argus-cloud-optimizer 0.4.0__tar.gz → 0.4.1__tar.gz

{argus_cloud_optimizer-0.4.0/argus_cloud_optimizer.egg-info → argus_cloud_optimizer-0.4.1}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: argus-cloud-optimizer
-Version: 0.4.0
-Summary: AI-powered multi-cloud cost optimization agent
+Version: 0.4.1
+Summary: Argus — AI Cloud Detective for AWS, GCP, and Azure
 Author-email: Vamshi Siddarth Gaddam <vamshisiddarth02@gmail.com>
 License-Expression: MIT
 Project-URL: Homepage, https://github.com/vamshisiddarth/argus
@@ -62,12 +62,12 @@ Requires-Dist: mkdocs-minify-plugin>=0.8.0; extra == "dev"
 Dynamic: license-file
 
 <p align="center">
-  <img src="docs/assets/images/logo-full.svg" alt="Argus" height="72">
+  <img src="docs/assets/images/logo.svg" alt="Argus" height="72">
 </p>
 
-<p align="center"><strong>AI-powered cloud cost optimization agent for AWS, GCP, and Azure.</strong></p>
+<p align="center"><strong>Argus — AI Cloud Detective for AWS, GCP, and Azure.</strong></p>
 
-Argus finds idle and wasted cloud resources — stopped EC2 instances, unattached EBS volumes, orphaned Elastic IPs, underutilized RDS databases — and delivers a prioritized, AI-reasoned report to Slack every week.
+Argus is an open-source AI Cloud Detective — it hunts idle and wasted cloud resources across AWS, GCP, and Azure (stopped EC2 instances, unattached EBS volumes, orphaned Elastic IPs, underutilized RDS databases) and delivers a prioritized, AI-reasoned report to Slack every week. No rules. No thresholds. Just reasoning.
 
 [![CI](https://github.com/vamshisiddarth/argus/actions/workflows/ci.yml/badge.svg)](https://github.com/vamshisiddarth/argus/actions/workflows/ci.yml)
 [![Publish](https://github.com/vamshisiddarth/argus/actions/workflows/publish.yml/badge.svg)](https://github.com/vamshisiddarth/argus/actions/workflows/publish.yml)
@@ -100,6 +100,20 @@ Top findings
 
 ---
 
+## 🆕 What's new in v0.4.1
+
+> **Biggest release yet.** Argus jumps from scheduled reporter to a conversational FinOps companion.
+
+- **Interactive chat mode** — `argus chat` lets you ask natural-language questions about your cloud costs and get intelligent, context-aware answers backed by real metrics and cost data.
+- **True multi-cloud, multi-account** — GCP multi-project and Azure multi-subscription support with unified reporting across all accounts in a single run.
+- **Expanded resource coverage** — 31 GCP resource types (up from 22) and 40 Azure resource types (up from 26), all with curated metric mappings.
+- **Chat UX polish** — live spinner updates per tool call, AI-generated context summaries, turn-safe history trimming, and rich terminal formatting via `pip install argus-cloud-optimizer[chat]`.
+- **Docs overhaul** — new architecture diagram, corrected IAM permissions for all three clouds, and accurate env-var reference.
+
+[Full changelog →](CHANGELOG.md)
+
+---
+
 ## What it does
 
 Every week (or on demand), Argus:
@@ -166,13 +180,13 @@ docker run --rm \
 
 ```bash
 pip install argus-cloud-optimizer
-argus --version   # argus 0.3.0
+argus --version   # argus 0.4.1
 argus --help
 ```
 
 One package — all three clouds included. No extras needed. `--cloud` auto-detects from your environment (GCP_PROJECT_ID / AZURE_SUBSCRIPTION_IDS / AWS credentials), or specify explicitly.
 
-> **Verified 2026-06-21:** `pip install argus-cloud-optimizer && argus --version` works on a clean Python 3.11/3.12/3.13 venv with no project files.
+> **Verified 2026-06-24:** `pip install argus-cloud-optimizer && argus --version` works on a clean Python 3.11/3.12/3.13 venv with no project files.
 
 **AWS:**
 
@@ -231,7 +245,7 @@ argus chat                                       # auto-detects cloud
 ```
 
 ```
-Argus v0.3.0 — Interactive Cloud Cost Assistant
+Argus v0.4.1 — AI Cloud Detective
 Cloud: AWS  |  Accounts: prod (123456789012)  |  Budget: $1.00/session
 Tip: end a line with \ to continue on the next line.
 Type your question, or /help for commands.

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/README.md

@@ -1,10 +1,10 @@
 <p align="center">
-  <img src="docs/assets/images/logo-full.svg" alt="Argus" height="72">
+  <img src="docs/assets/images/logo.svg" alt="Argus" height="72">
 </p>
 
-<p align="center"><strong>AI-powered cloud cost optimization agent for AWS, GCP, and Azure.</strong></p>
+<p align="center"><strong>Argus — AI Cloud Detective for AWS, GCP, and Azure.</strong></p>
 
-Argus finds idle and wasted cloud resources — stopped EC2 instances, unattached EBS volumes, orphaned Elastic IPs, underutilized RDS databases — and delivers a prioritized, AI-reasoned report to Slack every week.
+Argus is an open-source AI Cloud Detective — it hunts idle and wasted cloud resources across AWS, GCP, and Azure (stopped EC2 instances, unattached EBS volumes, orphaned Elastic IPs, underutilized RDS databases) and delivers a prioritized, AI-reasoned report to Slack every week. No rules. No thresholds. Just reasoning.
 
 [![CI](https://github.com/vamshisiddarth/argus/actions/workflows/ci.yml/badge.svg)](https://github.com/vamshisiddarth/argus/actions/workflows/ci.yml)
 [![Publish](https://github.com/vamshisiddarth/argus/actions/workflows/publish.yml/badge.svg)](https://github.com/vamshisiddarth/argus/actions/workflows/publish.yml)
@@ -37,6 +37,20 @@ Top findings
 
 ---
 
+## 🆕 What's new in v0.4.1
+
+> **Biggest release yet.** Argus jumps from scheduled reporter to a conversational FinOps companion.
+
+- **Interactive chat mode** — `argus chat` lets you ask natural-language questions about your cloud costs and get intelligent, context-aware answers backed by real metrics and cost data.
+- **True multi-cloud, multi-account** — GCP multi-project and Azure multi-subscription support with unified reporting across all accounts in a single run.
+- **Expanded resource coverage** — 31 GCP resource types (up from 22) and 40 Azure resource types (up from 26), all with curated metric mappings.
+- **Chat UX polish** — live spinner updates per tool call, AI-generated context summaries, turn-safe history trimming, and rich terminal formatting via `pip install argus-cloud-optimizer[chat]`.
+- **Docs overhaul** — new architecture diagram, corrected IAM permissions for all three clouds, and accurate env-var reference.
+
+[Full changelog →](CHANGELOG.md)
+
+---
+
 ## What it does
 
 Every week (or on demand), Argus:
@@ -103,13 +117,13 @@ docker run --rm \
 
 ```bash
 pip install argus-cloud-optimizer
-argus --version   # argus 0.3.0
+argus --version   # argus 0.4.1
 argus --help
 ```
 
 One package — all three clouds included. No extras needed. `--cloud` auto-detects from your environment (GCP_PROJECT_ID / AZURE_SUBSCRIPTION_IDS / AWS credentials), or specify explicitly.
 
-> **Verified 2026-06-21:** `pip install argus-cloud-optimizer && argus --version` works on a clean Python 3.11/3.12/3.13 venv with no project files.
+> **Verified 2026-06-24:** `pip install argus-cloud-optimizer && argus --version` works on a clean Python 3.11/3.12/3.13 venv with no project files.
 
 **AWS:**
 
@@ -168,7 +182,7 @@ argus chat                                       # auto-detects cloud
 ```
 
 ```
-Argus v0.3.0 — Interactive Cloud Cost Assistant
+Argus v0.4.1 — AI Cloud Detective
 Cloud: AWS  |  Accounts: prod (123456789012)  |  Budget: $1.00/session
 Tip: end a line with \ to continue on the next line.
 Type your question, or /help for commands.

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/adapters/aws/adapter.py

@@ -82,4 +82,6 @@ class AWSAdapter(CloudAdapter):
     ) -> "AWSAdapter":
         """Convenience constructor — resolves auth and returns a ready adapter."""
         session = auth.get_session(account=account, region=region)
-        return cls(session=session, aggregator_region=region)
+        # Pass aggregator_region=None so __init__ honours RESOURCE_EXPLORER_REGION
+        # env var before falling back to the default "us-east-1".
+        return cls(session=session, aggregator_region=None)

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/adapters/gcp/adapter.py

@@ -33,12 +33,16 @@ class GCPAdapter(CloudAdapter):
             )
         self._project_id: str = resolved
         self._bq_billing_table = bq_billing_table or os.environ.get("BILLING_BQ_TABLE")
+        # Populated after list_resources() — asset types skipped due to disabled APIs
+        self.skipped_asset_types: list[str] = []
 
     def list_resources(self, ignore_regions: list[str] | None = None) -> list[Resource]:
-        return asset_inventory.list_resources(
+        resources, skipped = asset_inventory.list_resources(
             project_id=self._project_id,
             ignore_regions=ignore_regions,
         )
+        self.skipped_asset_types = skipped
+        return resources
 
     def get_metrics(
         self,

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/adapters/gcp/asset_inventory.py

@@ -3,7 +3,12 @@ from __future__ import annotations
 from typing import Any
 
 import structlog
-from google.api_core.exceptions import GoogleAPICallError, PermissionDenied
+from google.api_core.exceptions import (
+    GoogleAPICallError,
+    InvalidArgument,
+    NotFound,
+    PermissionDenied,
+)
 from google.cloud import asset_v1
 
 from adapters.base import Resource
@@ -32,7 +37,7 @@ SCANNED_ASSET_TYPES: list[str] = [
     "compute.googleapis.com/VpnTunnel",
     "vpcaccess.googleapis.com/Connector",  # Serverless VPC
     # --- Databases ---
-    "sql.googleapis.com/Instance",  # Cloud SQL
+    "sqladmin.googleapis.com/Instance",  # Cloud SQL
     "spanner.googleapis.com/Instance",
     "bigtable.googleapis.com/Instance",
     "alloydb.googleapis.com/Cluster",  # AlloyDB (managed Postgres)
@@ -55,42 +60,76 @@ SCANNED_ASSET_TYPES: list[str] = [
 def list_resources(
     project_id: str,
     ignore_regions: list[str] | None = None,
-) -> list[Resource]:
+) -> tuple[list[Resource], list[str]]:
     """
-    Return all billable GCP resources in a project using Cloud Asset Inventory.
-    Uses a single paginated API call — no per-resource-type enumeration needed.
+    Return (resources, skipped_asset_types) for a GCP project.
+
+    Uses a single paginated Cloud Asset Inventory call. If any asset type's API
+    is not enabled in the project, INVALID_ARGUMENT is returned for that type.
+    We strip it and retry so a project without Bigtable/Spanner/etc. enabled
+    doesn't block the whole scan. Callers receive the list of skipped types so
+    they can surface a warning in the Slack digest.
     """
     client = asset_v1.AssetServiceClient()
     parent = f"projects/{project_id}"
     ignore_set = set(ignore_regions or [])
     resources: list[Resource] = []
-
-    request = asset_v1.ListAssetsRequest(
-        parent=parent,
-        asset_types=SCANNED_ASSET_TYPES,
-        content_type=asset_v1.ContentType.RESOURCE,
-    )
-
-    try:
-        for asset in retry_on_transient(
-            client.list_assets, request=request, timeout=60
-        ):
-            parsed = _parse_asset(asset, ignore_set)
-            if parsed:
-                resources.append(parsed)
-    except PermissionDenied as exc:
-        raise PermissionError(
-            f"Argus service account is missing cloudasset.assets.listAssets "
-            f"permission on project {project_id}."
-        ) from exc
-    except GoogleAPICallError as exc:
-        raise RuntimeError(f"Cloud Asset Inventory API error: {exc}") from exc
+    asset_types = list(SCANNED_ASSET_TYPES)  # mutable — bad types get stripped
+    skipped_types: list[str] = []
+
+    while True:
+        request = asset_v1.ListAssetsRequest(
+            parent=parent,
+            asset_types=asset_types,
+            content_type=asset_v1.ContentType.RESOURCE,
+        )
+        try:
+            for asset in retry_on_transient(
+                client.list_assets, request=request, timeout=60
+            ):
+                parsed = _parse_asset(asset, ignore_set)
+                if parsed:
+                    resources.append(parsed)
+            break  # success — exit retry loop
+        except InvalidArgument as exc:
+            # An asset type whose API is not enabled in this project causes
+            # INVALID_ARGUMENT. Strip it and retry with the remaining types.
+            bad_type = _extract_bad_asset_type(str(exc), asset_types)
+            if bad_type:
+                logger.warning(
+                    "asset_type_api_not_enabled_skipping",
+                    asset_type=bad_type,
+                    project_id=project_id,
+                )
+                skipped_types.append(bad_type)
+                asset_types.remove(bad_type)
+                resources = []  # reset — avoid duplicates from partial page
+            else:
+                raise RuntimeError(
+                    f"Cloud Asset Inventory INVALID_ARGUMENT (unknown type): {exc}"
+                ) from exc
+        except (PermissionDenied, NotFound) as exc:
+            raise PermissionError(
+                f"Project '{project_id}' not found or Argus lacks permission. "
+                f"Check the project ID and ensure cloudasset.assets.listAssets "
+                f"is granted to your credentials."
+            ) from exc
+        except GoogleAPICallError as exc:
+            raise RuntimeError(f"Cloud Asset Inventory API error: {exc}") from exc
 
     logger.info(
         "asset_inventory_complete",
         extra={"project_id": project_id, "total": len(resources)},
     )
-    return resources
+    return resources, skipped_types
+
+
+def _extract_bad_asset_type(error_msg: str, asset_types: list[str]) -> str | None:
+    """Return the first asset type string found in an INVALID_ARGUMENT error."""
+    for asset_type in asset_types:
+        if asset_type in error_msg:
+            return asset_type
+    return None
 
 
 def _parse_asset(asset: Any, ignore_set: set[str]) -> Resource | None:

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/adapters/gcp/cloud_logging.py

@@ -52,7 +52,6 @@ def get_last_activity(
                 filter_=log_filter,
                 order_by=gcp_logging.DESCENDING,
                 page_size=1,
-                timeout=60,
             )
         )
     except GoogleAPICallError as exc:
@@ -76,7 +75,7 @@ def _service_from_resource_type(resource_type: str) -> str | None:
     mapping: dict[str, str] = {
         "compute.googleapis.com/Instance": "compute.googleapis.com",
         "compute.googleapis.com/Disk": "compute.googleapis.com",
-        "sql.googleapis.com/Instance": "cloudsql.googleapis.com",
+        "sqladmin.googleapis.com/Instance": "cloudsql.googleapis.com",
         "container.googleapis.com/Cluster": "container.googleapis.com",
         "storage.googleapis.com/Bucket": "storage.googleapis.com",
         "bigquery.googleapis.com/Dataset": "bigquery.googleapis.com",

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/adapters/gcp/cloud_monitoring.py

@@ -26,7 +26,7 @@ _METRICS: dict[str, list[tuple[str, str]]] = {
         ("compute.googleapis.com/instance/disk/read_ops_count", "sum"),
         ("compute.googleapis.com/instance/disk/write_ops_count", "sum"),
     ],
-    "sql.googleapis.com/Instance": [
+    "sqladmin.googleapis.com/Instance": [
         ("cloudsql.googleapis.com/database/cpu/utilization", "mean"),
         ("cloudsql.googleapis.com/database/network/connections", "mean"),
         ("cloudsql.googleapis.com/database/network/received_bytes_count", "sum"),
@@ -394,7 +394,7 @@ def _resource_filter(resource_id: str, resource_type: str) -> str:
     match resource_type:
         case "compute.googleapis.com/Instance":
             return f'resource.labels.instance_id="{name}"'
-        case "sql.googleapis.com/Instance":
+        case "sqladmin.googleapis.com/Instance":
             return f'resource.labels.database_id="{name}"'
         case "container.googleapis.com/Cluster":
             return f'resource.labels.cluster_name="{name}"'

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/ai/azure_openai.py

@@ -128,6 +128,22 @@ class AzureOpenAIProvider(AIProvider):
         for attempt in range(MAX_RETRIES):
             try:
                 return self._client.chat.completions.create(**kwargs)
+            except openai.BadRequestError as exc:
+                # Reasoning models (o1/o3/o4) have two restrictions vs standard models:
+                #   1. Require max_completion_tokens instead of max_tokens.
+                #   2. Do not accept temperature (fixed at 1).
+                # Catch both in one retry so we don't need two round-trips.
+                err_msg = str(exc)
+                changed = False
+                if "max_completion_tokens" in err_msg and "max_tokens" in kwargs:
+                    kwargs["max_completion_tokens"] = kwargs.pop("max_tokens")
+                    changed = True
+                if "temperature" in err_msg and "temperature" in kwargs:
+                    kwargs.pop("temperature")
+                    changed = True
+                if changed:
+                    continue
+                raise
             except openai.RateLimitError:
                 if attempt < MAX_RETRIES - 1:
                     logger.warning(

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/ai/vertexai.py

@@ -9,7 +9,9 @@ No API key needed when running on Cloud Run / GCE with the right service account
 Environment variables:
     VERTEXAI_PROJECT   GCP project ID (required)
     VERTEXAI_LOCATION  GCP region (default: us-central1)
-    VERTEXAI_MODEL     Model name (default: gemini-1.5-pro-002)
+    VERTEXAI_MODEL     Model name (default: google/gemini-1.5-pro-002)
+                       Must include publisher prefix for the OpenAI-compat endpoint
+                       (e.g. google/gemini-1.5-pro-002, google/gemini-2.0-flash-001)
 """
 
 from __future__ import annotations
@@ -38,10 +40,11 @@ class VertexAIProvider(AIProvider):
     or directly via vertexai package. Falls back to the OpenAI-compatible
     Vertex AI endpoint so we can reuse the openai SDK already in requirements.txt.
 
-    Model: gemini-1.5-pro-002 (default) — supports function calling + large context.
+    Model: google/gemini-1.5-pro-002 (default) — the publisher prefix is required
+    by the OpenAI-compatible Vertex AI endpoint (/endpoints/openapi).
     """
 
-    DEFAULT_MODEL = "gemini-1.5-pro-002"
+    DEFAULT_MODEL = "google/gemini-1.5-pro-002"
     DEFAULT_LOCATION = "us-central1"
     DEFAULT_MAX_TOKENS = 4096
     DEFAULT_TEMPERATURE = 0.0

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1/argus_cloud_optimizer.egg-info}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: argus-cloud-optimizer
-Version: 0.4.0
-Summary: AI-powered multi-cloud cost optimization agent
+Version: 0.4.1
+Summary: Argus — AI Cloud Detective for AWS, GCP, and Azure
 Author-email: Vamshi Siddarth Gaddam <vamshisiddarth02@gmail.com>
 License-Expression: MIT
 Project-URL: Homepage, https://github.com/vamshisiddarth/argus
@@ -62,12 +62,12 @@ Requires-Dist: mkdocs-minify-plugin>=0.8.0; extra == "dev"
 Dynamic: license-file
 
 <p align="center">
-  <img src="docs/assets/images/logo-full.svg" alt="Argus" height="72">
+  <img src="docs/assets/images/logo.svg" alt="Argus" height="72">
 </p>
 
-<p align="center"><strong>AI-powered cloud cost optimization agent for AWS, GCP, and Azure.</strong></p>
+<p align="center"><strong>Argus — AI Cloud Detective for AWS, GCP, and Azure.</strong></p>
 
-Argus finds idle and wasted cloud resources — stopped EC2 instances, unattached EBS volumes, orphaned Elastic IPs, underutilized RDS databases — and delivers a prioritized, AI-reasoned report to Slack every week.
+Argus is an open-source AI Cloud Detective — it hunts idle and wasted cloud resources across AWS, GCP, and Azure (stopped EC2 instances, unattached EBS volumes, orphaned Elastic IPs, underutilized RDS databases) and delivers a prioritized, AI-reasoned report to Slack every week. No rules. No thresholds. Just reasoning.
 
 [![CI](https://github.com/vamshisiddarth/argus/actions/workflows/ci.yml/badge.svg)](https://github.com/vamshisiddarth/argus/actions/workflows/ci.yml)
 [![Publish](https://github.com/vamshisiddarth/argus/actions/workflows/publish.yml/badge.svg)](https://github.com/vamshisiddarth/argus/actions/workflows/publish.yml)
@@ -100,6 +100,20 @@ Top findings
 
 ---
 
+## 🆕 What's new in v0.4.1
+
+> **Biggest release yet.** Argus jumps from scheduled reporter to a conversational FinOps companion.
+
+- **Interactive chat mode** — `argus chat` lets you ask natural-language questions about your cloud costs and get intelligent, context-aware answers backed by real metrics and cost data.
+- **True multi-cloud, multi-account** — GCP multi-project and Azure multi-subscription support with unified reporting across all accounts in a single run.
+- **Expanded resource coverage** — 31 GCP resource types (up from 22) and 40 Azure resource types (up from 26), all with curated metric mappings.
+- **Chat UX polish** — live spinner updates per tool call, AI-generated context summaries, turn-safe history trimming, and rich terminal formatting via `pip install argus-cloud-optimizer[chat]`.
+- **Docs overhaul** — new architecture diagram, corrected IAM permissions for all three clouds, and accurate env-var reference.
+
+[Full changelog →](CHANGELOG.md)
+
+---
+
 ## What it does
 
 Every week (or on demand), Argus:
@@ -166,13 +180,13 @@ docker run --rm \
 
 ```bash
 pip install argus-cloud-optimizer
-argus --version   # argus 0.3.0
+argus --version   # argus 0.4.1
 argus --help
 ```
 
 One package — all three clouds included. No extras needed. `--cloud` auto-detects from your environment (GCP_PROJECT_ID / AZURE_SUBSCRIPTION_IDS / AWS credentials), or specify explicitly.
 
-> **Verified 2026-06-21:** `pip install argus-cloud-optimizer && argus --version` works on a clean Python 3.11/3.12/3.13 venv with no project files.
+> **Verified 2026-06-24:** `pip install argus-cloud-optimizer && argus --version` works on a clean Python 3.11/3.12/3.13 venv with no project files.
 
 **AWS:**
 
@@ -231,7 +245,7 @@ argus chat                                       # auto-detects cloud
 ```
 
 ```
-Argus v0.3.0 — Interactive Cloud Cost Assistant
+Argus v0.4.1 — AI Cloud Detective
 Cloud: AWS  |  Accounts: prod (123456789012)  |  Budget: $1.00/session
 Tip: end a line with \ to continue on the next line.
 Type your question, or /help for commands.

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/argus_cloud_optimizer.egg-info/SOURCES.txt

@@ -68,6 +68,7 @@ deploy/aws/single-account/template.yaml
 deploy/azure/function-app.bicep
 deploy/gcp/deploy.sh
 docs/index.md
+docs/roadmap.md
 docs/concepts/adapters.md
 docs/concepts/ai-providers.md
 docs/concepts/ai-reasoning.md

argus_cloud_optimizer-0.4.1/core/__version__.py

@@ -0,0 +1 @@
+__version__ = "0.4.1"

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/core/agent/loop.py

@@ -195,6 +195,7 @@ class AgentLoop:
         resources = self._adapter.list_resources(ignore_regions=ignore_regions or None)
         resources = _apply_exclusion_filters(resources)
         total_discovered = len(resources)
+        logger.info("resources_discovered", count=total_discovered)
 
         if not resources:
             self._prefiltered_payload = []
@@ -242,6 +243,11 @@ class AgentLoop:
             discovered=total_discovered,
             sent_to_ai=len(payload),
         )
+        logger.info(
+            "ai_analysis_starting",
+            resources=len(payload),
+            msg=f"Analyzing {len(payload)} resources with AI (this takes 2-5 minutes)…",
+        )
         return payload
 
     def _execute(self, tc: ToolCall) -> tuple[str, bool]:

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/core/agent/prompts.py

@@ -84,7 +84,7 @@ def build_system_prompt(
     """
     context = _build_context_header(cloud, ignore_regions, accounts)
 
-    return f"""You are Argus, an intelligent cloud cost optimization agent.
+    return f"""You are Argus, an AI Cloud Detective.
 
 MISSION
 ───────

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/core/config.py

@@ -48,7 +48,7 @@ class AISettings(BaseSettings):
     # Vertex AI
     vertexai_project: str | None = Field(None, alias="VERTEXAI_PROJECT")
     vertexai_location: str = Field("us-central1", alias="VERTEXAI_LOCATION")
-    vertexai_model: str = Field("gemini-1.5-pro-002", alias="VERTEXAI_MODEL")
+    vertexai_model: str = Field("google/gemini-1.5-pro-002", alias="VERTEXAI_MODEL")
 
     # Azure OpenAI
     azure_openai_endpoint: str | None = Field(None, alias="AZURE_OPENAI_ENDPOINT")

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/core/reports/delivery.py

@@ -219,7 +219,14 @@ def build_notification_providers() -> list[NotificationProvider]:
     return providers
 
 
-def notify_all(payload: dict[str, Any], dry_run: bool | None = None) -> None:
+def notify_all(payload: dict[str, Any], dry_run: bool | None = None) -> bool:
+    """
+    Send payload to all configured notification providers.
+
+    Returns True if at least one provider delivered successfully, False if
+    every provider failed. Callers should treat False as a delivery failure
+    and exit non-zero so the scheduler / CloudWatch alarm fires.
+    """
     from core.config import get_settings
 
     resolved_dry_run = dry_run if dry_run is not None else get_settings().report.dry_run
@@ -229,12 +236,14 @@ def notify_all(payload: dict[str, Any], dry_run: bool | None = None) -> None:
             "dry_run_notification_skipped",
             payload_preview=json.dumps(payload, indent=2)[:500],
         )
-        return
+        return True
 
     providers = build_notification_providers()
+    any_succeeded = False
     for provider in providers:
         try:
             provider.notify(payload)
+            any_succeeded = True
         except (NotificationDeliveryError, OSError) as exc:
             logger.error(
                 "notification_delivery_failed",
@@ -242,6 +251,11 @@ def notify_all(payload: dict[str, Any], dry_run: bool | None = None) -> None:
                 error=str(exc),
             )
 
+    if not any_succeeded and providers:
+        logger.error("all_notification_providers_failed", provider_count=len(providers))
+
+    return any_succeeded or not providers
+
 
 # ---------------------------------------------------------------------------
 # Legacy function — delegates to notify_all for backward compatibility

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/core/reports/generator.py

@@ -23,6 +23,7 @@ def build_report(
     agent_output_tokens: int = 0,
     scan_diff: dict[str, Any] | None = None,
     scan_errors: list[dict[str, str]] | None = None,
+    skipped_resource_types: list[str] | None = None,
 ) -> dict[str, Any]:
     """
     Convert a list of ResourceFinding objects into the canonical JSON report.
@@ -53,6 +54,7 @@ def build_report(
         ),
         "scan_diff": scan_diff,
         "scan_errors": scan_errors or [],
+        "skipped_resource_types": skipped_resource_types or [],
     }
 
 
@@ -207,6 +209,25 @@ def build_slack_payload(
         {"type": "divider"},
     ]
 
+    skipped_types = report.get("skipped_resource_types") or []
+    if skipped_types:
+        short_names = [t.split("/")[-1] for t in skipped_types]
+        blocks.append(
+            {
+                "type": "section",
+                "text": {
+                    "type": "mrkdwn",
+                    "text": (
+                        f":information_source: *Skipped {len(skipped_types)} resource "
+                        f"type{'s' if len(skipped_types) != 1 else ''} "
+                        f"(API not enabled in project):* "
+                        + ", ".join(f"`{n}`" for n in short_names)
+                    ),
+                },
+            }
+        )
+        blocks.append({"type": "divider"})
+
     scan_errors = report.get("scan_errors") or []
     if scan_errors:
         total_attempted = accounts + len(scan_errors)

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/deploy/aws/multi-account/spoke-role.yaml

@@ -66,6 +66,7 @@ Resources:
                 Effect: Allow
                 Action:
                   - cloudwatch:GetMetricData
+                  - cloudwatch:ListMetrics
                 Resource: "*"
 
               # Cost Explorer — cost data (read-only, per-resource billing breakdown)

{argus_cloud_optimizer-0.4.0 → argus_cloud_optimizer-0.4.1}/deploy/aws/single-account/template.yaml

@@ -150,6 +150,7 @@ Resources:
                 Effect: Allow
                 Action:
                   - cloudwatch:GetMetricData
+                  - cloudwatch:ListMetrics
                 Resource: "*"
 
               - Sid: CostExplorer