Offline BeyondTrust Remote Support evidence checker for CVE-2026-102764. Remote Support version and configuration evidence. Standalone Python; JSON/CSV/HTML reports; no exploit payloads.
BeyondTrust Remote Support: internal-message origin validation / unauthorized file operations
CVE-2026-102764 · advisory BT26-05 · offline, patch-evidence-based triage
beyondtrust_remote_support_cve_2026_102764_check.py is a single-file, standard-library-only Python 3.10+ checker for CVE-2026-102764. It reads operator-supplied facts (reported version, deployment type, and whether patch BT26-05-RS is applied) and reports an evidence status. It makes no network requests, runs no application code, and never tests exploitability.
[!IMPORTANT] A version number alone never shows remediation. BeyondTrust fixes this issue with the patch
BT26-05-RS, applied to releases that remain at version numbers up to26.2.3. The checker therefore reportsfixed_version_reportedfor a self-hosted appliance only when you supply patch evidence, and even then it is a report, not a guarantee. This CVE is listed in CISA's Known Exploited Vulnerabilities catalog (added 2026-10-16); treatpotentially_affectedandunknownresults as work to do.
This repository covers one vulnerability only. Related vulnerabilities belong in separate repositories.
| Field | Scope |
|---|---|
| Exploit identifier | CVE-2026-102764 |
| Aliases for this same exploit | None recorded |
| Vendor advisory / patch | BT26-05 / BT26-05-RS |
| Rule date | 2026-10-24 |
| Product | BeyondTrust Remote Support. Privileged Remote Access (PRA) is out of scope: the vendor blog says PRA shares the affected code and shipped the patch as a precaution, but the advisory and CVE record list only Remote Support. PRA input is unknown. |
The checker is offline only. The advisory is version- and patch-based, so no passive network mode can answer it, and none is provided.
self_hosted, saas or unknown) and patch evidence for BT26-05-RS.python beyondtrust_remote_support_cve_2026_102764_check.py --version 25.1.2 --deployment self_hosted --patch-evidence not_applied
python beyondtrust_remote_support_cve_2026_102764_check.py --version 26.2.3 --deployment self_hosted --patch-evidence applied --format html --output report.html
python beyondtrust_remote_support_cve_2026_102764_check.py --inventory appliances.json --format csv --output results.csv --fail-on-finding
python beyondtrust_remote_support_cve_2026_102764_check.py --csv-list appliances.csv --format json
python beyondtrust_remote_support_cve_2026_102764_check.py --help
Version-only input is supported but can only ever yield potentially_affected or unknown for self-hosted systems.
| Option | Meaning |
|---|---|
--version V | Reported version for a single observation |
--deployment {self_hosted,saas,unknown} | Deployment of that observation (default unknown) |
--patch-evidence {applied,not_applied,unknown} | Operator evidence about BT26-05-RS (default unknown) |
--custom-build {false,true,unknown} | Vendor or custom build (default false); true forces unknown, and unknown blocks fixed_version_reported |
--inventory FILE | JSON inventory (schema_version 1); repeatable |
--csv-list FILE | CSV list with columns label,version,deployment,patch_evidence,custom_build,evidence_source; repeatable |
Inputs can be combined; each record is a separate observation. custom_build defaults to false when omitted, which asserts a stock build. It must be a JSON boolean or "unknown" (CSV: true, false, unknown or empty); numbers and null are rejected. Set "unknown" when you do not know, which prevents fixed_version_reported.
The full input schema, size limits, strict parsing rules and decision order are in docs/REFERENCE.md.
| Option | Meaning |
|---|---|
--format {json,csv,html} | Report format (default json) |
--output FILE | Write the report to a new file instead of stdout |
--force | Replace an existing --output file |
--fail-on-finding | Exit 1 when any result is potentially_affected |
--tool-version | Print the checker version |
| Code | Meaning |
|---|---|
| 0 | Completed. Findings do not fail the run unless --fail-on-finding is given. |
| 1 | --fail-on-finding and at least one potentially_affected result. |
| 2 | Operational or input error (bad file, schema or field, existing output without --force, no input). |
| 130 | Interrupted. Recovery before publication retains completed observations and marks the report incomplete. Published files remain intact and interrupted stdout is never retried; see the reference for details. |
| Status | Meaning |
|---|---|
potentially_affected | Self-hosted or unspecified deployment, release 26.2.3 or earlier, and no evidence that BT26-05-RS is applied. Never confirmed exploitability. Releases older than 24.3.1 must upgrade before the patch can be applied. |
fixed_version_reported | Release 24.3.1 to 26.2.3 with operator-reported BT26-05-RS applied, or a deployment the operator declares as SaaS with either no version or a clean in-range version (vendor advisory BT26-05: SaaS patched automatically, applied as of 2026-10-09; the vendor blog says the patch was released 2026-10-08). The SaaS decision rests on the operator's word alone; the checker cannot verify it. A report, never a guarantee. |
unknown | Missing, malformed, prerelease or qualified version; custom/vendor build; conflicting evidence; patch reported on a release older than 24.3.1; SaaS with a malformed, qualified or above-range version; version above 26.2.3 (no fixed release number is published in the sources); other products. |
version_status, patch evidence and exploitability are separate fields. exploitability is always not_tested. There is no unsupported_runtime status: the sources give no lifecycle statement.
A version above 26.2.3 is unknown, not safe, because the sources name no fixed release number. Operational failures (exit 2) are not assessment results.