Scan unpacked values, binary artifacts, and generated text for sensitive-looking values
Source:R/safety.R
scan_secrets.RdReads all artifacts that would be copied into a bundle (captured input RDS files,
original_result.rds, original_error.rds, hidden_state.rds, user-added RDS/RData
files, and textual/dotfile artifacts) from an unpack_call() result and reports
strings that match common secret-shaped patterns or values held under
credential-named fields. Unreadable or unsupported binary artifacts produce an
explicit unknown/unassessed outcome. Reported previews are redacted and never include
the full matched value. RDS and RData files are deserialized and recursively
inspected when their contents are supported. Raw vectors, functions,
environments, external pointers,
and objects exceeding the traversal limits (64 levels or 10,000 nodes) are
unassessed; environments are not evaluated to inspect their bindings. Other
artifacts up to 1 MiB are treated as UTF-8 text when their
initial bytes contain no NUL, while oversized, binary, or unreadable artifacts
are unknown. Shape-based detection uses secret_patterns(); credential-named
fields are also checked for values without a recognizable shape.
Arguments
- x
An
insider_unpackobject returned byunpack_call().