Platform & taxonomy updates.
What changed on the platform and in the frameworks we support, newest first. Framework work and platform work are tagged separately.
How to read this. Entries are listed newest first. Each one records what changed and when it took effect. When a taxonomy or schema is superseded, the earlier version is kept on record, so filings already produced stay reproducible against the version they were built for.
-
Global TIN format validator, and a faster conversion tier
The Global TIN Format Validator is now public. It checks taxpayer identification numbers against per-jurisdiction structural rules — length, character class, checksum where one is defined — before a file ever reaches a tax authority portal. It runs as a stateless service with no submitted value written to disk, and it is edge-cached and rate limited so it can be used freely without an account.
Behind it, the conversion tier moved to pooled database connections, which removed the connection storms we were seeing when several large CRS batches were submitted in the same minute. Median conversion time for files above 50,000 account records improved noticeably; the tail improved more than the median, which is the part that actually hurt during filing season.
- TIN structural validation across the supported jurisdiction set
- Connection pooling on the conversion tier
- Reduced tail latency on large CRS and FATCA batches
-
MiCA supervisory reporting
Asset-referenced token and e-money token supervisory templates are now mapped end to end, covering the EBA MiCA reporting requirements for issuers. Reports are validated before export rather than at submission, so structural and cross-field problems surface while they are still cheap to fix.
On the engineering side, multi-step report generation was moved onto an orchestrated workflow. A failed step now retries in isolation instead of failing the whole run, which matters most for issuers producing several templates from one reference dataset.
- ART and EMT supervisory templates
- Pre-export validation with field-level messages
- Step-level retry on multi-template runs
-
Validation engine rewrite
The rule evaluator was rebuilt to evaluate independent rules in parallel rather than sequentially. For COREP submissions running the full rule set, this is the difference between waiting for a validation pass and getting one back while you are still looking at the screen.
Compute for the validation workers moved to ARM-based instances, which cut both cold-start time and cost per run. We also fixed a defect where a validation report could list the same cross-template error twice when a value participated in more than one rule — the underlying check was correct, but the duplicate entries made reports harder to read than they needed to be.
- Parallel rule evaluation
- ARM-based validation workers
- Fixed: duplicate cross-template errors in validation reports
-
MDES 3.4 portal support
A number of the jurisdictions we cover file through the MDES portal rather than accepting OECD CRS XML directly. Support for the 3.4 release covers the FATCA–CRS (FC) XML format alongside CRS XML, together with the portal's own file naming rules and its distinction between new, correction and NIL reports.
This removes a manual step for filers in the Caribbean and several smaller jurisdictions, where the common failure was a technically valid CRS file rejected on a portal-specific naming or message-type rule rather than on its contents.
- FC XML format alongside OECD CRS XML
- Portal file naming and message reference construction
- New, correction and NIL report types
-
xBRL-CSV output across the EBA reporting set
xBRL-CSV is now the default output format for EBA reporting, generated and validated against taxonomy 4.2 under DPM 2.0. For institutions moving off XBRL-XML, the practical gain is file size: xBRL-CSV report packages are dramatically smaller for the same submission, which removes the upload timeouts that large FINREP filers had been working around.
Both formats remain available. Reports produced against earlier taxonomy versions stay reproducible against the version they were built for, so a resubmission of a prior period does not silently move to a newer taxonomy.
- xBRL-CSV report packages across the EBA set
- Validation against taxonomy 4.2 / DPM 2.0
- Earlier taxonomy versions retained for reproducibility
-
DORA Register of Information
The Register of Information is now produced as a complete, validated submission rather than assembled by hand across spreadsheets. Contractual arrangements, ICT third-party providers, functions and the links between them are held as structured records, so the relational integrity checks that reject a register at submission are run before you export it.
The most common rejection we saw in testing was a referenced entity identifier that did not resolve within the register itself. That check now runs continuously as records are edited, not only at export.
- Full Register of Information export
- Continuous referential integrity checks
- LEI validation on third-party provider records
-
Filing season hardening
January is the month the platform is under most load, so the work went into durability rather than features. Submitted files and generated reports are encrypted at rest with customer-scoped keys, database storage runs across multiple availability zones with automated failover, and retention lifecycle rules now expire working artefacts on a defined schedule while keeping submitted reports for their required retention period.
We also fixed a defect in CRS batch handling where an account record carrying more than one controlling person could, in a specific ordering, produce a report that passed schema validation but grouped controlling persons under the wrong account holder. Affected customers were contacted directly and corrected files were reissued.
- Encryption at rest with customer-scoped keys
- Multi-AZ storage with automated failover
- Fixed: controlling person grouping in multi-CP CRS account records
Have a question about a change?
Ask us how an update affects your filings, or see how the platform keeps frameworks current.