strands-hub-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@strands-hub-mcp list all agents"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
strands-hub-mcp
MCP server that exposes strands-hub data from S3:
Registry - list/update agents
Prompts - get/create prompt versions
Sessions - browse session data (read-only)
Metrics - query run metrics (read-only)
Prerequisites
Python 3.10+
AWS credentials configured (
~/.aws/credentialsor env vars)S3 bucket with strands-hub data
Related MCP server: s3-mcp
Claude Code Integration
claude mcp add strands-hub-mcp --scope user \
-e USE_S3=true \
-e AGENT_HUB_BUCKET=your-bucket-name \
-e AGENT_HUB_REGION=us-east-1 \
-- uv --directory /path/to/strands-hub-mcp run strands-hub-mcpVerify:
claude mcp listAvailable Tools
Tool | Description |
| Show current hub configuration |
| List all registered agents |
| Get agent details |
| Update agent metadata |
| Get current prompt for an agent |
| Get specific prompt version |
| List all prompt versions |
| Create new prompt version (append-only) |
| List metrics by date/agent |
| Get specific metrics record |
| List all sessions |
| Get session metadata |
| List agents in a session |
| Get agent data from session |
| List messages in a session |
| Get specific message |
Environment Variables
Variable | Required | Description |
| Yes | Must be |
| Yes | S3 bucket name |
| Yes | AWS region |
Author
Built by Du'An Lightfoot (@labeveryday)
Available Tools
17 toolshub_statusA
Return the effective hub configuration (sanitized) used by this server.
Inputs: no named parameters.
metrics_getA
Fetch a metrics JSON object by S3 key.
Inputs: s3_key (required).
metrics_listC
List metrics objects in S3.
Inputs: limit, agent_id, date_prefix, continuation_token.
prompts_create_versionA
Create a new prompt version (append-only). Does NOT modify current.txt.
Fails if the version already exists.
Updates versions.json without changing its current pointer.
Inputs: content (required), version (required), agent_id (required), note.
prompts_get_currentC
Get the current system prompt content for an agent (cached locally by strands-hub).
Inputs: agent_id (required), force_refresh. Returns: result
prompts_get_versionC
Get a specific system prompt version content for an agent.
Inputs: version (required), agent_id (required). Returns: result
prompts_list_versionsB
List prompt versions for an agent from S3 (versions.json if present).
Inputs: agent_id (required).
registry_get_agentB
Get a single agent registry entry.
Inputs: agent_id (required). Returns: result
registry_list_agentsA
List registered agents (optionally filtered by tag).
Inputs: tag. Returns: result
registry_update_metadataB
Update allowlisted metadata fields for an existing agent.
Inputs: agent_id (required), tags, owner, model_id, +3 more. Returns: result
sessions_get_agent_jsonB
Fetch and parse sessions/<session_id>/agents/<agent_id>/agent.json.
Inputs: session_id (required), agent_id.
sessions_get_message_jsonC
Fetch and parse a message JSON object.
message_key may be either a basename like message_0.json or a full S3 key.
Inputs: session_id (required), message_key (required), agent_id.
sessions_get_rawA
Fetch a session object by S3 key and return parsed JSON when possible.
This is intentionally "raw" because session format is owned by strands-agents.
Inputs: s3_key (required).
sessions_get_session_jsonA
Fetch and parse sessions/<session_id>/session.json.
Inputs: session_id (required).
sessions_listC
List session IDs (prefixes) under sessions/.
Uses S3 CommonPrefixes (Delimiter="/") so it matches aws s3 ls s3://.../sessions/.
Inputs: limit, continuation_token.
sessions_list_agentsB
List agent IDs under sessions/<session_id>/agents/.
Inputs: session_id (required), limit, continuation_token.
sessions_list_messagesB
List message keys under sessions/<session_id>/agents/<agent_id>/messages/.
Inputs: session_id (required), limit, agent_id, continuation_token.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 17 tools
Most tools map to distinct resources and actions, with clear domain prefixes (registry, prompts, metrics, sessions). However, sessions_get_session_json and sessions_get_raw both fetch session data and could be confused, as could registry_get_agent vs sessions_get_agent_json despite their different contexts.
Naming follows a consistent <domain>_<verb>_<object> pattern for nearly all tools, such as registry_list_agents and sessions_get_message_json. The main deviation is hub_status, which lacks a verb unlike the rest, and sessions_get_raw uses an adjective instead of a specific object.
17 tools is slightly above the typical well-scoped range, but the count is justified by the five distinct domains and the deep session hierarchy. Each tool serves a specific retrieval task, making the count reasonable rather than bloated.
The read-focused surface is fairly complete for inspection: sessions are traversable from ID down to message content, prompts have version listing/creation, and metrics are accessible. Minor gaps include no delete/create in the registry and no write operations for sessions, but these may be intentionally outside the server's scope.
Maintenance
Related MCP Connectors
An MCP server for the STRG.com s3-compatible object storage network for all management operations. Full documentation access, ask questions, create buckets, configure scoped access keys, CORS rules, lifecycle policies and perform any S3 operation.
MCP server for building and testing AI agents with multi-model experimentation and insights.
An MCP server that provides an API to LLMs to manage their JumpCloud resources.
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
Related MCP Servers
AlicenseNot gradedqualityDmaintenanceMCP server that exposes Sceptre CloudFormation management operations as tools for AI agents.Apache 2.0- AlicenseNot gradedqualityDmaintenanceMCP server for AWS S3 and compatible object stores, enabling LLM clients to browse buckets and fetch objects.MIT No Attribution

AgentMark MCP Serverofficial
AlicenseNot gradedqualityAmaintenanceMCP server for AgentMark. Exposes the full AgentMark API to AI editors: list traces and spans, manage datasets, write scores, run experiments, manage apps, deployments, environments, and annotation queues.352AGPL 3.0- AlicenseAqualityCmaintenanceMCP server that exposes a pool of Claude SDK agents as tools, enabling deterministic agent creation, retrieval, and resumption from a capability corpus.10112 PyPI1MIT