Structured Logging
Turn print statements into searchable records with stable fields.
What it adds
Every log line becomes a structured event with a fixed field schema, a real severity level, and redaction applied before it leaves the process.
What your agent is told to do
5
What your agent is told to do
5-
1
Define the field schema once and write it down: timestamp, level, message, service, environment, release version, and the request or job context. Every log line carries the same field names, spelled the same way.
-
2
Use severity levels with an agreed meaning. If everything is 'info', the levels are decoration. Reserve error for things a human must look at, and warn for things that degraded but recovered.
-
3
Redact at the logging layer, not at the call sites. Maintain a deny list of field names — password, token, secret, authorization, card number, email where policy requires it — and truncate large payloads to a bounded size.
-
4
Do NOT put unbounded high-cardinality values in indexed fields. A user ID is fine; a full URL with query string, a raw SQL statement, or a UUID per line will make the log store slow and expensive.
-
5
Keep human-readable output in local development. Machine-readable JSON in production, pretty lines on a developer's terminal.
Edge cases it handles
6
Edge cases it handles
6- Log lines emitted before the logger is configured must not be lost or crash the boot — buffer them or fall back to plain output.
- Multi-line content like stack traces must stay in one event, not fragment into a line per frame.
- Redaction must survive nesting. A token buried three levels down inside a serialized object is still a token.
- An object that fails to serialize must not raise inside the logger and take down the request. Log the failure and move on.
- Very high log volume must not become the bottleneck. Sample noisy repeated events rather than dropping them silently.
- Third-party libraries and the web server write their own logs in their own format — decide whether to adapt them or accept two formats, but do not pretend the problem does not exist.
Definition of done
8
Definition of done
8- Every log event carries the same core fields with identical names.
- Severity levels are documented and used consistently.
- Secrets and personal data are redacted at the logging layer, including inside nested objects.
- No indexed field carries unbounded high-cardinality values.
- Development output is human-readable; production output is machine-parseable.
- A serialization failure inside the logger never breaks the request.
- The feature matches the existing design system.
- No existing functionality is broken.
Related features
AI Cost Budgets
AI Cost Budgets
Cap what AI features are allowed to spend before the bill arrives.
What it does
Monetary spending limits on AI work, scoped by workspace, feature, and time period, enforced before a run starts.
How it works
- 1 Find every place the app calls a model and route all of them through one accounting point that records estimated and actual spend against a scope. A budget that only covers the chat feature is not a budget.
- 2 Estimate the cost of a run from the size of its input before dispatching it, and refuse anything that would exceed the remaining budget on its own.
- 3 Reserve the estimate against the budget when the run starts, then reconcile to the real usage figures when it finishes, releasing whatever was over-reserved.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/ai-cost-budgets
Multi-Model Routing
Multi-Model Routing
Send each AI request to the right model using rules you can read and test.
What it does
A deterministic routing layer that picks a model per request from task type, context size, latency budget, and data sensitivity.
How it works
- 1 Express routing as explicit, ordered rules over inputs the app can measure: task type, estimated context size, latency budget, and the sensitivity classification of the data involved. A rule set that can be read line by line can be reviewed and tested.
- 2 Make routing deterministic. The same inputs must always produce the same route, so a bad output can be reproduced and a rule change can be evaluated. Randomised or load-based selection turns every incident into guesswork.
- 3 Classify data before routing and refuse to route restricted content to any destination not approved for it. This check is a hard block, not a preference, and it must run before the request is assembled.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/multi-model-routing
SEO Setup
SEO Setup
Make your app findable — titles, meta, Open Graph, sitemap, robots.
What it does
The baseline SEO and social-preview setup every public app should have, and most skip.
How it works
- 1 Give every public page a unique, descriptive title and meta description. Find the app's layout and add a mechanism for each page to set them.
- 2 Add Open Graph and Twitter Card tags so shared links render a preview instead of a bare URL.
- 3 Generate a sitemap.xml covering every public, indexable page, and a robots.txt pointing at it.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/seo-setup
How it works
-
1
Copy the link
Grab the Markdown instruction URL for this feature.
-
2
Give it to your AI
Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.
-
3
It inspects, then implements
Your agent reads your existing app first, then adds the feature to fit it.
Works with your stack
These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.
Need it tighter than that? Customize the feature and tell it exactly what you're running.