@@ -2,10 +2,11 @@ name: datasets-validate-all
 
 # Runs data_validation/run_all_checks.py: every dataset's own validate.py
 # (including ones in sub-collections), then the shared checks in
-# data_validation/validators/ on every data file. Then the data_validation
-# test suite, and the cross-dataset consistency check. The coverage report
-# and anomaly list go to the job summary for reviewers; they never fail
-# the job.
+# data_validation/validators/ on every data file, and the cross-dataset
+# consistency check. The tests job runs the data_validation test suite on
+# each supported Python. The coverage report and anomaly list go to the job
+# summary for reviewers; they never fail the job.
+
 on:
   push:
     branches: [main]
@@ -43,9 +44,6 @@ jobs:
       - name: Cross-dataset consistency
         run: python data_validation/qc_tools/consistency_check.py
 
-      - name: data_validation tests
-        run: python -m pytest data_validation/tests -v
-
       - name: Coverage report and anomalies (job summary)
         if: always()
         run: |
@@ -58,3 +56,25 @@ jobs:
             python data_validation/qc_tools/anomaly_detector.py
             echo '```'
           } >> "$GITHUB_STEP_SUMMARY"
+
+  # The tooling's own tests on every Python release it supports.
+  tests:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ['3.11', '3.12', '3.13']
+    steps:
+      - name: Checkout
+        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install requirements
+        run: python -m pip install --quiet -r data_validation/requirements.txt
+
+      - name: data_validation tests
+        run: python -m pytest data_validation/tests -v
@@ -0,0 +1,69 @@
+name: reference-data-refresh
+
+# Monthly: regenerate the MITRE ATLAS and CWE tables in
+# data_validation/reference_data/ from MITRE's latest releases. If anything
+# changed, push the result to a branch and open an issue so a maintainer can
+# open the PR (this repository does not let workflows open pull requests).
+on:
+  schedule:
+    - cron: '17 6 2 * *'   # 06:17 UTC on the 2nd of each month
+  workflow_dispatch:
+
+permissions:
+  contents: write
+  issues: write
+
+jobs:
+  refresh:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: '3.11'
+
+      - name: Install requirements
+        run: python -m pip install --quiet -r data_validation/requirements.txt
+
+      - name: Regenerate ATLAS and CWE tables
+        run: python data_validation/reference_data/update_reference_data.py
+
+      - name: Run the checks against the new tables
+        id: checks
+        continue-on-error: true
+        run: python data_validation/run_all_checks.py --no-shared && python data_validation/validators/crossref_validator.py
+
+      - name: Push a branch and open an issue if anything changed
+        env:
+          GH_TOKEN: ${{ github.token }}
+          CHECKS: ${{ steps.checks.outcome }}
+        run: |
+          if git diff --quiet -- data_validation/reference_data; then
+            echo "Reference data is current."
+            exit 0
+          fi
+          branch="automation/reference-data-$(date -u +%Y-%m)"
+          git config user.name "github-actions[bot]"
+          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
+          git switch -c "$branch"
+          git add data_validation/reference_data
+          git commit -m "reference_data: refresh MITRE ATLAS and CWE tables"
+          git push --force origin "$branch"
+          versions=$(grep -E '^\| (MITRE ATLAS|CWE) ' data_validation/reference_data/SOURCES.md | cut -d'|' -f2,4)
+          {
+            echo "The monthly refresh regenerated the MITRE ATLAS and CWE tables."
+            echo
+            echo "Versions: $versions"
+            echo
+            echo "Identifier check against the existing data: **$CHECKS**. A failure usually means MITRE retired or renamed an ID the datasets cite; the crossref output in the run log names the entries."
+            echo
+            echo "Open the PR: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/compare/main...${branch}?expand=1"
+            echo
+            echo "Run: ${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"
+          } > body.md
+          if gh issue list --state open --search "in:title Reference data refresh $(date -u +%Y-%m)" --json number --jq 'length' | grep -q '^0$'; then
+            gh issue create --title "Reference data refresh $(date -u +%Y-%m)" --body-file body.md
+          fi
@@ -40,7 +40,8 @@ The tooling has two jobs: keep the research data behind OWASP deliverables (the
 |---|---|---|
 | `schema_validator.py` | Schema violations, for files no dataset validator covers | File with no schema to check against |
 | `dsgai_mapping_check.py` | DSGAI ID not in the taxonomy; a mapping value that is not a DSGAI ID | Same ID twice in one mapping; more than 6 mappings |
-| `crossref_validator.py` | Malformed CVE/GHSA/CWE/ATLAS ID; CVE year in the future; unknown CWE or ATLAS ID | Deprecated CWE; retired ATLAS ID; malformed ID in free text (often a placeholder) |
+| `crossref_validator.py` | Malformed CVE/GHSA/CWE/ATLAS ID; CVE year in the future; unknown CWE, ATLAS or OWASP Top 10 ID (`LLM01:2026`, `ASI01:2026`) | Deprecated CWE; retired ATLAS ID; OWASP ID without its edition year; malformed ID in free text (often a placeholder) |
+| `severity_check.py` | CVSS v3.x/v4.0 vector that doesn't parse; score that doesn't match its vector; qualitative rating outside the score's FIRST band | Score with no vector |
 | `anonymization_scanner.py` | — | Email addresses, keys and tokens, private or (in a network context) public IPs, internal hostnames, home paths, SSNs |
 | `dedup_checker.py` | Two records in one dataset with the same ID | Records with identical or ≥ 80% similar text |
 
@@ -75,7 +76,7 @@ After the automated checks pass, contributions are reviewed by cybersecurity and
 
 ## Running the checks
 
-Python 3.10 or higher.
+Python 3.11 or higher (CI tests 3.11, 3.12 and 3.13).
 
 ```bash
 pip install -r data_validation/requirements.txt
@@ -111,13 +112,13 @@ data_validation/
 ├── qc_tools/                Reports for reviewers
 ├── schemas/                 Schemas for datasets without their own schema.json
 │                            (the rest are pointers to the dataset-local schemas)
-├── reference_data/          DSGAI, MITRE ATLAS and CWE lookup tables; see SOURCES.md
+├── reference_data/          DSGAI, MITRE ATLAS, CWE and OWASP Top 10 lookup tables; see SOURCES.md
 └── tests/                   Tests and deliberately valid/invalid fixtures
 ```
 
 Dataset-local schemas (`datasets/<name>/schema.json`) are authoritative. `schemas/exploit`, `vulnerability`, `riskassessment` and `agentdataflow_trace` are kept only as `$ref` pointers to them; `schemas/incident`, `rag`, `crossframework_mapping` and `promptinj_testcase` are the only schemas for their data.
 
-The ATLAS and CWE tables are generated from MITRE's releases by `reference_data/update_reference_data.py`; don't edit them by hand.
+The ATLAS and CWE tables are generated from MITRE's releases by `reference_data/update_reference_data.py`; don't edit them by hand. The `reference-data-refresh` workflow runs it on the 2nd of each month and, when MITRE has published something new, pushes a branch and opens an issue for a maintainer to turn into a PR. `owasp_top10.csv` is maintained by hand when OWASP publishes a new edition.
 
 ---
 
 
@@ -17,17 +17,17 @@ The data validation pipeline is a set of Python scripts that check contributed d
 
 ## Prerequisites
 
-### Python 3.10+
+### Python 3.11+
 
-The scripts require Python 3.10 or higher. Most modern systems have Python pre-installed.
+The scripts require Python 3.11 or higher. Most modern systems have Python pre-installed.
 
 **Check if you have it:**
 
 ```bash
 python3 --version
 ```
 
-If you see `Python 3.10.x` or higher, you're good. If not:
+If you see `Python 3.11.x` or higher, you're good. If not:
 
 | Platform | How to install |
 |---|---|
@@ -95,7 +95,7 @@ You'll know it's active when your terminal prompt changes to show `(venv)` at th
 pip install -r requirements.txt
 ```
 
-> **What this does:** Reads the `requirements.txt` file and installs all the Python libraries the validation scripts need. That is `jsonschema` (schema validation) and `pytest` (the tests), plus `pyyaml`, which only the reference-data refresh script uses.
+> **What this does:** Reads the `requirements.txt` file and installs all the Python libraries the validation scripts need. That is `jsonschema` with its format checkers (schema validation, including dates and URIs) and `pytest` (the tests), plus `pyyaml`, which only the reference-data refresh script uses.
 
 **If you get a permissions error:** Make sure your virtual environment is activated (Step 2). If you're not using a virtual environment, add `--user` to the command: `pip install --user -r requirements.txt`
 
@@ -169,7 +169,7 @@ Python might be installed as `python` instead of `python3` on your system (commo
 python --version
 ```
 
-If that shows 3.10+, use `python` everywhere this guide says `python3`.
+If that shows 3.11+, use `python` everywhere this guide says `python3`.
 
 ### "No module named 'jsonschema'" (or any other module)
 
 
@@ -6,8 +6,11 @@ Lookup tables the validators check identifiers against.
 |---|---|---|---|
 | DSGAI | `dsgai_entries.json` | 1.0 | OWASP GenAI Data Security Risks and Mitigations 2026 (kept identical to `datasets/_shared/dsgai_taxonomy.json`; a test enforces it) |
 | MITRE ATLAS | `mitre_atlas_techniques.csv` | 2026.09 | latest release listed in `https://raw.githubusercontent.com/mitre-atlas/atlas-data/main/dist/manifest.yaml` |
+| OWASP Top 10s | `owasp_top10.csv` | LLM 2025, LLM 2026, Agentic 2026 | LLM: `https://github.com/GenAI-Security-Project/GenAI-LLM-Top10` (2026 published 2026-08-04). Agentic: `https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/` (published 2025-12-09) |
 | CWE | `cwe_ids.csv` | 4.20 | `https://cwe.mitre.org/data/xml/cwec_latest.xml.zip` (weaknesses, categories and views) |
 
+`owasp_top10.csv` is maintained by hand. The 2026 LLM list renumbered the entries (LLM03 was Supply Chain in 2025 and is Excessive Agency in 2026), so IDs always carry the edition year; `equivalent_2025` gives the 2025 ID each 2026 entry continues (LLM08:2026 Hidden Context Exposure broadens LLM07:2025 System Prompt Leakage). OWASP writes the Agentic IDs without a year (ASI01); this repository adds `:2026` so a future edition cannot collide.
+
 Regenerate the ATLAS and CWE tables with `python update_reference_data.py`; it fills in the Version column. Do not edit those two CSVs by hand.
 
 `datasets/_shared/dsgai_taxonomy.json` is the DSGAI list the validators use. `dsgai_entries.json` is kept because the Turkish prompt-injection validator reads it.
 
@@ -0,0 +1,31 @@
+id,list,edition,name,equivalent_2025
+LLM01:2025,LLM,2025,Prompt Injection,
+LLM02:2025,LLM,2025,Sensitive Information Disclosure,
+LLM03:2025,LLM,2025,Supply Chain,
+LLM04:2025,LLM,2025,Data and Model Poisoning,
+LLM05:2025,LLM,2025,Improper Output Handling,
+LLM06:2025,LLM,2025,Excessive Agency,
+LLM07:2025,LLM,2025,System Prompt Leakage,
+LLM08:2025,LLM,2025,Vector and Embedding Weaknesses,
+LLM09:2025,LLM,2025,Misinformation,
+LLM10:2025,LLM,2025,Unbounded Consumption,
+LLM01:2026,LLM,2026,Prompt Injection,LLM01:2025
+LLM02:2026,LLM,2026,Sensitive Information Disclosure,LLM02:2025
+LLM03:2026,LLM,2026,Excessive Agency,LLM06:2025
+LLM04:2026,LLM,2026,Supply Chain,LLM03:2025
+LLM05:2026,LLM,2026,Data and Model Poisoning,LLM04:2025
+LLM06:2026,LLM,2026,Unbounded Consumption,LLM10:2025
+LLM07:2026,LLM,2026,Misinformation,LLM09:2025
+LLM08:2026,LLM,2026,Hidden Context Exposure,LLM07:2025
+LLM09:2026,LLM,2026,Vector and Embedding Weaknesses,LLM08:2025
+LLM10:2026,LLM,2026,Improper Output Handling,LLM05:2025
+ASI01:2026,ASI,2026,Agent Goal Hijack,
+ASI02:2026,ASI,2026,Tool Misuse and Exploitation,
+ASI03:2026,ASI,2026,Identity and Privilege Abuse,
+ASI04:2026,ASI,2026,Agentic Supply Chain Vulnerabilities,
+ASI05:2026,ASI,2026,Unexpected Code Execution,
+ASI06:2026,ASI,2026,Memory and Context Poisoning,
+ASI07:2026,ASI,2026,Insecure Inter-Agent Communication,
+ASI08:2026,ASI,2026,Cascading Failures,
+ASI09:2026,ASI,2026,Human-Agent Trust Exploitation,
+ASI10:2026,ASI,2026,Rogue Agents,
@@ -1,5 +1,7 @@
 # Validators, QC tools and their tests
-jsonschema>=4.18.0
+# [format-nongpl] adds the checkers for "uri" and other formats; without it
+# jsonschema treats those formats as annotations and never fails them.
+jsonschema[format-nongpl]>=4.18.0
 pytest>=7.0.0
 # Only for reference_data/update_reference_data.py (refreshing the ATLAS table)
 pyyaml>=6.0
@@ -9,7 +9,8 @@
    silently passing.
 2. Shared checks (data_validation/validators/) on every data file:
    schema (for files no dataset validator covers), DSGAI IDs, CVE/CWE/ATLAS
-   IDs, anonymization patterns and duplicates. Errors fail the run;
+   and OWASP Top 10 IDs, CVSS scores and ratings, anonymization patterns
+   and duplicates. Errors fail the run;
    heuristic warnings (anonymization, near-duplicates, retired IDs) are
    counted, and printed with --verbose.
 
@@ -34,6 +35,7 @@
 
 from validators import (  # noqa: E402
     anonymization_scanner, crossref_validator, dedup_checker, dsgai_mapping_check, schema_validator,
+    severity_check,
 )
 from validators._common import (  # noqa: E402
     ERROR, SKIP_DIRS, Finding, ensure_utf8_stdout, iter_data_files, load_json, load_taxonomy_ids,
@@ -100,6 +102,7 @@ def shared_checks(dataset: Path, datasets_root: Path) -> tuple[int, list[Finding
             findings += schema_validator.validate_file(path, dataset_root=dataset)
         findings += dsgai_mapping_check.check_data(path, data, taxonomy)
         findings += crossref_validator.check_data(path, data)
+        findings += severity_check.check_data(path, data)
         findings += anonymization_scanner.check_data(path, data)
     records, load_errors = dedup_checker.load_records(files)
     findings += load_errors + dedup_checker.check_records(records)
@@ -162,7 +165,7 @@ def main(argv: list[str] | None = None) -> int:
 
     checked_files = errors = warnings = 0
     if not args.no_shared:
-        print("\nShared checks (schema, DSGAI IDs, CVE/CWE/ATLAS IDs, anonymization, duplicates):")
+        print("\nShared checks (schema, DSGAI, CVE/CWE/ATLAS/OWASP IDs, CVSS, anonymization, duplicates):")
         for dataset in datasets:
             n, findings = shared_checks(dataset, datasets_root)
             if not n:
 
@@ -1,5 +1,5 @@
 {
-  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
   "title": "Cross-Framework Control Mapping Entry",
   "type": "object",
   "required": ["dsgai_id", "dsgai_name", "framework", "control_id", "control_name", "relationship"],
 
@@ -1,5 +1,5 @@
 {
-  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
   "title": "Incident Dataset Entry",
   "type": "object",
   "required": ["incident_id", "date", "summary", "dsgai_mapping", "attack_vector", "data_types_affected", "impact"],
Original file line number	Diff line number	Diff line change
`@@ -1,5 +1,5 @@`
`1`	`1`	`{`
`2`		`- "$schema": "http://json-schema.org/draft-07/schema#",`
	`2`	`+ "$schema": "https://json-schema.org/draft/2020-12/schema",`
`3`	`3`	`"title": "Cross-Framework Control Mapping Entry",`
`4`	`4`	`"type": "object",`
`5`	`5`	`"required": ["dsgai_id", "dsgai_name", "framework", "control_id", "control_name", "relationship"],`