Privacy Policy
We take the security of your codebase seriously. This policy outlines how Janitor Agent handles your source code, production logs, and AI interactions.
1. Information We Collect
To provide autonomous error resolution, we collect and process the following data categories:
- Production LogsStack traces, error messages, and environment context sent via Webhook.
- Source Code SnippetsRead-only access to files relevant to the specific error being debugged.
- Git MetadataBranch names, commit hashes, and repository names to generate Pull Requests.
2. Your Responsibility (Secrets & PII)
Important Safety Notice: You are responsible for sanitizing your logs before they reach our webhook.
We do not want your secrets.
Janitor Agent is not designed to process Personally Identifiable Information (PII) or sensitive secrets (API Keys, Passwords). We disclaim all liability for any secrets you accidentally include in the stack traces or log payloads sent to our infrastructure.
3. How We Process Data
We use your data strictly to generate fixes. We do not sell your code to third parties.
The "Transient" Promise
Your source code and logs are processed transiently. They are sent to our AI inference provider (e.g., OpenAI/Anthropic) solely for the purpose of generating a fix and are discarded from the inference context window immediately after, subject to the provider's own retention policies.
4. AI Model Training
A common concern with AI coding tools is model training. Our stance is clear:
- Zero Training: We enforce "Zero Retention" policies via API agreements with our LLM partners. To the best of our ability, we ensure your private codebase is never used to train base models.
- Opt-In Improvements: Only if you explicitly tag a fix as "Helpful" in the dashboard, may we store the anonymized pattern to improve future agent performance.
5. Third-Party Subprocessors
We utilize the following infrastructure partners. We are not liable for data breaches occurring on third-party infrastructure.
| Provider | Purpose |
|---|---|
| OpenAI / Anthropic | LLM Inference |
| Vercel / AWS | Hosting & Log Ingestion |
| GitHub | Pull Request Management |
6. Security Measures
We employ industry-standard security practices to protect your intellectual property:
- Webhook Verification: We validate signatures on incoming logs (where supported).
- Encryption: Data is encrypted in transit (TLS 1.3).
- Scope limitation: The agent only requests permission to modify specific repositories you authorize.
To request data deletion or export, email privacy@janitor.help