feat: owasp threat dragon threat model import support

README.md

@@ -72,6 +72,9 @@ open as-is; see [Formats & interoperability](docs/formats.md).
 - **Version like code**: semantic `diff`, three-way `merge`, and `git-setup` to wire both into
   your repo, plus declarative `apply`/`export` manifests for reproducible models.
 - **Convert** between `.tm7`, `tmforge-json`, draw.io, and Visio.
+- **Import OWASP Threat Dragon v2 JSON** with stable identities and authored threats. The initial
+  [bounded subset](docs/formats.md#threat-dragon-owasp-threat-dragon-v2-import) supports rectangular
+  boundaries and directed flows; native Threat Dragon export is not supported.
 - **Report** to self-contained HTML (with inline SVG diagrams), or `render` the diagram right
   in your terminal.
 - **Analyze in CI** with the `tmforge` CLI (`tmforge analyze`), gating builds on SARIF-reported

docs/cli-reference.md

@@ -887,6 +887,11 @@ tmforge report payments.tm7 --rules ./corporate.tmrules.json --out payments.html
 
 Convert between formats. The target is chosen by `--to` or inferred from the `--out` extension.
 
+OWASP Threat Dragon v2 JSON is an additional **input-only** format, detected from its content.
+Use `tmforge convert dragon.json --to tmforge-json --out imported.tmforge.json` or `--to tm7`.
+The initial [supported subset and refusal rules](formats.md#threat-dragon-owasp-threat-dragon-v2-import)
+are deliberate: unsupported trust-boundary geometry and threat treatments are not silently changed.
+
 ```text
 tmforge convert [--to <format>] [--out <path>] [--knowledge-base <file.tb7>] [--json] <input>
 ```

docs/formats.md

@@ -12,6 +12,7 @@ consume them, and each provider declares how faithfully it round-trips.
 | `tmforge-json` | `.tmforge.json` | Threat Model Forge JSON (canvas wire model) | Yes | Yes | Structural |
 | `drawio` | `.drawio` | draw.io / diagrams.net | Yes | Yes | Structural |
 | `vsdx` | `.vsdx` | Microsoft Visio | Yes | Yes | Structural |
+| `threat-dragon` | `.json` (content-detected) | OWASP Threat Dragon v2 | Yes, bounded subset | No | Import only |
 
 List these at runtime with `tmforge` (via conversion targets) or the API's `GET /v1/formats`, which
 returns each provider's capabilities and fidelity note.
@@ -57,8 +58,11 @@ validated the same as one authored with custom attributes.
 The shape Studio and the API speak: elements, flows, trust boundaries, names, and geometry, plus an
 optional analysis selection (disabled packs/rules) and an author-owned threat overlay, risk
 acceptance, and per-threat edits (state, priority, mitigation, description), plus manually-authored
-threats. Multi-page models carry a `diagrams` array (one entry per page, with its name); single-page
-models keep the flat `elements`/`flows` shape, so older readers keep working. Knowledge-base
+threats. An optional `metadata` object retains the model name, owner, description, contributors, and
+reviewer. Imported manual threats carry informational provenance in their optional `source` object.
+Multi-page models carry a `diagrams` array (one entry per page, with its name); a named or explicitly
+identified single page also retains that array. The flat `elements`/`flows` mirror the first page for
+older readers, and legacy flat documents remain readable. Knowledge-base
 attributes and the full generated-threat register are not represented, the register is regenerated
 from the rules on demand, and only the author-owned overlay round-trips so it's structural rather
 than lossless. Use it to bridge Studio and the CLI.
@@ -78,6 +82,58 @@ per-shape **Visio Shape Data** (visible in Visio's Shape Data pane) and re-impor
 properties. The rich threat model itself is not reconstructed, so the mapping is structural. Import
 recognizes packages this provider wrote and the documented master/shape convention.
 
+### `threat-dragon` (OWASP Threat Dragon v2 import)
+
+This is a **read-only, bounded import**, not native Threat Dragon editing or a lossless conversion.
+The reader recognizes the `summary` / `detail.diagrams` JSON envelope and accepts major version 2.
+It does not claim the generic `.json` extension, execute foreign rules, or fetch external resources.
+An unsupported construct refuses the entire import rather than returning a partial diagram.
+
+**Preserved:** named pages and stable identities; actors, processes, stores and rectangular trust
+boundaries; directed, page-local flows; model title/owner/description/contributors/reviewer; and
+authored threats with title, description, category, mitigation, priority, supported state and scope.
+Both `data.threats` and the cell-level `threats` schema variant are accepted, but two nonempty threat
+arrays on the same cell are ambiguous and refused.
+
+Imported threats use `manual:threat-dragon.<source-id>` keys and remain independent of generated
+tmforge threats. Source format/version, threat/cell/diagram ids, methodology, original status,
+severity and custom score are retained as provenance. Non-STRIDE category text is preserved, not
+coerced into STRIDE; importing a LINDDUN category does not install a LINDDUN evaluator. Source ids
+must fit the manual-id vocabulary; malformed or duplicate identities are refused.
+
+**Control mapping:** explicit store booleans map `storesCredentials` to `StoresCredentials`, `isALog`
+to `StoresLogData`, `isSigned` to `Signed`, and `isEncrypted` to `Encrypted=At-rest` or `No`.
+Actor `providesAuthentication` maps to `AuthenticatesItself`; flow `protocol` is preserved verbatim.
+Missing values remain missing. Flow encryption alone does not invent a protocol such as TLS.
+Other cell data is preserved under `ThreatDragon.data.*`, alongside source identity properties.
+**Out-of-scope flags are retained as information, not applied as tmforge suppressions.** The imported
+model is analyzed under the selected tmforge rules, not Threat Dragon's detection semantics.
+
+**Refused in this first delivery:** curved/line trust boundaries, unknown cell kinds, bidirectional
+or detached flows, cross-page endpoints, fractional/out-of-range rectangles, and unmappable threat
+statuses or priorities. Supported statuses are `Open`, `Mitigated`, and `Accepted`; supported
+priorities are `Critical`, `High`, `Medium`, and `Low`. For example, `NA`, `Transferred`, `Avoided`
+and `Eliminated` are not silently rewritten to `Accepted`. Geometry must use integer coordinates
+within +/-1,000,000 and dimensions from 1 to 100,000; rounding could alter a trust claim.
+
+Limits are 8 MiB of UTF-8 JSON, depth 64, 128 diagrams, 10,000 cells, and 20,000 authored threats.
+Style, thumbnails, free-form routing/vertices and unlisted document fields are not retained.
+Keep the original file. In particular, the upstream v2 demo includes curved boundaries and is
+deliberately refused by this first delivery; replacing those boundaries requires a human review,
+not automatic conversion to rectangles.
+
+```bash
+tmforge open dragon.json
+tmforge convert dragon.json --to tmforge-json --out imported.tmforge.json
+tmforge convert dragon.json --to tm7 --out imported.tm7
+```
+
+Studio **Open File**, the HTTP read endpoint, WASM and MCP use the same reader. Studio detaches the
+source file handle and proposes a new `.tmforge.json` name for Save. Native `--to threat-dragon`
+output is refused without overwriting the destination. Schema/sample grounding uses
+[Threat Dragon v2.6.2](https://github.com/OWASP/threat-dragon/tree/v2.6.2/ThreatDragonModels); tests use
+a synthetic fixture covering the supported subset, not a claim that every v2 model is importable.
+
 ## Converting
 
 ### CLI
@@ -99,7 +155,7 @@ See the [CLI reference](cli-reference.md#convert).
 
 ### API
 
-```
+```http
 POST /v1/model/convert?to=<format>     # convert to any format
 POST /v1/model/export/tm7              # export a .tm7 specifically
 POST /v1/detect                        # sniff a file's format from its bytes
@@ -109,9 +165,10 @@ See the [API reference](api-reference.md).
 
 ### Studio
 
-Studio round-trips through `tmforge-json` (**Export tmforge-json** / **Import JSON**). For `.tm7`,
-`.drawio`, or `.vsdx`, convert with the CLI or the API. The browser canvas itself doesn't parse file
-formats. See the [Studio guide](studio-guide.md#importing-and-exporting).
+Studio round-trips through `tmforge-json`. **Open File** uses the active API or in-browser WASM
+engine to read `.tm7`, `.drawio`, `.vsdx`, and supported Threat Dragon JSON. Read-only inputs save
+as new tmforge files, not back to their original format. The browser canvas itself does not parse
+foreign file formats. See the [Studio guide](studio-guide.md#importing-and-exporting).
 
 ## Choosing a format
 

docs/studio-guide.md

@@ -283,6 +283,23 @@ it. To change the model, edit the manifest and re-apply, or save the model as it
 > `tmforge apply`, but Studio cannot recognize it: every manifest field is optional, so a recognizer
 > that accepted an absent envelope would claim any JSON file. Add the envelope to open it here.
 
+### Opening Threat Dragon v2 JSON
+
+**Open File** recognizes supported OWASP Threat Dragon v2 JSON through the active engine. Imported
+threats appear as manual entries alongside tmforge-generated threats after **Analyze**, retaining
+their original category, text, treatment and scope. Model metadata and threat provenance survive
+saving, reopening and exporting to `.tm7`.
+
+This is import-only: Studio does not bind the original file for overwriting. **Save** offers a new
+`.tmforge.json` filename, and Threat Dragon is not offered as an export target. Keep the original
+for Threat Dragon-specific content such as styling and routing.
+
+The first delivery supports rectangular boundaries and directed flows. Curved boundaries,
+bidirectional flows, fractional rectangles and unsupported treatment states are refused, with no
+partial replacement of the current workspace. Out-of-scope flags remain source information and do
+not disable tmforge rules. See the [full import contract](formats.md#threat-dragon-owasp-threat-dragon-v2-import)
+before migrating a model.
+
 ## Merging edits from two branches
 
 When two people edit the same model on different branches, click **Merge** in the toolbar to

src/ThreatModelForge.Api/InputErrorHandler.cs

@@ -1,6 +1,7 @@
 namespace ThreatModelForge.Api
 {
     using System;
+    using System.IO;
     using System.Text.Json;
     using System.Threading;
     using System.Threading.Tasks;
@@ -81,6 +82,7 @@ private static bool IsCallerInput(Exception exception)
 
                 // Uploaded bytes that are not the model document they claim to be.
                 case JsonException:
+                case InvalidDataException:
                     return true;
                 default:
                     return false;

src/ThreatModelForge.Api/openapi/v1.json

@@ -1244,6 +1244,61 @@
           }
         }
       },
+      "MetaInformation": {
+        "type": "object",
+        "properties": {
+          "assumptions": {
+            "type": [
+              "null",
+              "string"
+            ],
+            "description": "Gets or sets the recorded assumptions."
+          },
+          "contributors": {
+            "type": [
+              "null",
+              "string"
+            ],
+            "description": "Gets or sets the contributors."
+          },
+          "externalDependencies": {
+            "type": [
+              "null",
+              "string"
+            ],
+            "description": "Gets or sets the external dependencies."
+          },
+          "highLevelSystemDescription": {
+            "type": [
+              "null",
+              "string"
+            ],
+            "description": "Gets or sets the high-level system description."
+          },
+          "owner": {
+            "type": [
+              "null",
+              "string"
+            ],
+            "description": "Gets or sets the model owner."
+          },
+          "reviewer": {
+            "type": [
+              "null",
+              "string"
+            ],
+            "description": "Gets or sets the reviewer."
+          },
+          "threatModelName": {
+            "type": [
+              "null",
+              "string"
+            ],
+            "description": "Gets or sets the threat model name."
+          }
+        },
+        "description": "Free-text metadata describing the modeled system (owner, reviewer, assumptions, and so on)."
+      },
       "PackDto": {
         "type": "object",
         "properties": {
@@ -1511,6 +1566,15 @@
               "string"
             ]
           },
+          "source": {
+            "type": [
+              "null",
+              "object"
+            ],
+            "additionalProperties": {
+              "type": "string"
+            }
+          },
           "manual": {
             "type": "boolean"
           }
@@ -1656,6 +1720,15 @@
               "string"
             ]
           },
+          "source": {
+            "type": [
+              "null",
+              "object"
+            ],
+            "additionalProperties": {
+              "type": "string"
+            }
+          },
           "priority": {
             "type": [
               "null",
@@ -1832,6 +1905,16 @@
               "string"
             ]
           },
+          "metadata": {
+            "oneOf": [
+              {
+                "type": "null"
+              },
+              {
+                "$ref": "#/components/schemas/MetaInformation"
+              }
+            ]
+          },
           "elements": {
             "type": [
               "null",

src/ThreatModelForge.Engine/Authoring/AuthoringService.cs

@@ -324,6 +324,7 @@ public static AuthoringResultDto EditThreat(TmForgeModelDto? model, EditThreatRe
                 Priority = priority ?? existing.Priority,
                 Description = request.Description ?? existing.Description,
                 Mitigation = request.Mitigation ?? existing.Mitigation,
+                Source = existing.Source,
                 Category = request.Category ?? existing.Category,
                 Title = Resolve(request.Title, existing.Title),
                 ElementIds = existing.ElementIds,
@@ -393,6 +394,7 @@ private static TmForgeModelDto WithThreats(TmForgeModelDto source, List<ThreatSt
             {
                 Schema = source.Schema,
                 Version = source.Version,
+                Metadata = source.Metadata,
                 Elements = source.Elements,
                 Flows = source.Flows,
                 Diagrams = source.Diagrams,

src/ThreatModelForge.Engine/EngineService.cs

@@ -1340,6 +1340,7 @@ private static void AppendManualThreats(
                     Interaction = DescribeScope(ids, idToName),
                     State = NormalizeState(entry.State),
                     Justification = entry.Justification,
+                    Source = entry.Source,
                     Manual = true,
                 });
             }

src/ThreatModelForge.Engine/ThreatDto.cs

@@ -59,6 +59,9 @@ public sealed class ThreatDto
         /// <summary>Gets the author-set description, when present.</summary>
         public string? Description { get; init; }
 
+        /// <summary>Gets informational source provenance when this is an imported threat.</summary>
+        public IReadOnlyDictionary<string, string>? Source { get; init; }
+
         /// <summary>Gets a value indicating whether this threat was authored manually (not projected from a rule).</summary>
         public bool Manual { get; init; }
     }

src/ThreatModelForge.Engine/ThreatStateDto.cs

@@ -35,6 +35,9 @@ public sealed class ThreatStateDto
         /// <summary>Gets the author-set mitigation override.</summary>
         public string? Mitigation { get; init; }
 
+        /// <summary>Gets informational source provenance for an imported manual threat, never rule input.</summary>
+        public IReadOnlyDictionary<string, string>? Source { get; init; }
+
         /// <summary>Gets the author-set priority (<c>High</c> / <c>Medium</c> / <c>Low</c>).</summary>
         public string? Priority { get; init; }
 

src/ThreatModelForge.Engine/TmForgeModelDto.cs

@@ -1,6 +1,7 @@
 namespace ThreatModelForge.Engine
 {
     using System.Collections.Generic;
+    using ThreatModelForge.Model;
 
     /// <summary>
     /// The canonical tmforge-json model posted by the editor.
@@ -13,6 +14,9 @@ public sealed class TmForgeModelDto
         /// <summary>Gets the schema version.</summary>
         public string? Version { get; init; }
 
+        /// <summary>Gets the author-owned model description, owner and review metadata.</summary>
+        public MetaInformation? Metadata { get; init; }
+
         /// <summary>Gets the DFD elements.</summary>
         public IReadOnlyList<TmForgeElementDto>? Elements { get; init; }