наблюдаемость: измерять бюджет каталога MCP

crates/crank-community-mcp/src/catalog.rs

@@ -8,7 +8,9 @@ use crank_core::{CacheScope, CoordinationStateStore, CoordinationStateValue};
 use crank_registry::{PostgresRegistry, PublishedAgentTool, RegistryError};
 use serde::{Deserialize, Serialize};
 use tokio::sync::{Mutex, RwLock};
-use tracing::info;
+use tracing::{info, warn};
+
+use crate::manifest::analyze_published_tool_catalog;
 
 #[derive(Clone)]
 pub struct PublishedToolCatalog {
@@ -105,6 +107,7 @@ impl PublishedToolCatalog {
         }
 
         if let Some((tools, age)) = self.load_shared_snapshot(workspace_slug, agent_slug).await {
+            log_catalog_analysis(workspace_slug, agent_slug, "shared_cache", &tools);
             let mut guard = self.cached.write().await;
             guard.insert(
                 key,
@@ -125,6 +128,7 @@ impl PublishedToolCatalog {
             Err(RegistryError::PublishedAgentNotFound { .. }) => Vec::new(),
             Err(error) => return Err(error),
         };
+        log_catalog_analysis(workspace_slug, agent_slug, "postgres", &tools);
         self.store_shared_snapshot(workspace_slug, agent_slug, &tools)
             .await;
         let mut guard = self.cached.write().await;
@@ -208,6 +212,42 @@ impl PublishedToolCatalog {
     }
 }
 
+fn log_catalog_analysis(
+    workspace_slug: &str,
+    agent_slug: &str,
+    source: &str,
+    tools: &[PublishedAgentTool],
+) {
+    let analysis = match analyze_published_tool_catalog(tools) {
+        Ok(analysis) => analysis,
+        Err(error) => {
+            warn!(workspace_slug, agent_slug, source, %error, "published catalog analysis failed");
+            return;
+        }
+    };
+    let warning_count = analysis
+        .quality
+        .findings
+        .iter()
+        .filter(|finding| finding.severity == crank_core::ToolQualitySeverity::Warning)
+        .count();
+
+    info!(
+        workspace_slug,
+        agent_slug,
+        source,
+        tool_count = analysis.budget.tool_count,
+        serialized_bytes = analysis.budget.serialized_bytes,
+        estimated_context_tokens = analysis.budget.estimated_context_tokens,
+        largest_tool_estimated_context_tokens =
+            analysis.budget.largest_tool_estimated_context_tokens,
+        recommended_context_tokens = analysis.budget.recommended_context_tokens,
+        exceeds_recommended_budget = analysis.budget.exceeds_recommended_budget,
+        catalog_quality_warning_count = warning_count,
+        "published agent catalog analyzed"
+    );
+}
+
 fn now_unix_ms() -> u64 {
     SystemTime::now()
         .duration_since(UNIX_EPOCH)

crates/crank-community-mcp/src/manifest.rs

@@ -1,4 +1,7 @@
-use crank_core::{HttpMethod, OperationSafetyClass, OperationSafetyPolicy, Target};
+use crank_core::{
+    HttpMethod, OperationSafetyClass, OperationSafetyPolicy, Target, ToolCatalogAnalysis,
+    ToolCatalogAnalysisError, ToolQualityCatalogTool, analyze_tool_catalog,
+};
 use crank_registry::PublishedAgentTool;
 use serde_json::{Value, json};
 
@@ -27,6 +30,37 @@ pub fn tool_definitions(tool: &PublishedAgentTool) -> Vec<Value> {
     )]
 }
 
+pub fn analyze_published_tool_catalog(
+    tools: &[PublishedAgentTool],
+) -> Result<ToolCatalogAnalysis, ToolCatalogAnalysisError> {
+    let mut quality_tools = Vec::new();
+    let mut definitions = Vec::new();
+
+    for tool in tools {
+        for definition in tool_definitions(tool) {
+            quality_tools.push(ToolQualityCatalogTool {
+                name: definition_string(&definition, "name")?,
+                display_name: definition_string(&definition, "title")?,
+                description: definition_string(&definition, "description")?,
+            });
+            definitions.push(definition);
+        }
+    }
+
+    analyze_tool_catalog(&quality_tools, &definitions)
+}
+
+fn definition_string(
+    definition: &Value,
+    field: &'static str,
+) -> Result<String, ToolCatalogAnalysisError> {
+    definition
+        .get(field)
+        .and_then(Value::as_str)
+        .map(ToOwned::to_owned)
+        .ok_or(ToolCatalogAnalysisError::DefinitionFieldMissing(field))
+}
+
 pub fn tool_definition(name: &str, title: &str, description: &str, input_schema: Value) -> Value {
     json!({
         "name": name,

crates/crank-community-mcp/tests/unit/manifest.rs

@@ -1,6 +1,6 @@
 use std::collections::BTreeMap;
 
-use crank_community_mcp::manifest::tool_definitions;
+use crank_community_mcp::manifest::{analyze_published_tool_catalog, tool_definitions};
 use crank_core::{
     ExecutionConfig, HttpMethod, Operation, OperationId, OperationSecurityLevel, OperationStatus,
     Protocol, RestTarget, Target, ToolDescription, WorkspaceId,
@@ -57,6 +57,20 @@ fn marks_destructive_tools_as_two_step_confirmation_calls() {
     assert_eq!(definition["inputSchema"]["required"], json!(["base"]));
 }
 
+#[test]
+fn catalog_budget_uses_the_same_definitions_as_tools_list() {
+    let tool = published_tool();
+    let definitions = tool_definitions(&tool);
+    let analysis = analyze_published_tool_catalog(&[tool]).unwrap();
+
+    assert_eq!(analysis.budget.tool_count, definitions.len());
+    assert_eq!(
+        analysis.budget.serialized_bytes,
+        serde_json::to_vec(&definitions[0]).unwrap().len()
+    );
+    assert!(!analysis.budget.exceeds_recommended_budget);
+}
+
 fn published_tool() -> PublishedAgentTool {
     PublishedAgentTool {
         workspace_id: WorkspaceId::new("ws_01"),