Guide · Agent skill
How to use the Server-Side Tracking Audit agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The ads-server-side-tracking skill gives an AI agent a structured checklist for auditing how a website or app sends conversion data to advertising platforms through server-side tagging, such as server-side Google Tag Manager (sGTM) or Meta and TikTok conversion APIs. This guide explains what the skill does, how to install it, how to ask for an audit, and what to prepare so the agent can work with your setup. By the end you will know how to run an audit that returns a prioritised list of findings with verification steps.
Server-Side Tracking AuditAgriciDaniel/claude-ads · MIT Download the skill Quick start
- 1Download the ZIP. Download the skill ZIP file from this directory page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ in a project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Gather the tracking materials. Collect container exports, sample event payloads, diagnostic logs and business conversion totals for the period you want to audit. Redact raw identifiers and credentials.
- 4Ask for a specific audit. Describe the platform, the symptom and the data you are sharing, using terms such as Meta CAPI, event_id or deduplication.
- 5Verify the findings. Work through the prioritised findings and their verification steps yourself, because the agent reports issues rather than fixing them.
What does the Server-Side Tracking Audit skill make the agent do?
An agent skill is a set of instructions that an AI agent loads when a request matches the skill's description. This one teaches the agent to review measurement from collection to destination. It maps the collection, consent, transport, transformation, destination, storage and observability components of your setup.
The agent then compares browser and server events, checking taxonomy, parameters, IDs, timestamps, values, currency, user data and consent state. It verifies deduplication, replay handling, retries, latency, diagnostics and test events, and checks hashing and data minimisation before data is transmitted. Finally it reconciles destination diagnostics with source logs and business conversions.
- Output: schema-valid findings, failure modes, an owner and a priority for each issue, and verification steps.
- Hashing does not make unnecessary data collection acceptable, so the agent also checks whether data should be sent at all.
- The skill instructs the agent not to change production tracking during an audit.
How do you install the skill?
Download the ZIP file from this directory page. Unzip it into one of two locations, depending on how you want to use the skill. For a personal setup, unzip it into ~/.claude/skills/. For a single project, unzip it into the .claude/skills/ folder inside that project.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills. The skill's package contains a SKILL.md file, and the agent must be SKILL.md-compatible, such as Claude Code.
- Personal use: ~/.claude/skills/
- Project use: .claude/skills/ inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload
How do you trigger the skill with a natural request?
You do not need a special command. Describe the problem in plain language and include terms the skill's description covers. The description mentions server-side tracking, sGTM, server-side tagging, Meta CAPI, TikTok Events API, event_id, deduplication, pixel debugging, first-party measurement and conversion data loss.
Be specific about the platform and the symptom. A request that names the platform, the suspected problem and the data you can share will produce a more useful audit than a general question.
- Ask for a review of Meta CAPI deduplication when Meta reports more conversions than your backend records.
- Ask for a check of event_id matching between the browser pixel and the server events.
- Ask for a sGTM container audit before a site migration.
- Ask for a pixel debugging review when TikTok conversions drop after a consent banner change.
What do you need to give the agent before it can audit?
The skill requires access to the tracking setup being audited. In practice this means material the agent can read, such as container exports, payload samples or diagnostic logs. The agent treats these as untrusted data and does not retain raw identifiers or credentials in its artifacts, so you can redact personal data before sharing.
Business conversion figures help the agent reconcile destination diagnostics with what actually happened on your site. Without them, it can check consistency between tracking systems but cannot confirm that the numbers match real sales or leads.
- Container exports or configuration from server-side tag management
- Sample event payloads from the browser and the server
- Diagnostic or server logs covering the period in question
- Business conversion totals from your own system, for reconciliation
What safety limits should you know about?
The skill runs no scripts and uses no network access, according to its listing. The agent works from the material you supply. It is designed to audit and report rather than to change production systems, and it is instructed not to modify production tracking during an audit.
Debug pages, tag payloads, logs, exports and vendor responses are treated as untrusted data, so the agent should not follow instructions embedded in them. Still, review what you share, because the audit depends on the samples you provide.
- No scripts are run and no network calls are made by the skill.
- The agent should not change production tracking during an audit.
- Remove raw identifiers and credentials from any files before sharing them.
What tips help you get a useful audit?
Start with one platform and one symptom, such as lost Meta conversions or doubled Google Ads conversions. Narrow audits produce findings you can act on. Then use the verification steps in the output to confirm each finding yourself, since the agent reports issues but does not fix them.
The skill is not useful for simple browser-only pixel setups with no server component, where its checklist adds little value. Its output is a technical review. Consent and privacy decisions need review by counsel, not by the agent.
- Use the priority ordering to decide which findings to fix first.
- Check each finding with the verification steps before making changes.
- Do not rely on the audit alone for legal compliance decisions on consent or privacy.
What people use it for
- Finding why Meta reports more conversions than your backend records, by checking deduplication between browser and server events.
- Checking that event_id values match between the browser pixel and the server-side conversion API for TikTok or Meta.
- Reviewing an sGTM container before a site migration to catch taxonomy or consent problems.
- Reviewing what personal data leaves your site through conversion APIs, as part of a privacy review.
Frequently asked questions
How do I install the Server-Side Tracking Audit skill?
Download the ZIP from this directory page and unzip it into ~/.claude/skills/ for personal use, or into .claude/skills/ inside a project. In the Claude apps you can instead upload it under Settings → Capabilities → Skills.
What do I need to run the ads-server-side-tracking skill?
You need access to the tracking setup being audited, such as container exports, payload samples or diagnostic logs. You also need a SKILL.md-compatible agent, for example Claude Code.
Is the ads-server-side-tracking skill free to use?
The skill is licensed under the MIT licence, as listed on this page. Check the licence file in the source repository for the full terms.
Does the skill fix my tracking automatically?
No. The skill is designed to audit and report. It returns prioritised findings with verification steps and is instructed not to change production tracking during an audit.
Why does the agent say my setup is out of scope?
The skill is not designed for simple browser-only pixel setups with no server component, where its checklist adds little value. It is aimed at setups that send conversion data through server-side tagging or conversion APIs.
Does the skill run scripts or send data over the network?
According to its listing, the skill runs no scripts and uses no network access. The agent works from the files and samples you provide.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits marketing and analytics teams troubleshooting lost or doubled conversions in Meta, TikTok or Google Ads, and developers maintaining a server-side GTM container or Conversions API integration. It is a poor fit for browser-only pixel setups, for anyone expecting automatic fixes, or for legal consent and privacy decisions, which should be checked by counsel.
Download the skill