Research Evidence handling
IOC provenance is part of the finding.
An extracted indicator becomes more useful when the raw match, normalized value, source context, transformation, and caveats remain attached for later review.
Indicators of compromise are commonly moved from logs and notes into tickets, reports, scripts, and blocklists. During that movement it is easy to preserve only a cleaned value and lose the details that make it possible to understand what was actually observed.
A normalized value can be useful for matching. The raw form and source context are useful for review. Strong evidence handling preserves both.
Normalization changes the evidence.
Defanging and refanging are familiar examples. A report may intentionally alter a URL to reduce accidental clicks. A tool can normalize the value for structured output, but that transform should not erase the original representation or imply independent validation.
raw_match: hxxps://example[.]test/login?source=report
normalized_value: https://example.test/login?source=report
indicator_type: url
source_context: analyst-note.txt:14
transform: refanged
validation: syntax only; no enrichment performed
That small record carries more information than a standalone URL. It tells a reviewer why strings differ, where the match appeared, and where the process stopped.
Build a reviewable record.
The right model depends on the workflow, but a compact record often includes:
- the raw match as it appeared in input;
- a normalized value when a safe, documented transform applies;
- the indicator category and validation outcome;
- the source location or surrounding context available to the tool; and
- the limitations, including an absent enrichment step or unsupported input form.
Safe sharing is deliberate.
Defanged representations can make reports safer to copy and discuss, but they are not a security guarantee. Output should make the state clear—especially where someone could paste a value into another tool. A visible label is safer than silently modifying a value and assuming the next reader notices.
Keep the limits visible.
IOC Sanitizer is a public local Python utility for extracting, validating, normalizing, defanging, and refanging supported common indicators. It does not claim that an indicator is malicious, active, or relevant to a particular incident.
That boundary is not a weakness. It is what allows the output to remain deterministic and honest enough for someone else to inspect, reproduce, and build on.
Provenance is part of engineering quality.
The same pattern appears in incident cases and detection outputs: preserve input, transform it deliberately, explain the rule, and state where automation ends. This makes a result more transferable than a high-level score because it gives the next analyst material to assess.