Sanitize Server Logs for ChatGPT & AI
Secrets + JWT + IPSanitize server logs for ChatGPT and AI—stripping API keys, JWTs, Postgres/Mongo URIs, MACs, IPv4/IPv6 addresses, and .internal hosts in browser RAM.
- 0 bytes uploaded — verify offline
- API keys, JWT & DB URIs
- Ignores v1.2.3.4 versions
Instant Before & After Preview
API keys, JWT, DB URIs & IPs scrubbed[ERROR] build=v2.4.1.0 client=192.168.10.42 upstream=http://auth.corp.internal:8080/v1/token
db=postgres://app_user:s3cr3tPass!@db.internal:5432/billing
Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiJ1c3JfMTIzIiwibmFtZSI6Ik1heWEgTGluIiwibG9jYWwiOiJ0cnVlIn0.dGVzdF9zaWduYXR1cmU
openai_key=sk-proj-demoKey1234567890abcdef[ERROR] build=v2.4.1.0 client=[[IP-01]] upstream=[[URL-01]]
db=[[DB-01]]
Authorization: Bearer [[JWT-01]]
openai_key=[[KEY-01]]How it works
How to scrub secrets from server logs before asking AI
Debug production stack traces with LLMs without leaking database passwords, cloud keys, or internal VPC topology.
- Paste stack trace or server logDrop .log files or paste request logs containing Authorization headers, JWTs, DB connection strings, or private IPs.
- Review scrubbed tokensCredentials become [[KEY-01]], [[JWT-01]], [[DB-01]], and [[IP-01]] while semantic versions like v1.2.3.4 stay untouched.
- Copy to AI & restore patchesShare the sanitized log with AI, then paste suggested config or fix commands into Restore to plug real hostnames back in locally.
FAQ
Frequently asked questions
Which developer secrets and token prefixes are detected?
AWS keys (AKIA/ASIA), OpenAI/Anthropic/Stripe keys (sk-, sk-ant-, sk-proj-), GitHub (ghp_, gho_, github_pat_), Slack (xox...), Google (AIza...), Bearer tokens, JWTs, and DB URIs.
Will package versions like v1.2.3.4 or 10.0.2.15-alpha be mistaken for IPs?
No. IPv4 validation checks strict 0–255 octets and ignores version prefixes like v1.2.3.4 so stack trace versions stay readable for AI debugging.
Are private internal domains like db.prod.internal masked too?
Yes. Hostnames and URLs ending in .internal, .local, .corp, localhost with ports, and RFC1918 private IPs are flagged and tokenized.