weavatrix 1.11.0 → 1.14.2

package/README.md

@@ -10,9 +10,13 @@
 **Give your coding agent repository evidence before it starts guessing.**
 
 Weavatrix is the native MCP product for repository intelligence. It gives
-Codex, Claude Code, and other coding agents 47 read-only operations over one
+Codex, Claude Code, and other coding agents 60 read-only operations over one
 revision-bound evidence graph: impact, architecture, APIs, Git history,
-duplicates, dead code, search, semantic links, and temporal memory.
+duplicates, dead code, search, semantic links, temporal memory, exported
+n8n workflows, and Dify YAML apps. Those parsers are domains on Weavatrix
+Core, not new MCP products. See
+[n8n](https://weavatrix.com/n8n), [Dify](https://weavatrix.com/dify), and
+[parsers](https://weavatrix.com/parsers).
 
 It does not answer from a larger grep or an invented confidence score. Every
 bounded result can carry the repository revision, file, line, extractor,
@@ -42,21 +46,44 @@ weavatrix mcp .
 ```toml
 [mcp_servers.weavatrix]
 command = "npx"
-args = ["-y", "weavatrix", "mcp", "."]
+args = ["-y", "weavatrix@1.14.2", "mcp", "."]
 ```
 
+Pass an absolute repository path when the Codex process cwd is not the project
+you intend to analyze.
+
 ### Claude Code
 
 ```sh
-claude mcp add weavatrix -- npx -y weavatrix mcp .
+claude mcp add weavatrix -- npx -y weavatrix@1.14.2 mcp .
+```
+
+### Cursor
+
+Prefer the plugin, or pin the open workspace explicitly and do not also enable
+the plugin for the same server name:
+
+```json
+{
+  "mcpServers": {
+    "weavatrix": {
+      "command": "npx",
+      "args": ["-y", "weavatrix@1.14.2", "mcp", "${workspaceFolder}"]
+    }
+  }
+}
 ```
 
 Profiles expose bounded views of the same engine:
 
 ```sh
-npx -y weavatrix mcp . --profile=all
-npx -y weavatrix mcp . --profile=code
-npx -y weavatrix mcp . --profile=seo
+npx -y weavatrix@1.14.2 mcp . --profile=all
+npx -y weavatrix@1.14.2 mcp . --profile=code
+npx -y weavatrix@1.14.2 mcp . --profile=seo
+npx -y weavatrix@1.14.2 mcp . --profile=n8n
+npx -y weavatrix@1.14.2 mcp . --profile=dify
+npx -y weavatrix@1.14.2 mcp . --profile=agent
+npx -y weavatrix@1.14.2 mcp . --profile=diagram
 ```
 
 The package contains native binaries for Windows x64/arm64, macOS x64/arm64,
@@ -94,6 +121,8 @@ Find every GraphQL, gRPC, Kafka, RabbitMQ, NATS, JMS, SQS, or SNS
 contract affected by this branch.
 Build the smallest source bundle needed to edit this symbol safely.
 Show me this file as it was two commits ago, without a checkout.
+If I change the email recipient in this exported n8n workflow, who reads it?
+If I change start_node.query in this Dify YAML, which nodes consume it?
 ```
 
 ## See it answer
@@ -124,10 +153,12 @@ answered:
 }
 ```
 
-Every answer carries that `repository_context` block. Add
-`"expected_repository": "weavatrix"` to any call and a server that was
-retargeted elsewhere fails loudly instead of answering about the wrong
-repository.
+Every answer carries that `repository_context` block. From **1.11.1**, each MCP
+process pins its launch root for the session: `open_repo` to another path is
+refused unless you started with `--allow-retarget`, and omitted
+`expected_repository` is filled from that pin. Still pass
+`"expected_repository": "weavatrix"` when you already know the folder name — a
+mismatched server fails loudly instead of answering about the wrong repository.
 
 **Blast radius before you edit:**
 
@@ -169,7 +200,7 @@ Three commits before `ec8bf30` this package was 1.9.2; the agent reads that
 follow-up to a diff without touching the worktree. Binary blobs fail closed
 instead of being decoded into garbage.
 
-## The 47 read-only operations
+## The 60 read-only operations
 
 | Workflow | Operations |
 | --- | --- |
@@ -181,6 +212,10 @@ instead of being decoded into garbage.
 | Architecture | `get_architecture_contract`, `verify_architecture`, `verify_capabilities`, `explain_architecture_violation`, `propose_architecture_exception` |
 | Git and repositories | `git_history`, `git_read_blob`, `cross_repo_git`, `open_repo`, `list_known_repos`, `rebuild_graph` |
 | Native extensions | `vector_search`, `semantic_link`, `seo_link_suggestions`, `memory_context` |
+| n8n workflows | `n8n_inventory`, `n8n_trace`, `n8n_context` |
+| Dify apps | `dify_inventory`, `dify_trace`, `dify_context` |
+| Agent packages | `agent_inventory`, `agent_trace`, `agent_context`, `agent_change_impact` |
+| Mermaid diagrams | `diagram_inventory`, `diagram_trace`, `diagram_context` |
 
 Every operation is read-only with respect to the analyzed repository.
 Pagination and explicit limits bound large neighborhoods, histories, searches,
@@ -191,13 +226,39 @@ and contract inventories.
 | Group | Surfaces |
 | --- | --- |
 | Code | Rust; JavaScript/JSX; TypeScript/TSX; Python; Go; Java; C#; C; C++; SQL; Bash/Zsh; Swift; Solidity |
-| Contracts and configuration | GraphQL; Protobuf/gRPC; JSON/JSONC; YAML/Kubernetes; Terraform/HCL; XML |
+| Contracts and configuration | GraphQL; Protobuf/gRPC; JSON/JSONC (n8n after parse); YAML/Kubernetes (Dify after parse); Terraform/HCL; XML |
 | Documents and UI | HTML/Vue/Svelte; CSS/SCSS/Sass/Less; Markdown/MDX; reStructuredText; AsciiDoc |
 
 Cross-surface analysis connects HTTP, GraphQL, gRPC, Kafka, RabbitMQ/AMQP,
 JMS, NATS, SQS, and SNS evidence. Dynamic dispatch that cannot be proved stays
 unresolved; static reachability is never presented as measured coverage.
 
+## Workflow export domains
+
+n8n and Dify are after-parse domains on the same graph, not new MCP servers.
+
+```text
+export in git → existing JSON/YAML parse → typed domain → inventory / trace / context
+```
+
+| Domain | Tools | Status |
+| --- | --- | --- |
+| [n8n](https://weavatrix.com/n8n) | `n8n_inventory`, `n8n_trace`, `n8n_context` | Product 1.12.0 |
+| [Dify](https://weavatrix.com/dify) | `dify_inventory`, `dify_trace`, `dify_context` | Product 1.13.0 |
+| Agent packages | `agent_inventory`, `agent_trace`, `agent_context`, `agent_change_impact` | Product 1.14.0 |
+| Mermaid diagrams | `diagram_inventory`, `diagram_trace`, `diagram_context` | Product 1.14.0 |
+
+| Need | Use |
+| --- | --- |
+| Change impact on an export in git | Weavatrix `n8n_*` / `dify_*` |
+| Live n8n instance | [n8n-mcp](https://github.com/czlonkowski/n8n-mcp) |
+| Live Dify console | [dify-mcp](https://github.com/alexjiaguo/dify-mcp) |
+| CI lint/diff of Dify YAML | [difyctl](https://github.com/JSLEEKR/difyctl) |
+
+No invented n8n/Dify millisecond benches. Those tools share the measured
+installed MCP boundary (2026-08-03): cold **32.21x** (157.34 ms vs 5,068.22 ms)
+and warm **36.85x** (7.94 ms vs 292.55 ms) versus `weavatrix-js` 0.3.15.
+
 ## Product and engine are separate
 
 ```text
@@ -205,12 +266,12 @@ coding agent
     |
     | MCP over stdio
     v
-weavatrix 1.11.0
+weavatrix 1.14.2
     profile catalog · refresh · watcher · MCP framing
     |
     v
-weavatrix-rust 2.10.0
-    typed graph · analysis · 47 read-only operations
+weavatrix-rust 2.14.3
+    typed graph · analysis · 60 product operations including n8n, Dify, agent packages, and Mermaid
 ```
 
 This npm product owns MCP transport and native distribution. The
@@ -276,6 +337,7 @@ Full evidence and methodology:
 - [Tool reference](https://github.com/Weavatrix/weavatrix/blob/main/docs/tool-reference.md)
 - [Language support](https://github.com/Weavatrix/weavatrix/blob/main/docs/language-support.md)
 - [Architecture](https://github.com/Weavatrix/weavatrix/blob/main/docs/architecture.md)
+- [n8n](https://weavatrix.com/n8n) · [Dify](https://weavatrix.com/dify) · [parsers](https://weavatrix.com/parsers)
 - [Benchmarks](https://github.com/Weavatrix/weavatrix/blob/main/docs/benchmarks.md)
 
 ## License

package/package.json

@@ -1,8 +1,8 @@
 {
   "name": "weavatrix",
-  "version": "1.11.0",
-  "weavatrixEngineVersion": "2.10.0",
-  "description": "Native repository intelligence for coding agents: 47 read-only MCP operations backed by typed evidence graphs, impact, architecture, APIs, Git, search, semantics, and memory.",
+  "version": "1.14.2",
+  "weavatrixEngineVersion": "2.14.3",
+  "description": "Native repository intelligence for coding agents: 60 read-only MCP operations backed by typed evidence graphs, impact, architecture, APIs, Git, search, semantics, memory, n8n workflows, Dify apps, agent packages, and Mermaid diagrams.",
   "author": "Sergii Ziborov <sergii.ziborov@gmail.com>",
   "license": "MIT",
   "homepage": "https://weavatrix.com",

package/server.json

@@ -2,19 +2,19 @@
   "$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json",
   "name": "io.github.Weavatrix/weavatrix",
   "title": "Weavatrix",
-  "description": "Native Weavatrix MCP: 47 read-only repository tools across 24 code and configuration surfaces.",
+  "description": "Native Weavatrix MCP: 60 read-only tools for code, n8n, Dify, agent packages, and Mermaid.",
   "repository": {
     "url": "https://github.com/Weavatrix/weavatrix",
     "source": "github"
   },
-  "version": "1.11.0",
+  "version": "1.14.2",
   "websiteUrl": "https://weavatrix.com",
   "packages": [
     {
       "registryType": "npm",
       "registryBaseUrl": "https://registry.npmjs.org",
       "identifier": "weavatrix",
-      "version": "1.11.0",
+      "version": "1.14.2",
       "runtimeHint": "npx",
       "transport": {
         "type": "stdio"

package/skill/SKILL.md

@@ -4,7 +4,9 @@ description: >-
   Use the Weavatrix MCP when a repository task benefits from indexed evidence:
   codebase orientation, symbol/source search, dependency or call graphs, change
   impact, API/event tracing, architecture checks, Git/coverage evidence,
-  duplicates, dead-code review, or semantic context. Skip it for trivial
+  duplicates, dead-code review, semantic context, exported n8n
+  workflows, exported Dify YAML, Agent Plugins/Skills/MCP
+  catalogs, or Mermaid flowcharts. Skip it for trivial
   single-file edits or when a known source file and native tests already answer
   the task.
 ---
@@ -17,15 +19,28 @@ contracts. Do not call it merely because the server is available.
 
 ## Minimal workflow
 
-1. Call `graph_stats` to confirm the active repository and graph revision.
+1. Call `graph_stats` and confirm `repository_context.root` matches the workspace
+   you intend. The MCP process is pinned to its launch root; it will not silently
+   answer for another repository.
 2. Use `module_map` for orientation or `search_code` for a known literal.
 3. Pin decisive evidence with `inspect_symbol`, `go_to_definition`,
    `find_references`, `context_bundle`, or `read_source`.
 4. Expand only when needed: `get_dependents` or `change_impact` for risk,
    `trace_endpoint` or `trace_api_contract` for runtime contracts,
-   `verify_architecture` for policy, and `run_audit` for a broad health pass.
+   `verify_architecture` for policy, `run_audit` for a broad health pass,
+   `n8n_inventory` / `n8n_trace` / `n8n_context` for exported n8n
+   workflows, and `dify_inventory` / `dify_trace` / `dify_context` for
+   exported Dify YAML, and `agent_inventory` / `agent_trace` /
+  `agent_context` / `agent_change_impact` for Agent Plugins, Skills, and
+  supplied MCP catalogs, and `diagram_inventory` / `diagram_trace` /
+  `diagram_context` for Mermaid flowcharts. Those paths are static graph
+  evidence, not a live n8n, Dify, MCP, or Mermaid runtime.
 5. Use repository-native tests or benchmarks for behavioral proof.
 
+Do not call `open_repo` "just in case". Cross-repository tools take explicit
+roots in their arguments. Multi-root retarget requires launching with
+`--allow-retarget`.
+
 Prefer `output_format:"text"` and a small `token_budget` for conversational
 work. Use JSON only for automation or retained evidence. Call `rebuild_graph`
 only when the repository changed before automatic refresh completed or when

package/skill/references/tool-routing.md

@@ -3,8 +3,10 @@
 Open only the card for the method you are about to call. Each card explains
 when to use the method, its exact published inputs, and a minimal compact call.
 The live MCP `tools/list` schema is authoritative. Every answer carries
-`repository_context` (root, scan revision, Git HEAD, graph age); pass
-`expected_repository` to fail fast when the active repository differs.
+`repository_context` (root, scan revision, Git HEAD, graph age). The process
+pins its launch root and injects `expected_repository` when omitted, so answers
+cannot silently switch to another repository. Pass an explicit
+`expected_repository` to fail fast when you already know the intended folder.
 
 ## Graph, source, and orientation
 
@@ -60,7 +62,7 @@ The live MCP `tools/list` schema is authoritative. Every answer carries
 
 ## Repositories and refresh
 
-- [`open_repo`](tools/open_repo.md) — Retarget to another local repository. Graphs unused for 20 minutes unload; asking for one again rescans that folder.
+- [`open_repo`](tools/open_repo.md) — Retarget to another local repository. Default launches refuse this unless started with `--allow-retarget`. Graphs unused for 20 minutes unload; asking for one again rescans that folder.
 - [`list_known_repos`](tools/list_known_repos.md) — List repositories with an in-process graph.
 - [`rebuild_graph`](tools/rebuild_graph.md) — Rebuild the derived in-memory graph without source writes.
 
@@ -70,3 +72,38 @@ The live MCP `tools/list` schema is authoritative. Every answer carries
 - [`vector_search`](tools/vector_search.md) — Exact or bounded approximate nearest-neighbor search.
 - [`seo_link_suggestions`](tools/seo_link_suggestions.md) — Directional SEO internal-link evidence from supplied page profiles.
 - [`memory_context`](tools/memory_context.md) — Compile bounded temporal memory context from supplied events.
+
+## n8n workflows
+
+- [`n8n_inventory`](tools/n8n_inventory.md) — List exported workflows, nodes, entries, and coverage.
+- [`n8n_trace`](tools/n8n_trace.md) — Bounded flow, output-dependency, error, and subworkflow walk.
+- [`n8n_context`](tools/n8n_context.md) — Bounded context for one node or workflow, without secrets.
+
+## Dify apps
+
+Local YAML export only. Not the Dify console API. Shipped in product 1.13.0
+on engine `weavatrix-rust` 2.13.2. See [weavatrix.com/dify](https://weavatrix.com/dify).
+
+- [`dify_inventory`](tools/dify_inventory.md) — List exported apps, nodes, modes, and coverage.
+- [`dify_trace`](tools/dify_trace.md) — Bounded control and data walk, including iteration scope.
+- [`dify_context`](tools/dify_context.md) — Bounded context for one node or app, without secrets.
+
+## Agent packages
+
+Local plugin, skill, catalog, and observation files only. Not a plugin
+runtime, gateway, or policy authority. Requires engine `weavatrix-rust`
+2.14.3.
+
+- [`agent_inventory`](tools/agent_inventory.md) — List plugins, skills, MCP bindings, catalog tools, and supplied observations.
+- [`agent_trace`](tools/agent_trace.md) — Declared origin, profile, transport, and package bindings for one identity.
+- [`agent_context`](tools/agent_context.md) — Bounded context; `allowed-tools` is not a grant.
+- [`agent_change_impact`](tools/agent_change_impact.md) — Compare two catalog snapshots; a known inject transform can keep one exposure compatible.
+
+## Mermaid diagrams
+
+Local flowchart files and Markdown fences only. Not a renderer. Requires
+engine `weavatrix-rust` 2.14.3.
+
+- [`diagram_inventory`](tools/diagram_inventory.md) — List diagrams, native elements, and explicit sidecar bindings.
+- [`diagram_trace`](tools/diagram_trace.md) — Walk `declared_architecture` arrows only.
+- [`diagram_context`](tools/diagram_context.md) — Fragments and binding status; display names are not IDs.

package/skill/references/tools/agent_change_impact.md

@@ -0,0 +1,33 @@
+# `agent_change_impact`
+
+Compare two supplied MCP catalog snapshots and report contract changes,
+adapter compensation, and declared consumers.
+
+## When to use
+
+After adding `before` and `after` catalog JSON files to the repository. A
+known inject transform can keep one exposure compatible while the upstream
+required list grew. Incomplete pagination is not a deletion.
+
+## Inputs
+
+- `before` (string, required) — repository-relative previous catalog snapshot.
+- `after` (string, required) — repository-relative current catalog snapshot.
+- `max_results` (number, optional) — change page size, 1–500, default 100.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "agent_change_impact",
+  "arguments": {
+    "before": "catalogs/before.json",
+    "after": "catalogs/after.json",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/agent_context.md

@@ -0,0 +1,32 @@
+# `agent_context`
+
+Bounded context for one agent package node: declared bindings, source
+fragments, and explicit authorization gaps.
+
+## When to use
+
+Inspect one plugin, skill, or server after `agent_inventory` or `agent_trace`.
+`allowed-tools` stays a declaration. Commands are not executed.
+
+## Inputs
+
+- `label` (string, required) — exact node id or an unambiguous label.
+- `task` (string, optional) — what the caller intends to change or inspect.
+- `max_related` (number, optional) — relation page size, 1–200, default 24.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "agent_context",
+  "arguments": {
+    "label": "lookup",
+    "task": "inspect",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/agent_inventory.md

@@ -0,0 +1,30 @@
+# `agent_inventory`
+
+List Agent Plugins, Skills, MCP server bindings, catalog exposures, and
+supplied observations from local package files.
+
+## When to use
+
+Orient on plugin packages, `SKILL.md`, `mcp.json`, or a supplied catalog
+snapshot already in the repository. This does not launch servers or grant
+permissions.
+
+## Inputs
+
+- `path` (string, optional) — repository-relative package file or path fragment.
+- `max_results` (number, optional) — inventory page size, 1–500, default 200.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "agent_inventory",
+  "arguments": {
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/agent_trace.md

@@ -0,0 +1,30 @@
+# `agent_trace`
+
+Show the declared origin, profile, transport, and package bindings for one
+plugin, skill, MCP server, catalog tool, or observation.
+
+## When to use
+
+After `agent_inventory`, when a visible name is not enough: the same `search`
+label can belong to two packages. Use an exact id if the label is ambiguous.
+
+## Inputs
+
+- `label` (string, required) — exact node id or an unambiguous label.
+- `max_related` (number, optional) — relation page size, 1–200, default 48.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "agent_trace",
+  "arguments": {
+    "label": "search-a",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/diagram_context.md

@@ -0,0 +1,32 @@
+# `diagram_context`
+
+Bounded Mermaid context: source fragments, explicit bindings, and gaps.
+
+## When to use
+
+Inspect one diagram or element after inventory. Exact implementation
+bindings come from `.weavatrix/diagram-links.json`, never from a shared
+display name.
+
+## Inputs
+
+- `label` (string, required) — graph id or unique native element label.
+- `task` (string, optional) — what the caller intends to change or inspect.
+- `max_related` (number, optional) — related row cap, 1–200, default 24.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "diagram_context",
+  "arguments": {
+    "label": "ORDERS",
+    "task": "change createOrder",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/diagram_inventory.md

@@ -0,0 +1,29 @@
+# `diagram_inventory`
+
+List Mermaid flowchart diagrams, native elements, and explicit sidecar
+bindings.
+
+## When to use
+
+Orient on `.mmd`, `.mermaid`, or fenced Mermaid already in the repository.
+Drawn arrows are declared architecture, not production calls.
+
+## Inputs
+
+- `path` (string, optional) — repository-relative diagram file or path fragment.
+- `max_results` (number, optional) — inventory page size, 1–500, default 200.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "diagram_inventory",
+  "arguments": {
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/diagram_trace.md

@@ -0,0 +1,31 @@
+# `diagram_trace`
+
+Walk `declared_architecture` arrows inside one Mermaid diagram.
+
+## When to use
+
+Follow drawn paths from a native element id. This does not treat the arrow
+as a production `calls` edge.
+
+## Inputs
+
+- `label` (string, required) — graph id or unique native element label.
+- `depth` (number, optional) — walk depth, 1–32, default 8.
+- `max_nodes` (number, optional) — page size, 1–500, default 100.
+- `cursor` (string, optional) — opaque page token from a previous page.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "diagram_trace",
+  "arguments": {
+    "label": "ORDERS",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/dify_context.md

@@ -0,0 +1,32 @@
+# `dify_context`
+
+Bounded Dify context for one node: proven consumers, selector sites, and explicit gaps.
+
+## When to use
+
+Inspect one exported Dify node or variable after `dify_inventory` or
+`dify_trace`. Answers include source fragments and consumer lists for impact
+tasks. Secrets stay omitted.
+
+## Inputs
+
+- `label` (string, required) — node, app, or variable id or graph label.
+- `task` (string, optional) — `explain` or `impact`. Impact adds consumer lists.
+- `max_related` (number, optional) — related-item page size.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "dify_context",
+  "arguments": {
+    "label": "LLM",
+    "task": "impact",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/dify_inventory.md

@@ -0,0 +1,28 @@
+# `dify_inventory`
+
+List Dify apps, nodes, modes, and analysis bounds from exported YAML.
+
+## When to use
+
+Orient on one or more exported Dify apps already in the repository. This does
+not call a live Dify console API.
+
+## Inputs
+
+- `path` (string, optional) — repository-relative YAML file or path fragment.
+- `max_results` (number, optional) — inventory page size, 1–500, default 200.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "dify_inventory",
+  "arguments": {
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/dify_trace.md

@@ -0,0 +1,32 @@
+# `dify_trace`
+
+Upstream and downstream Dify port flow, selectors, and typed data relations.
+
+## When to use
+
+Walk declared control flow and typed data edges from one exported Dify node
+or app. A DAG merge is a merge, not a cycle. Missing consumers stay missing.
+
+## Inputs
+
+- `label` (string, required) — node or app id, title, or graph label.
+- `depth` (number, optional) — hop limit, 1–32, default 8.
+- `max_nodes` (number, optional) — page size, 1–500, default 100.
+- `cursor` (string, optional) — `v1:<offset>` or `v1:<offset>:<revision>`.
+- `direction` ("outgoing" | "incoming", optional) — walk direction.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "dify_trace",
+  "arguments": {
+    "label": "LLM",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/n8n_context.md

@@ -0,0 +1,33 @@
+# `n8n_context`
+
+Bounded n8n context for one node: proven dependencies, expression sites, and
+explicit gaps.
+
+## When to use
+
+Inspect one exported node or workflow before changing it. Secrets, cookies,
+auth headers, URL credentials, `pinData`, `staticData`, and `$env` values are
+not placed on the default context.
+
+## Inputs
+
+- `label` (string, required) — exact node or workflow identity.
+- `task` (string, optional) — what the caller intends to change or inspect.
+- `max_related` (number, optional) — related-item bound, default 24.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "n8n_context",
+  "arguments": {
+    "label": "Send Invoice",
+    "task": "change email recipient",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/n8n_inventory.md

@@ -0,0 +1,28 @@
+# `n8n_inventory`
+
+List n8n workflows, nodes, entry points, and analysis bounds from exported JSON.
+
+## When to use
+
+Orient on one or more exported n8n workflows already in the repository. This
+does not call a live n8n API.
+
+## Inputs
+
+- `path` (string, optional) — repository-relative workflow file or path fragment.
+- `max_results` (number, optional) — inventory page size, 1–500, default 200.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "n8n_inventory",
+  "arguments": {
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.

package/skill/references/tools/n8n_trace.md

@@ -0,0 +1,33 @@
+# `n8n_trace`
+
+Upstream and downstream n8n port flow, output dependencies, and static
+subworkflow links.
+
+## When to use
+
+Walk `flows_to`, `depends_on_output`, `handles_error_with`, and
+`calls_workflow` from one node. Do not treat those relations as the same
+walk. Missing subworkflows are not provided, not deleted.
+
+## Inputs
+
+- `label` (string, required) — exact node or workflow identity.
+- `depth` (number, optional) — hop bound, 1–32, default 8.
+- `max_nodes` (number, optional) — page size, 1–500, default 100.
+- `cursor` (string, optional) — opaque `v1:<offset>` token from `page.next_cursor`.
+- `output_format` ("text" | "json" | "structured", default "json") — text returns the concise text block only; json returns structured output and mirrors it into text for clients that read only content; structured drops that mirror, which is the larger copy, and is safe only where the client reads structuredContent.
+
+## Minimal call
+
+```json
+{
+  "name": "n8n_trace",
+  "arguments": {
+    "label": "Send Invoice",
+    "output_format": "text"
+  }
+}
+```
+
+Use `output_format:"text"` for compact agent interaction. The live MCP
+`tools/list` schema remains authoritative for this installed version.