fix: remove legacy service-connection MCP field (#233)

Remove the dead `service_connection` field from `McpOptions` and fix
incorrect documentation claiming 1ES uses service connections for MCP
routing. Both targets use MCPG — the field was parsed but never read.

- Remove `service_connection` field from `McpOptions` in types.rs
- Remove `service-connection:` from AGENTS.md MCP config docs
- Fix create/update prompts: 1ES supports containerized MCPs via MCPG

Closes #233

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: jamesadevine

AGENTS.md

@@ -1561,7 +1561,6 @@ mcp-servers:
 **Common (both types):**
 - `allowed:` - Array of tool names the agent is permitted to call (required for security)
 - `env:` - Environment variables for the MCP server process. Use `""` (empty string) for passthrough from the pipeline environment.
-- `service-connection:` - (1ES target only) Override the service connection name. Defaults to `mcp-<name>-service-connection`
 
 ### Environment Variable Passthrough
 

prompts/create-ado-agentic-workflow.md

@@ -174,7 +174,7 @@ target: 1es
 | Value | Generates |
 |---|---|
 | `standalone` | Full 3-job pipeline with AWF network sandbox and Squid proxy |
-| `1es` | Pipeline extending `1ES.Unofficial.PipelineTemplate.yml`; no custom proxy; MCPs via service connections |
+| `1es` | Pipeline extending `1ES.Unofficial.PipelineTemplate.yml`; no custom proxy; MCPs via MCPG |
 
 ### Step 8 — MCP Servers
 
@@ -216,8 +216,6 @@ mcp-servers:
 
 > **Security**: Custom MCPs must have an explicit `allowed:` list. Built-in MCPs default to all tools when set to `true`.
 >
-> **1ES target**: Custom containerized MCPs are not supported — only built-ins with service connections.
->
 > **Standalone target** (the default): Built-in MCPs (entries without a `container:` or `url:` field) are silently skipped at compile time — they have no effect and will not be available to the agent. For the standalone target, use only **custom** containerized MCPs with a `container:` field.
 
 ### Step 9 — Safe Outputs
@@ -542,4 +540,4 @@ safe-outputs:
 - **No direct writes**: All mutations go through safe outputs — the agent cannot push code or call write APIs directly.
 - **Compile before committing**: Always compile with `ado-aw compile` and commit both the `.md` source and generated `.yml` together.
 - **Check validation**: The compiler will error if write safe-outputs are configured without `permissions.write`.
-- **1ES target limits**: No custom MCPs, no custom network allow-lists — these are handled by OneBranch infrastructure.
+- **1ES target limits**: No custom network allow-lists — these are handled by OneBranch infrastructure.

prompts/update-ado-agentic-workflow.md

@@ -250,7 +250,7 @@ Before finalizing any update, verify:
 
 6. **Network domains**: If new MCPs or external services are added, ensure required domains are in `network.allowed`.
 
-7. **Target compatibility**: If `target: 1es`, custom MCPs (with `container:`) are not supported — only built-in MCPs with service connections.
+7. **Target compatibility**: Both `standalone` and `1es` targets support containerized MCPs via MCPG.
 
 8. **Safe output `target` fields**: `comment-on-work-item` requires an explicit `target` field. `update-work-item` fields require explicit opt-in (`status: true`, `title: true`, etc.).
 

src/compile/types.rs

@@ -707,9 +707,6 @@ pub struct McpOptions {
     /// Environment variables for the MCP server process
     #[serde(default)]
     pub env: HashMap<String, String>,
-    /// Service connection name (1ES only, auto-generated if not specified)
-    #[serde(default, rename = "service-connection")]
-    pub service_connection: Option<String>,
 }
 
 /// Trigger configuration for the pipeline