bc-telemetry-buddy-mcp 3.1.0 → 3.1.2

package/CHANGELOG.md

@@ -7,6 +7,25 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [3.1.2] - 2026-03-02
+
+### Added
+- **Retry Logic with Exponential Backoff**: LLM API calls now automatically retry on transient errors (HTTP 429, 529, 503) with configurable exponential backoff. Default: 10 retries, 2s initial delay, 2x multiplier, 60s max delay. Configurable via `retry` section in agent config.
+
+### Improved
+- **Dynamic Pipeline Templates**: `generateAzureDevOpsYaml()` and `generateGitHubActionsYaml()` now accept `PipelineOptions` (LLM provider, branch name, variable group name) for customizable CI/CD pipeline generation
+- **Azure DevOps Pipeline Template**: Updated to use inline `export` for npm prefix/PATH (self-hosted agent compatibility), `BCTB_WORKSPACE_PATH`, branch-based triggers, and dual LLM key support (Azure OpenAI / Anthropic)
+- **Template READMEs**: Added sections for custom branch, self-hosted agents, Anthropic usage, and new troubleshooting entries (EACCES, git push)
+
+### Changed
+- **Agentic Monitoring marked as Preview**: All user-facing references now include "(Preview)" label with disclaimer notes in README
+
+## [3.1.1] - 2025-06-04
+
+### Improved
+- **Pipeline Templates — Slimmed Secrets**: GitHub Actions and Azure DevOps templates now list only 3 required secrets (`BCTB_CLIENT_ID`, `BCTB_CLIENT_SECRET`, LLM API key) plus commented-out action secrets. Non-sensitive config values (tenant ID, App Insights ID, Kusto URL, LLM endpoint/deployment) are read from `.bctb-config.json` at runtime.
+- **Template READMEs — Split Secrets Tables**: Secrets tables split into "Always Required" and "Only If You Configured These Actions" sections with clear provenance notes
+
 ## [3.1.0] - 2026-02-26
 
 ### Added

package/README.md

@@ -15,7 +15,7 @@ BC Telemetry Buddy MCP Server is a **standalone NPM package** that enables AI as
 - 🔧 **CLI Interface**: `bctb-mcp` command with init, validate, test-auth, start, and **agent** subcommands
 - 📁 **File-Based Config**: Simple `.bctb-config.json` with schema validation and environment variable substitution
 - 👥 **Multi-Profile Support**: Manage multiple customers/environments in single config file with profile switching
-- 🤖 **Agentic Monitoring**: Autonomous scheduled telemetry monitoring via CI/CD pipelines — LLM-powered reasoning, issue tracking, Teams/email alerts
+- 🤖 **Agentic Monitoring (Preview)**: Autonomous scheduled telemetry monitoring via CI/CD pipelines — LLM-powered reasoning, issue tracking, Teams/email alerts
 - 🔌 **Optional for VSCode**: VSCode extension works standalone; MCP only required for chat participant features
 - 🔄 **Automatic Updates**: Update notifications on startup when newer versions are available on NPM
 - 🧪 **Comprehensive Testing**: 70%+ test coverage with dedicated test suites for Claude Desktop workflows
@@ -264,7 +264,9 @@ bctb-mcp --version
 bctb-mcp --help
 ```
 
-## Agentic Monitoring
+## Agentic Monitoring (Preview)
+
+> **Note:** Agentic Monitoring is currently a preview feature. APIs and behavior may change in future releases.
 
 BC Telemetry Buddy MCP includes a built-in **autonomous agent runtime** for scheduled telemetry monitoring. Agents use an LLM (Azure OpenAI or Anthropic) to query telemetry, reason about findings, track issues across runs, and take action — all without manual intervention.
 

package/config-schema.json

@@ -170,6 +170,32 @@
                             ],
                             "description": "Tool access scope for agents. 'read-only' prevents save_query and other write tools.",
                             "default": "read-only"
+                        },
+                        "retry": {
+                            "type": "object",
+                            "description": "Retry configuration for transient LLM API errors (429, 529, 503). Uses exponential backoff.",
+                            "properties": {
+                                "maxRetries": {
+                                    "type": "number",
+                                    "description": "Maximum number of retry attempts before failing.",
+                                    "default": 10
+                                },
+                                "initialDelayMs": {
+                                    "type": "number",
+                                    "description": "Initial delay in milliseconds before the first retry.",
+                                    "default": 2000
+                                },
+                                "backoffMultiplier": {
+                                    "type": "number",
+                                    "description": "Multiplier applied to the delay after each retry attempt.",
+                                    "default": 2
+                                },
+                                "maxDelayMs": {
+                                    "type": "number",
+                                    "description": "Maximum delay in milliseconds between retries (cap).",
+                                    "default": 60000
+                                }
+                            }
                         }
                     }
                 },