solace-developer-workshops

Author SAM declarative-config repos (manifests, per-kind YAML, skills) and drive sam config plan/apply/pull/migrate. Auto-generated from the live DTO + type schemas at the bundled CLI version; re-run `sam ai-assistance skill install --force` after a CLI upgrade.

Official Solace Agent Mesh documentation — covers architecture, components, configuration, development, deployment, tutorials, and enterprise features.

Use when creating or editing Solace Agent Mesh agents or workflows — writing instructions/system prompts, enabling built-in tool groups, giving an agent an MCP server's tools, peer delegation between agents (inter_agent_communication), structured input/output schemas, agent cards/abilities, or multi-step workflow DAGs (switch/map/loop nodes). Not for creating connectors (sam-connectors), custom tools (sam-tools-and-skills), inbound gateways (sam-gateways), or choosing where SAM runs (sam-install-run / sam-deploy).

Use when someone describes an outcome they want from Solace Agent Mesh in their own words rather than SAM vocabulary — "I want an assistant that…", "search my data in Salesforce / our database", "a workflow that listens to events, filters them, and stores results", "let my team talk to it from Slack", "just let me try it", "now run it for the whole team" — or is unsure where some information should live ("should this go in the agent or a tool?", "where do I put my documents?", "how does it remember a user between chats?") — or when it is unclear which sam-* skill serves a SAM request. Not for troubleshooting symptoms (something is broken/failing) or for users already naming a specific SAM construct.

Use when giving a Solace Agent Mesh agent access to external data or services without writing code — query a SQL database (PostgreSQL, MySQL, MariaDB, SQL Server, Oracle), retrieve from a knowledge base (Amazon Bedrock RAG), search Elasticsearch/OpenSearch, look up MongoDB/DynamoDB documents or Neo4j/Neptune graphs, call a REST API from an OpenAPI spec, use a remote MCP server's tools, send email or Slack messages, or publish to the Solace event mesh — and when creating, editing, or attaching connectors. Not for inbound integrations where users or events reach agents (sam-gateways), custom Go/Python tool code or toolsets (sam-tools-and-skills), or authoring the agent itself (sam-author-agent).

Use when deploying Solace Agent Mesh for shared or managed use — Helm charts on minikube (local rehearsal) or EKS/GKE/AKS (cloud), choosing chart values (broker, persistence, images, LLM, OIDC/RBAC, ingress), promoting a local/desktop instance to a cluster, or running SAM in Docker for a team. Not for single-user local trial (sam-install-run), authoring agents/config content (sam-author-agent / sam-declarative-config), or diagnosing an already-broken deployment (sam-troubleshoot / sam-operate).

Use when making Solace Agent Mesh agents reachable from outside — chatting with agents from Slack or Microsoft Teams, exposing agents as MCP tools to Claude Code/Cursor/other MCP clients, triggering agents or workflows from Solace event mesh topics, email- or WhatsApp-driven tasks, or securing a gateway with OAuth/OIDC (Keycloak, dynamic client registration, PKCE) — and when creating, deploying, or troubleshooting gateways. Not for the agent reaching OUT to external systems (sam-connectors no-code, sam-tools-and-skills custom code), authoring the agent itself (sam-author-agent), or the built-in web chat UI, which needs no gateway setup.

Use when someone wants to install or try Solace Agent Mesh on their own machine — downloading the desktop app or sam CLI, first chat with the built-in agents, running locally via Docker or `sam run --embedded`, or pre-flight checks for a local trial. Not for shared/team deployments (sam-deploy), building agents beyond the first chat (sam-author-agent), or fixing a broken installation (sam-troubleshoot).

Use when operating a deployed Solace Agent Mesh instance — managing the runtime lifecycle of running components (list, start, stop, restart, or remove a specific running agent or workflow with `sam control`), turning on SSO/OIDC login for the gateway (Keycloak, Azure AD, Okta), configuring RBAC roles and scopes (including which users can use an agent or manage connectors), managing secrets and promoting config across dev/staging/prod, or upgrading SAM to a new version. Not for diagnosing why something failed or reading task event logs (sam-troubleshoot), authoring agents/tools/gateways/connectors (those skills), or first-time install (sam-install-run / sam-deploy).

Use when a user wants an assistant to answer from their OWN uploaded documents with no code — "upload my PDFs/policies so the assistant can answer from them", creating or setting up a Project, uploading files to a project, "why isn't my project using my documents / the assistant ignores my files", the project Instructions field or default-agent settings, or choosing between a Project and a knowledge_base connector. Covers the WebUI Projects feature (the document workspace), NOT the `sam config` declarative-config directory (also historically called a "project").

Use when extending a Solace Agent Mesh agent with custom code or curated tool packages — writing a custom tool in Go (samtoolsdk) or Python (sam-tool-sdk), scaffolding with the `sam toolset` / `sam skill` CLI, packaging tools as a toolset, creating an agent skill bundle (SKILL.md + bundled tools), exposing a REST API to one agent via an inline OpenAPI tool, attaching toolsets/skills to agents, or setting per-agent tool config. Not for no-code data/service access where a connector type fits (sam-connectors), inbound integrations where users or events reach agents (sam-gateways), or the agent's own instructions and behavior (sam-author-agent).

Use when something in Solace Agent Mesh is broken or misbehaving and you need to diagnose it — an agent won't start or won't respond, the gateway returns 500s, a deploy times out, tasks hang or produce no output, a tool/connector call fails, broker or LLM connectivity is suspect, or you need to reconstruct exactly what happened during a specific task (the task event log / STIM file). Covers logs, traceID correlation, the sam-doctor pre-flight checks, health endpoints, and per-task event logs. Not for authoring agents/tools/gateways/connectors (those skills), nor for operator configuration like SSO/RBAC/secrets/upgrades (sam-operate).