6-test-generator

Generate runtime validation test harnesses for confirmed zeroize-audit findings: C test harnesses, MemorySanitizer tests, Valgrind targets, and stack canary tests.

Input

You receive these values from the orchestrator:

Process

Step 0 — Load Configuration

Read config_path to load the merged config.

Step 1 — Read Final Report

Load {workdir}/report/findings.json and filter to confirmed findings (confidence = confirmed or likely).

Step 2 — Generate Test Harnesses

For each confirmed finding, generate:

1. C test harness: Allocates the sensitive object, calls the function under test, and verifies all bytes are zero at the expected wipe point.
2. MemorySanitizer test (-fsanitize=memory): Detects reads of un-zeroed memory after the wipe point.
3. Valgrind invocation target: Builds the test without sanitizers for Valgrind leak and memory error detection.
4. Stack canary test: For STACK_RETENTION findings, places canary values around the sensitive object and checks for retention after function return.

Step 3 — Generate Makefile

Produce a Makefile in the output directory that:

• Builds all test harnesses with appropriate compiler flags
• Includes sanitizer targets (test-msan, test-asan)
• Includes Valgrind targets (test-valgrind)
• Has a run-all target that executes everything and reports results
• Uses compile flags from compile_commands.json where applicable

Step 4 — Generate Manifest

Produce test_manifest.json listing all generated tests with:

• Test file path
• Finding ID it validates
• Test type (harness, msan, valgrind, canary)
• Expected behavior

Output

Write all output files to {workdir}/tests/:

Error Handling

• No confirmed findings: Write empty manifest and note in notes.md. Not an error.
• Missing final report: Fatal — cannot generate tests. Write error to notes.md.
• Always write test_manifest.json and Makefile — even if empty/no-op.