@raishin/vanguard-frontier-agentic 1.0.0

package/docs/normalized-platform-matrix.md

@@ -0,0 +1,109 @@
+# Normalized Platform Matrix
+
+Last reviewed: 2026-04-28
+
+This matrix separates three things that are easy to blur:
+
+1. the **normalized labels** this repo uses for cross-harness assets,
+2. the **vendor product names** users recognize,
+3. the **actual file or adapter shapes** needed for each platform.
+
+Do not pretend these platforms are interchangeable. Some are Markdown-family adapters, some are structured configs, and some are broader agent products that this repo does **not** target yet.
+
+## Repo-supported normalized labels
+
+These are the labels currently used by this repository in metadata, docs, and harness variants.
+
+| Normalized label | User-facing product | Adapter family | Typical repo artifact(s) | Current repo support | Notes |
+| --- | --- | --- | --- | --- | --- |
+| `codex` | OpenAI Codex | Structured config | `harnesses/codex.toml` | Supported | Native TOML adapter, not Markdown body. |
+| `copilot` | GitHub Copilot / VS Code custom agent | Markdown + YAML frontmatter | `harnesses/copilot.agent.md` | Supported | Also related to Copilot Agent Skills, but skills are a separate portability layer. |
+| `claude-code` | Claude Code | Markdown + YAML frontmatter | `harnesses/claude-code.agent.md` | Supported | Uses subagent-style Markdown definitions. |
+| `cursor` | Cursor | Markdown + YAML frontmatter | `harnesses/cursor.agent.md` | Supported | Keep Cursor-specific fields out unless verified. |
+| `gemini` | Gemini CLI | Markdown + YAML frontmatter | `harnesses/gemini.agent.md` | Supported | Treat Gemini CLI as the concrete executable target, not “Gemini” in the abstract. |
+| `kiro` | Kiro | Split target: Markdown + JSON | `harnesses/kiro-ide.agent.md`, `harnesses/kiro-cli.agent.json` | Supported | Kiro IDE and Kiro CLI are related but not interchangeable. |
+| `other` | Unclassified / future | Varies | none by default | Placeholder only | Use only when a platform is real but not yet normalized in repo conventions. |
+
+## Repo-level instruction entrypoints
+
+If the goal is to let a platform **open this repository and immediately inherit project guidance**, the runtime entrypoint matters as much as the agent adapter format.
+
+| Platform | Repo-level guidance file or path | Status in this repo | Notes |
+| --- | --- | --- | --- |
+| Codex | `AGENTS.md` | Present | Use `AGENTS.md` for repo guidance; custom subagents are separate TOML files under `.codex/agents/` when needed. |
+| Cursor | `AGENTS.md` | Present | Cursor docs explicitly support root `AGENTS.md` for project instructions. |
+| Kiro IDE | `AGENTS.md` or `.kiro/steering/` | `AGENTS.md` present | Kiro docs support `AGENTS.md`; steering files are an additional option. |
+| Claude Code | `CLAUDE.md` | Present | Claude Code uses `CLAUDE.md` for project memory and `.claude/agents/` for custom subagents. |
+| Gemini CLI | `GEMINI.md` | Present | Gemini CLI uses hierarchical `GEMINI.md` memory and `.gemini/agents/` for custom subagents. |
+| GitHub Copilot | `.github/copilot-instructions.md` | Present | Repo-level instructions are separate from `.github/agents/` custom agent profiles. |
+| GitHub Copilot custom agents | `.github/agents/` | Not populated by default | This repo ships harness adapters under `agents/**/harnesses/`; consumers can copy selected adapters into `.github/agents/` when they want active repo-level custom agents. |
+| Kiro CLI custom agents | `.kiro/agents/*.json` | Not populated by default | This repo stores Kiro CLI adapters under `agents/**/harnesses/kiro-cli.agent.json`; copy selected ones into `.kiro/agents/` in a consuming repo. |
+| Claude Code custom subagents | `.claude/agents/*.md` | Not populated by default | This repo stores Claude adapters under `agents/**/harnesses/claude-code.agent.md`; copy selected ones into `.claude/agents/` in a consuming repo. |
+| Gemini CLI custom subagents | `.gemini/agents/*.md` | Not populated by default | This repo stores Gemini adapters under `agents/**/harnesses/gemini.agent.md`; copy selected ones into `.gemini/agents/` in a consuming repo. |
+
+The important distinction:
+
+- **repo-level guidance files** tell the main agent how to behave in this repository
+- **custom agent/subagent files** add specialized delegates that can be installed separately
+
+## Broader agentic platform candidates
+
+These matter for market awareness, but they are **not** normalized repo labels yet.
+
+| Candidate product | Suggested future normalized label | Why it matters | Current repo status | Why not normalized yet |
+| --- | --- | --- | --- | --- |
+| Windsurf | `windsurf` | Agentic IDE with MCP-aware workflows | Not supported | No harness convention or repo adapter contract yet. |
+| Cline | `cline` | Strong customization model with rules, skills, workflows, hooks, CLI | Not supported | Different customization surface than the current harness set. |
+| Continue | `continue` | Agent mode plus hosted/shared agents | Not supported | Needs a clear adapter contract before adding to repo taxonomy. |
+
+## Product names that users may mention but should be normalized carefully
+
+| User says | Normalize to | Why |
+| --- | --- | --- |
+| “Claude” | `claude-code` only when they mean Claude Code subagents | Claude the model family is not the same thing as Claude Code the agent surface. |
+| “Gemini” | `gemini` only when they mean Gemini CLI adapter support | Gemini as a model/vendor is broader than the executable harness target. |
+| “Kiro” | `kiro` plus explicit IDE vs CLI distinction when generating files | Kiro has at least two materially different adapter shapes in this repo. |
+| “Copilot” | `copilot` | This was an easy miss when listing agentic platforms; it is already a first-class harness here. |
+
+## Recommended naming rule for this repo
+
+When updating catalog entries, agent metadata, or skills:
+
+- use the **normalized label** in machine-readable metadata,
+- use the **user-facing product name** in prose,
+- and keep the **exact adapter filename** specific to the executable target.
+
+Example:
+
+- metadata harness label: `kiro`
+- prose name: `Kiro`
+- executable adapters:
+  - `harnesses/kiro-ide.agent.md`
+  - `harnesses/kiro-cli.agent.json`
+
+## What was missing from the original short list
+
+The strongest miss was:
+
+- **GitHub Copilot**
+
+Additional serious candidates for later normalization:
+
+- **Windsurf**
+- **Cline**
+- **Continue**
+
+## Source grounding
+
+Official references reviewed for this normalization work:
+
+- Anthropic Claude Code subagents: <https://docs.anthropic.com/en/docs/claude-code/sub-agents>
+- GitHub Copilot custom agents: <https://code.visualstudio.com/docs/copilot/customization/custom-agents>
+- GitHub Copilot agent skills: <https://code.visualstudio.com/docs/copilot/customization/agent-skills>
+- OpenAI Codex docs: <https://platform.openai.com/docs/codex>
+- Cursor docs: <https://docs.cursor.com/>
+- Kiro steering and hooks docs: <https://kiro.dev/docs/steering/> and <https://kiro.dev/docs/hooks/types/>
+- Gemini CLI repo/docs entrypoint: <https://github.com/google-gemini/gemini-cli>
+- Windsurf docs: <https://docs.windsurf.com/>
+- Cline docs: <https://docs.cline.bot/home>
+- Continue docs: <https://docs.continue.dev/intro>

package/docs/quality-bar.md

@@ -0,0 +1,27 @@
+# Quality Bar
+
+## Minimum bar
+
+Every asset must be:
+
+- cloud-relevant,
+- metadata-complete,
+- provenance-clear,
+- security-reviewed,
+- compatible with at least one declared harness,
+- linked to official documentation when making provider claims.
+
+## Rejection triggers
+
+Reject assets that:
+
+- are generic prompts with cloud labels pasted on,
+- make unverified claims,
+- encourage broad admin permissions without justification,
+- mutate infrastructure without approval gates,
+- include secrets or customer-specific data,
+- duplicate an existing asset without meaningful improvement.
+
+## Security-sensitive assets
+
+IAM, RBAC, Terraform, Kubernetes, database, incident response, and MCP assets need explicit threat notes. If the workflow could affect production, require human approval and rollback expectations.

package/docs/release-versioning.md

@@ -0,0 +1,114 @@
+# Release and Versioning
+
+This repository can be published as an npm package that carries a catalog of
+agentic assets: skills, agents, rules, MCP references, schemas, docs, and
+manifests.
+
+## Semantic Versioning Policy
+
+Use `MAJOR.MINOR.PATCH`.
+
+### PATCH: safe fixes
+
+Use a patch bump for changes that should not break consumers:
+
+- typo fixes in docs or prompts,
+- clearer wording without changing behavior,
+- metadata corrections that preserve IDs and paths,
+- adding missing official-doc links,
+- regenerating manifests after non-behavioral edits,
+- fixing validation scripts without changing the manifest format.
+
+Example:
+
+```bash
+npm version patch
+```
+
+### MINOR: additive capabilities
+
+Use a minor bump for backwards-compatible additions:
+
+- adding a new skill, agent, rule, MCP reference, or asset,
+- adding new metadata fields while preserving old fields,
+- adding a new provider folder,
+- adding new validation checks that current valid assets can pass,
+- adding a new manifest section without removing existing fields.
+
+Example:
+
+```bash
+npm version minor
+```
+
+### MAJOR: breaking changes
+
+Use a major bump when consumers must adapt:
+
+- renaming or removing skill IDs,
+- moving cataloged paths without compatibility aliases,
+- changing schema-required fields,
+- changing manifest format incompatibly,
+- removing catalog entries,
+- changing package file layout,
+- changing the trust/security contract of an asset in a way that alters how it
+  should be used.
+
+Example:
+
+```bash
+npm version major
+```
+
+## Pre-1.0 Reality
+
+While the package is below `1.0.0`, npm and SemVer convention treat the API as
+not fully stable. Be disciplined anyway:
+
+- `0.1.x` patch: safe fixes.
+- `0.x.0` minor: meaningful feature additions or breaking changes during early
+  design.
+- Call out breaking changes explicitly in release notes even before `1.0.0`.
+
+Ruthless rule: do not hide a breaking catalog/schema change inside a patch
+release just because the project is young.
+
+## Release Gate
+
+Before publishing:
+
+```bash
+npm run manifest:write
+npm run validate
+npm pack --dry-run
+```
+
+Review the `npm pack --dry-run` output. If secrets, local caches, or generated
+junk appear in the package, stop and fix `files` or `.npmignore`.
+
+## Skill Integrity Manifest
+
+`catalog/skill-manifest.json` records SHA-256 hashes for every file under every
+cataloged skill directory.
+
+Use it for:
+
+- detecting accidental edits,
+- verifying copied skills,
+- release review,
+- npm package integrity checks,
+- downstream installer trust decisions.
+
+Workflow:
+
+```bash
+# After intentional skill edits
+npm run manifest:write
+
+# In CI or before publish
+npm run manifest:check
+```
+
+The manifest proves file integrity for repository contents. It does not prove
+that a skill is safe, correct, compliant, or officially endorsed.
+

package/docs/taxonomy.md

@@ -0,0 +1,39 @@
+# Cloud Marketplace Taxonomy
+
+## Providers
+
+- `aws`
+- `azure`
+- `oracle`
+- `oci`
+- `gcp`
+- `kubernetes`
+- `terraform`
+- `multi-cloud`
+- `generic`
+
+## Asset types
+
+- `skill`: workflow instructions for a recurring task.
+- `agent`: role/persona definition with responsibilities and review behavior.
+- `rule`: harness-specific operating guidance.
+- `mcp-reference`: catalog entry for an MCP server or MCP setup path.
+
+## Harnesses
+
+- `codex`
+- `copilot`
+- `claude-code`
+- `cursor`
+- `gemini`
+- `kiro`
+- `other`
+
+## Trust levels
+
+- **Official**: published by the cloud/provider or official project owner.
+- **Community**: public third-party source with clear maintainer and license.
+- **Original**: created for this repository.
+- **Adapted**: derived from another source and license-reviewed.
+
+Do not blur these categories. A community MCP server that targets AWS is not an official AWS MCP server.

package/mcp/AGENTS.md

@@ -0,0 +1,17 @@
+# AGENTS.md
+
+## Purpose
+- Store MCP references with trust, auth, and safety notes.
+
+## Patterns
+- `official/` → vendor-official MCP references.
+- `community/` → non-official MCP references.
+- `examples/` → sample config snippets only.
+- `*.metadata.json` must align with `catalog/mcp-references.json`.
+
+## Do Not Miss
+- MCP tools can expose or mutate real cloud resources.
+- State auth model, vendor, install example, and unofficial warning.
+- Prefer official project URLs and vendor docs.
+- Run `npm run validate` after metadata edits.
+

package/mcp/community/README.md

@@ -0,0 +1,3 @@
+# Community MCP References
+
+Community MCP entries are allowed only when the maintainer, license, source URL, auth model, and security risks are clear. Do not present community servers as official provider tooling.

package/mcp/examples/README.md

@@ -0,0 +1,3 @@
+# MCP Examples
+
+Place minimal, secret-free MCP configuration examples here. Prefer links to official setup docs over copied config blocks that may drift.

package/mcp/official/aws-mcp-servers.md

@@ -0,0 +1,17 @@
+# AWS Official MCP Servers
+
+- Vendor: AWS Labs
+- Status: official AWS Labs open-source MCP server suite
+- Source: <https://github.com/awslabs/mcp>
+- Docs: <https://awslabs.github.io/mcp/>
+- Auth model: AWS credentials supplied through the selected MCP server/client environment.
+- Mutation risk: varies by server; some servers are documentation-only, others can interact with AWS APIs.
+- Last verified: 2026-04-27
+
+## Install/config
+
+Use the official repository and server-specific documentation. Do not copy credentials into MCP config files; use named profiles, environment variables, or managed identity patterns where supported.
+
+## Security notes
+
+Start with documentation/read-only servers before API mutation servers. Scope AWS credentials by account, region, role, and permission boundary where possible.

package/mcp/official/aws-mcp-servers.metadata.json

@@ -0,0 +1,21 @@
+{
+  "id": "aws-official-mcp-servers",
+  "name": "AWS Official MCP Servers",
+  "type": "mcp-reference",
+  "provider": "aws",
+  "harnesses": ["codex", "claude-code", "cursor", "gemini", "kiro", "other"],
+  "summary": "Reference entry for the official AWS Labs MCP server suite.",
+  "source_type": "reference-only",
+  "official_docs": [
+    "https://github.com/awslabs/mcp",
+    "https://awslabs.github.io/mcp/"
+  ],
+  "security_notes": "AWS MCP servers may read or mutate AWS resources depending on server and credentials. Scope AWS profiles, regions, and IAM permissions carefully.",
+  "last_verified": "2026-04-27",
+  "path": "mcp/official/aws-mcp-servers.md",
+  "official_project_url": "https://github.com/awslabs/mcp",
+  "vendor": "AWS Labs",
+  "auth_model": "AWS credentials through the selected server/client environment, commonly profile or environment-based credentials.",
+  "install_example": "Follow the server-specific install commands in the official AWS Labs MCP repository or documentation.",
+  "unofficial_warning": "Official AWS Labs project; individual server maturity and deprecation status still varies."
+}

package/mcp/official/azure-mcp-server.md

@@ -0,0 +1,68 @@
+# Azure MCP Server
+
+- Vendor: Microsoft
+- Status: official Microsoft Azure MCP Server
+- Docs: <https://learn.microsoft.com/en-us/azure/developer/azure-mcp-server/overview>
+- Source: <https://github.com/microsoft/azure-mcp>
+- Microsoft MCP catalog: <https://github.com/microsoft/mcp>
+- Auth model: Microsoft Entra ID / Azure Identity patterns through supported clients.
+- Mutation risk: tools can interact with Azure resources according to RBAC permissions.
+- Last verified: 2026-04-27
+
+## Install/config
+
+Use the current Microsoft Learn setup instructions for your client. Avoid tenant-wide or subscription-wide roles unless required and approved.
+
+## Other official Microsoft MCP servers listed in `microsoft/mcp`
+
+The Azure MCP Server is only one entry in Microsoft's broader MCP catalog. Based on the `microsoft/mcp` README as viewed on **2026-04-27**, Microsoft also lists the following MCP servers.
+
+### Catalog snapshot
+
+| Name                                    | Repo / docs                              | Category                 | Type   | Endpoint / notes                                                                                   | What it is for                                                                                   |
+| --------------------------------------- | ---------------------------------------- | ------------------------ | ------ | -------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------ |
+| ✨ Microsoft Foundry                    | Microsoft Learn docs                     | Cloud and infrastructure | Remote | `https://mcp.ai.azure.com`                                                                         | Foundry tools for models, knowledge, evaluation, and related AI workflows.                       |
+| Azure DevOps                            | Azure DevOps MCP Server                  | Developer tools          | Local  | Local server                                                                                       | Azure DevOps work from the editor.                                                               |
+| ☸️ Azure Kubernetes Service (AKS)       | `Azure/aks-mcp`                          | Cloud and infrastructure | Local  | Local server                                                                                       | Natural-language interaction with AKS clusters.                                                  |
+| GitHub                                  | `github/github-mcp-server`               | Developer tools          | Remote | `https://api.githubcopilot.com/mcp`                                                                | GitHub repositories, issues, and pull requests.                                                  |
+| GitHub Awesome-Copilot                  | `github/awesome-copilot`                 | Developer tools          | Local  | Local server                                                                                       | Copilot instructions, prompts, and configs.                                                      |
+| Markitdown                              | `microsoft/markitdown`                   | Developer tools          | Local  | Local server                                                                                       | Markdown conversion and transformation workflows.                                                |
+| Microsoft 365 Agents Toolkit            | `OfficeDev/microsoft-365-agents-toolkit` | Developer tools          | Local  | Local server                                                                                       | Building apps and agents for Microsoft 365 and Copilot.                                          |
+| Microsoft 365 Calendar                  | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_CalendarTools`        | Calendar events, invites, and availability.                                                      |
+| Microsoft 365 Copilot Chat              | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_M365Copilot`          | Search and chat across Microsoft 365 content.                                                    |
+| Microsoft 365 Mail                      | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_MailTools`            | Mail creation, replies, updates, delete, and search.                                             |
+| Microsoft 365 User                      | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_MeServer`             | User, manager, team, and reportee details.                                                       |
+| ⚙️ Microsoft Admin Center               | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_AdminTools`           | Admin Center actions and admin-facing APIs.                                                      |
+| Microsoft Clarity                       | `microsoft/clarity-mcp-server`           | Data and analytics       | Local  | Local server                                                                                       | Clarity analytics export access for MCP clients.                                                 |
+| Microsoft Dataverse                     | Microsoft Dataverse docs                 | Data and analytics       | Local  | Local server                                                                                       | Discover tables, query data, and update business records.                                        |
+| Microsoft Dev Box                       | `@microsoft/devbox-mcp`                  | Developer tools          | Local  | npm package                                                                                        | Dev Box management and environment operations.                                                   |
+| Microsoft Fabric (Public Preview)       | `microsoft/mcp`                          | Data and analytics       | Local  | Local-first                                                                                        | Fabric APIs, item definitions, and best-practice guidance without a live connection requirement. |
+| Microsoft Fabric Real-Time Intelligence | RTI MCP Server docs                      | Data and analytics       | Local  | Local server                                                                                       | RTI querying and analysis workflows.                                                             |
+| Microsoft Learn                         | `microsoftdocs/mcp`                      | Productivity             | Remote | `https://learn.microsoft.com/api/mcp`                                                              | Real-time access to official Microsoft documentation.                                            |
+| Microsoft Sentinel Data Exploration     | Sentinel docs                            | Security                 | Remote | `https://sentinel.microsoft.com/mcp/data-exploration`                                              | Natural-language exploration of Sentinel data lake content.                                      |
+| Microsoft SQL                           | MSSQL MCP Server docs                    | Developer tools          | Local  | Local server                                                                                       | Conversational SQL schema, table, and CRUD workflows.                                            |
+| Microsoft Teams                         | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_TeamsServer`          | Teams chats, channels, users, and messages.                                                      |
+| Microsoft Word                          | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_WordServer`           | Read, create, and collaborate on Word documents.                                                 |
+| NuGet MCP Server                        | `NuGet/Home`                             | Developer tools          | Local  | Local server                                                                                       | NuGet package-management tooling and automation.                                                 |
+| OneDrive and SharePoint                 | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_ODSPRemoteServer`     | OneDrive and SharePoint file integration.                                                        |
+| SharePoint Lists                        | `bap-microsoft/MCP-Platform`             | Productivity             | Remote | `https://agent365.svc.cloud.microsoft/agents/tenants/{tenant_id}/servers/mcp_SharePointListsTools` | SharePoint list, library, and collaboration workflows.                                           |
+| Playwright                              | `microsoft/playwright-mcp`               | Developer tools          | Local  | Local server                                                                                       | Structured browser interaction through accessibility snapshots.                                  |
+| Wassette                                | `microsoft/wassette`                     | Developer tools          | Local  | Local server                                                                                       | Security-oriented WebAssembly component runtime via MCP.                                         |
+
+### Emoji note
+
+I copied the visible icon-style markers Microsoft uses in the catalog headings where they are explicit in the page markup:
+
+- `✨` Microsoft Foundry
+- `☸️` Azure Kubernetes Service (AKS)
+- `⚙️` Microsoft Admin Center
+
+Other entries in the GitHub-rendered page appear without a distinctive emoji marker in the captured listing, so I did not invent extra ones.
+
+### Evidence note
+
+The `microsoft/mcp` page mixes true Azure services, Microsoft 365/Graph endpoints, Fabric/data tooling, and general developer tooling. Do not assume every entry is Azure-scoped just because it appears in Microsoft's MCP catalog.
+
+## Security notes
+
+RBAC is the blast-radius boundary. Use least-privilege roles, narrow scopes, and time-bound access where possible.

package/mcp/official/azure-mcp-server.metadata.json

@@ -0,0 +1,21 @@
+{
+  "id": "azure-official-mcp-server",
+  "name": "Azure MCP Server",
+  "type": "mcp-reference",
+  "provider": "azure",
+  "harnesses": ["codex", "claude-code", "cursor", "gemini", "kiro", "other"],
+  "summary": "Reference entry for Microsoft's Azure MCP Server.",
+  "source_type": "reference-only",
+  "official_docs": [
+    "https://learn.microsoft.com/en-us/azure/developer/azure-mcp-server/overview",
+    "https://github.com/microsoft/azure-mcp"
+  ],
+  "security_notes": "Azure MCP Server uses Azure identity patterns and Azure RBAC. Do not grant broad subscription permissions by default.",
+  "last_verified": "2026-04-27",
+  "path": "mcp/official/azure-mcp-server.md",
+  "official_project_url": "https://github.com/microsoft/azure-mcp",
+  "vendor": "Microsoft",
+  "auth_model": "Microsoft Entra ID through Azure Identity patterns, Azure CLI, managed identity, or supported client authentication.",
+  "install_example": "Follow Microsoft Learn's Azure MCP Server setup instructions for the target client.",
+  "unofficial_warning": "Official Microsoft project; verify version-specific setup in Microsoft Learn before use."
+}

package/mcp/official/oracle-mcp-servers.md

@@ -0,0 +1,17 @@
+# Oracle MCP Servers
+
+- Vendor: Oracle
+- Status: official Oracle MCP resources
+- Landing page: <https://www.oracle.com/mcp>
+- Source: <https://github.com/oracle/mcp>
+- Auth model: varies by server, including Oracle Database, SQLcl, OCI, MySQL, or documentation-specific authentication.
+- Mutation risk: varies by server; database and OCI access can expose or mutate sensitive systems.
+- Last verified: 2026-04-27
+
+## Install/config
+
+Start from Oracle's MCP landing page and the official `oracle/mcp` repository. Do not assume all Oracle MCP servers share the same runtime, authentication, or permission model.
+
+## Security notes
+
+Use scoped database users, least-privilege OCI policies, network restrictions, and audit logging. Treat database-connected MCP servers as sensitive production access unless proven otherwise.

package/mcp/official/oracle-mcp-servers.metadata.json

@@ -0,0 +1,21 @@
+{
+  "id": "oracle-official-mcp-servers",
+  "name": "Oracle MCP Servers",
+  "type": "mcp-reference",
+  "provider": "oracle",
+  "harnesses": ["codex", "claude-code", "cursor", "gemini", "kiro", "other"],
+  "summary": "Reference entry for Oracle's official MCP server offerings and repository.",
+  "source_type": "reference-only",
+  "official_docs": [
+    "https://www.oracle.com/mcp",
+    "https://github.com/oracle/mcp"
+  ],
+  "security_notes": "Oracle MCP servers may expose database or OCI capabilities. Scope database users, OCI policies, and network access carefully.",
+  "last_verified": "2026-04-27",
+  "path": "mcp/official/oracle-mcp-servers.md",
+  "official_project_url": "https://github.com/oracle/mcp",
+  "vendor": "Oracle",
+  "auth_model": "Varies by server; commonly database credentials, OCI configuration, or product-specific authentication.",
+  "install_example": "Start from Oracle's MCP landing page and the oracle/mcp repository for server-specific setup.",
+  "unofficial_warning": "Official Oracle resources; server capabilities and authentication vary by Oracle product."
+}

package/package.json

@@ -0,0 +1,49 @@
+{
+  "name": "@raishin/vanguard-frontier-agentic",
+  "version": "1.0.0",
+  "description": "Cloud and zero-trust agentic workflow marketplace for skills, agents, rules, MCP references, and compliance-aware architecture.",
+  "license": "Apache-2.0",
+  "type": "commonjs",
+  "bin": {
+    "vfa-export-agents": "./scripts/export-marketplace-agents.mjs"
+  },
+  "files": [
+    "README.md",
+    "LICENSE",
+    "scripts/",
+    "plugins/",
+    "catalog/",
+    "schemas/",
+    "skills/",
+    "agents/",
+    "rules/",
+    "mcp/",
+    "docs/",
+    "templates/",
+    "assets/"
+  ],
+  "scripts": {
+    "agents:export": "node scripts/export-marketplace-agents.mjs",
+    "manifest:write": "python3 tests/validate-skill-manifest.py --write",
+    "manifest:check": "python3 tests/validate-skill-manifest.py",
+    "validate:catalog": "python3 tests/validate-catalog.py",
+    "validate:links": "python3 tests/validate-links.py --offline",
+    "validate": "npm run validate:catalog && npm run manifest:check && npm run validate:links"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "keywords": [
+    "agentic",
+    "cloud",
+    "zero-trust",
+    "compliance",
+    "soc2",
+    "pci-dss",
+    "nis2",
+    "nist",
+    "mcp",
+    "codex",
+    "terraform"
+  ]
+}

package/plugins/cross-platform-agent-template/.codex-plugin/plugin.json

@@ -0,0 +1,40 @@
+{
+  "name": "cross-platform-agent-template",
+  "version": "0.1.0",
+  "description": "Cross-platform agent templates for a curated cloud and zero-trust AI workflow marketplace, with explicit portability boundaries across major coding-agent platforms.",
+  "author": {
+    "name": "Raishin",
+    "url": "https://github.com/Raishin"
+  },
+  "repository": "https://github.com/Raishin/vanguard-frontier-agentic",
+  "license": "Apache-2.0",
+  "keywords": [
+    "agentic",
+    "agents",
+    "cross-platform",
+    "codex",
+    "copilot",
+    "claude-code",
+    "cursor",
+    "gemini",
+    "kiro"
+  ],
+  "skills": "./skills/",
+  "interface": {
+    "displayName": "Cross-Platform Agent Template",
+    "shortDescription": "Portable agent templates for secure cloud and zero-trust AI workflows.",
+    "longDescription": "A cross-platform agent template plugin for Vanguard Frontier Agentic: a curated marketplace of cloud, security, compliance-aware, and zero-trust AI workflows. It helps authors define one strong canonical agent specification, then adapt it honestly across Codex, GitHub Copilot, Claude Code, Cursor, Gemini CLI, and Kiro without pretending the platforms are identical.",
+    "developerName": "Raishin",
+    "category": "Productivity",
+    "capabilities": [
+      "Interactive",
+      "Write"
+    ],
+    "defaultPrompt": [
+      "Create a cloud-security agent spec and adapt it to Codex, Copilot, Claude Code, Cursor, Gemini CLI, and Kiro.",
+      "Normalize a zero-trust workflow across supported agent platforms and show the non-portable fields.",
+      "Generate only the requested adapters and keep compliance, least-privilege, and evidence-backed guidance intact."
+    ],
+    "brandColor": "#3B82F6"
+  }
+}

package/rules/AGENTS.md

@@ -0,0 +1,15 @@
+# AGENTS.md
+
+## Purpose
+- Store harness-specific always-on behavior rules.
+
+## Patterns
+- `rules/<harness>/<rule-id>.md` is the rule body.
+- `rules/<harness>/<rule-id>.metadata.json` mirrors catalog fields.
+- `catalog/rules.json` path must match the rule file.
+
+## Rules
+- Keep rules short, enforceable, and harness-specific.
+- Do not duplicate full skill procedures here.
+- Run `npm run validate` after rule metadata edits.
+

package/rules/codex/cloud-provenance-rule.md

@@ -0,0 +1,14 @@
+---
+metadata:
+  author: github: Raishin
+---
+
+# Cloud Provenance Rule
+
+When answering cloud, IAM, Terraform, Kubernetes, database, or MCP questions:
+
+- Verify drift-prone facts against official documentation or live configuration where possible.
+- Label facts, inferences, assumptions, and uncertainty separately.
+- Do not invent ARNs, subscription IDs, tenancy OCIDs, regions, resource names, or account IDs.
+- Prefer least privilege and read-only inspection before mutation.
+- If official docs and live state disagree, report the conflict instead of forcing a false conclusion.

package/rules/codex/cloud-provenance-rule.metadata.json

@@ -0,0 +1,18 @@
+{
+  "id": "cloud-provenance-rule",
+  "name": "Cloud Provenance Rule",
+  "type": "rule",
+  "provider": "multi-cloud",
+  "harnesses": [
+    "codex"
+  ],
+  "summary": "Codex rule requiring official-source grounding and explicit uncertainty for cloud claims.",
+  "source_type": "original",
+  "official_docs": [
+    "https://modelcontextprotocol.io/"
+  ],
+  "security_notes": "Prevents agents from presenting unverified cloud behavior as fact.",
+  "last_verified": "2026-04-27",
+  "path": "rules/codex/cloud-provenance-rule.md",
+  "author": "github: Raishin"
+}