copycat handles sensitive clipboard contents (passwords, API tokens, private keys) in memory. Security disclosures are taken seriously.
Do not open public issues for security vulnerabilities.
Please use GitHub's private vulnerability reporting feature:
- Navigate to the Security tab of this repository.
- Click Report a vulnerability.
When submitting a report, please include:
- Repository version/commit SHA, OS, and launch configuration.
- Steps to reproduce or a test case.
- Impact assessment (e.g., memory exposure, DoS).
| Version | Supported |
|---|---|
main |
Yes |
| Latest release | Yes |
| Older releases | No |
- IPC Channel Security: Permissions on unix sockets (
0600/0700) and Windows named-pipe DACLs. - Masking Policy Enforcement: Preventing
Secretitems from reaching model outputs in any privacy posture. - Memory Zeroization & Logging: Ensuring credentials/secrets are zeroized on drop and never emitted in application logs.
- IPC Denial of Service: Panics or unbounded allocations reachable over the wire.
- Same-User Process Access: Processes running as the same user can access the socket by design.
- Heuristic Classifier Misses:
pii::classifyoperates on best-effort pattern matching. - Debug Builds: Unsafe reveal paths explicitly enabled via
--features debug-unsafeorCOPYCAT_INSECURE.
- Set
COPYCAT_PRIVACY=strictwhen using third-party/hosted model endpoints. - Store sockets only in secure, restricted directories (
0700). - Do not enable
debug-unsafefeatures in production builds.