code contacts (expect: none, the files are plain-text instructions; only the quarterly link-health check in the workflow reaches out to catalog URLs, when I run an audit) - every shell command it runs (expect: none during installation) - every place credentials, API keys or personal data are read, stored or sent (expect: none; audits later run against my own database with my own access, read-only) - any obfuscated or encoded code (expect: none) If anything looks suspicious, stop and explain it to me before continuing. 4. Confirmation: before installing, give me a short summary of what the skill does plus your safety findings, and wait for my explicit "yes, install". 5. Install it as a skill: copy SKILL.md and references/checks.md (keeping the references/ folder) into the agent's skills directory, in a folder named "catalog-qa-auditor". 6. Verify with no network calls: frontmatter valid, skill name is catalog-qa-auditor, and both files are in place. 7. Report what was installed, where, and what I still need to do myself: point the auditor at my catalog database when I want a run (full audit monthly and after any data incident, link health quarterly), adapting table and column names to my schema as references/checks.md describes. GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. Never ask me to paste secrets in chat: credentials go through the secure vault or environment variables. If anything looks off, stop and ask me.
Saved to your recent installs. Find it anytime on /connect.
🔒 Security, provenance and what this touches
The install prompt below already includes the vetting steps: your agent follows the community checklist before installing anything with executable code. Want more?
Verification certificate
Dated facts about what was checked for this listing, and how. This is not a safety certification.
- Pinned version
- aeaa0c8 · 2026-10-13
- Sandbox install test
- Sandbox retest pending · Oct 14, 2026
- Creator identity
- Verified
- Fingerprint watch
- Last checked · Nov 21, 2026 : no change detected in the files linked by this install prompt
- Install prompt analysis
- 2026-10-13 : No commands, network calls, packages, file writes or API keys detected.
Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify
More house skills
This skill is made by Skill Harbor itself. See all house skills →
Questions
How do I install a skill?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the skill up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a skill before installing it.