CVE-2026-46439

CVE-2026-46439 in compliance-trestle is being framed as an SSTI bug, but that mischaracterizes the actual risk. The vulnerability isn't about untrusted template authorship — it's about recursive output re-processing. When compliance authors write `{{ ssp.metadata.title }}` in a template, they expect string interpolation. What they get is a second Jinja2 render pass over the already-processed output. A title field that should be inert data becomes executable code if an attacker controls any upstream data source feeding the pipeline. The critical detail most defenders are missing: this isn't a one-pass data-to-execution model. Traditional SSTI defenses assume you either trust the template or you sanitize the input. Compliance-trestle breaks that binary by feeding rendered output back into the template engine. The compliance officer writing the template has no mental model for "output that becomes code in a second pass" — and they shouldn't need one. This is an architectural flaw, not a user error. Your priority actions: First, verify whether your automated compliance pipelines pull data from external sources (vendor feeds, shared libraries, APIs) that could carry injection payloads in metadata fields. If so, treat any data ingestion point as a potential attack vector, not just the template itself. Second, examine whether Lookup Tables and SSP documents are the only re-processed outputs — the recursive rendering architecture may apply to other author commands you rely on. Third, the patch (3.12.3/4.0.3) must break this loop, but you need to determine whether it prevents recursive re-rendering entirely or adds sanitization that could be bypassed. The former eliminates the attack class; the latter defers it. The blast radius here is the real problem. Compromised compliance documents don't sit idle — they feed into FedRAMP packages, NIST assessments, and audit trails that third parties consume without re-verification. You may not just have a code-execution problem; you may have corrupted compliance evidence that downstream reviewers trust. Runtime monitoring of compliance pipeline outputs becomes necessary, not optional, because version-patching alone doesn't fix an architecture that can re-introduce this pattern with the next flexibility feature.

Reviewed through automated stages and approved by a human before publication.