security-secret-boundary-testinglisted

# Secret Boundary Testing Test which component may see each secret and which observations must exclude it. Use generated fixture secrets that cannot authenticate to a real service. Never use real credentials merely to make a leakage test realistic. ## Define The Observation Boundary Read the custody and telemetry contract. Separate authorized storage/transport of a secret from accidental exposure. Ciphertext, identifiers, lengths and timing can also be restricted by a privacy contract; do not assume only plaintext matters. Inventory channels before running the scenario: ordinary and pretty formatting, nested errors/results, stdout/stderr, child-process diagnostics, logs/traces, temporary files, IPC, crash output and CI artifacts. Identify who owns collection of each channel. A caller-selected subset cannot prove complete capture. Keep fixture source, expected-value buffers and authorized vault data separate from forbidden observations. Their intentional canaries are not leakage findings. ## Build The Regression 1. Generate distinct synthetic values for each secret class and owner. Exercise relevant encodings and binary values, including empty and non-UTF8 input where supported. Never print the secret to explain an assertion failure. 2. Drive the real producer, formatter and boundary under test, including refusal, cancellation and restart paths. A test of a redaction helper alone does not cover wrappers that bypass it. 3. Capture bounded output from all decl