concierge

Name: concierge
Author: concierge-hq

113installs/mo · pypi533stars97forks

How to install concierge?

concierge-hq/concierge

$git clone https://github.com/concierge-hq/concierge

Installs into the current project.

Files · 1

View on GitHub

README.md

1	<div align="center">
2
3	<!-- omit in toc -->
4
5	<picture>
6	<img width="900" alt="Concierge Banner" src="assets/concierge-banner.png" />
7	</picture>
8
9	# Concierge AI 🚀
10
11	<strong>The fabric for reliable MCP servers and AI applications.</strong>
12
13	[![Docs](https://img.shields.io/badge/docs-getconcierge.app-blue)](https://docs.getconcierge.app)
14	[![Discord](https://img.shields.io/badge/community-discord-5865F2?logo=discord&logoColor=white)](https://discord.gg/Y3ayRa33Pg)
15	[![PyPI - Version](https://img.shields.io/pypi/v/concierge-sdk.svg)](https://pypi.org/project/concierge-sdk)
16	[![Python](https://img.shields.io/badge/python-3.9+-8B5CF6?logo=python&logoColor=white)](https://pypi.org/project/concierge-sdk)
17
18	</div>
19
20	The [Model Context Protocol](https://modelcontextprotocol.io) (MCP) is a standardized way to connect AI agents to tools. Instead of exposing a flat list of every tool on every request, Concierge progressively discloses only what's relevant. Concierge guarantees deterministic results and reliable tool invocation.
21
22	## Getting Started
23
24	> [!NOTE]
25	> Concierge requires Python 3.9+. We recommend installing with [uv](https://docs.astral.sh/uv/) for faster dependency resolution, but pip works just as well.
26
27	```bash
28	pip install concierge-sdk
29	```
30
31	Scaffold a new project:
32
33	```bash
34	concierge init my-store # Generate a ready to run project
35	cd my-store # Enter project
36	python main.py # Start the MCP server
37	```
38
39	Or wrap an existing MCP server two lines, nothing else changes:
40
41	```python
42	# Before
43	from mcp.server.fastmcp import FastMCP
44	app = FastMCP("my-server")
45
46	# After: just wrap it
47	from concierge import Concierge
48	app = Concierge(FastMCP("my-server"))
49	```
50
51	> [!TIP]
52	> Concierge works at the MCP protocol level. It dynamically changes which tools are returned by `tools/list` based on the current workflow step. The agent and client don't need to know Concierge exists, they just see fewer, more relevant tools at each point.
53
54	<br />
55
56	```python
57	from concierge import Concierge
58	from mcp.server.fastmcp import FastMCP
59
60	app = Concierge(FastMCP("my-server"))
61
62	# Your @app.tool() decorators stay exactly the same.
63	# You can additionally add app.stages and app.transitions.
64	```
65
66	> [!NOTE]
67	> The wrap and go gives you progressive tool disclosure immediately. Add `app.stages` and `app.transitions` when you want full workflow control, no code changes required.
68
69	<br />
70
71	## Usage
72
73	### Group tools into steps
74
75	Instead of exposing everything at once, group related tools together. Only the current step's tools are visible to the agent:
76
77	```python
78	app.stages = {
79	"browse": ["search_products", "view_product"],
80	"cart": ["add_to_cart", "remove_from_cart", "view_cart"],
81	"checkout": ["apply_coupon", "complete_purchase"],
82	}
83	```
84
85	### Define transitions
86
87	Control which steps can follow which. The agent moves forward (or backward) only along paths you allow:
88
89	```python
90	app.transitions = {
91	"browse": ["cart"], # Can only move to cart
92	"cart": ["browse", "checkout"], # Can go back or proceed
93	"checkout": [], # Terminal step
94	}
95	```
96
97	<details>
98	<summary><b>Share state between steps</b></summary>
99	<br>
100
101	Pass data between workflow steps without round-tripping through the LLM. State is session-scoped and works across distributed replicas:
102
103	```python
104	# In the "browse" step - save a selection
105	app.set_state("selected_product", {"id": "p1", "name": "Laptop"})
106
107	# In the "cart" step retrieve it directly
108	product = app.get_state("selected_product")
109	```
110	</details>
111
112	<details>
113	<summary><b>Scale with semantic search</b></summary>
114	<br>
115
116	When you have hundreds of tools, enable semantic search to collapse your entire API behind two meta-tools:
117
118	```python
119	from concierge import Concierge, Config, ProviderType
120
121	app = Concierge("large-api", config=Config(
122	provider_type=ProviderType.SEARCH,
123	max_results=5,
124	))
125	```
126
127	No matter how many tools you register, the agent only ever sees:
128
129	```
130	search_tools(query: st

People who liked this also viewed

6 picks

Type

fastmcp🚀 The fast, Pythonic way to build MCP servers and clients.MCP ServersJul 202686M27k
playwright-mcpPlaywright MCP serverMCP ServersJul 202626M35k
ext-appsOfficial repo for spec & SDK of MCP Apps protocol - standard for UIs embedded AI chatbots, served by MCP serversMCP ServersJul 20268.5M2.6k
langchain-mcp-adaptersLangChain 🔌 MCPMCP ServersJul 20267.0M3.6k
mcporterCall MCPs via TypeScript, masquerading as simple TypeScript API. Or package them as cli.MCP ServersJul 20262.5M4.8k
mcp-proxyA bridge between Streamable HTTP and stdio MCP transportsMCP ServersJul 20262.2M2.7k

concierge

113installs/mo · pypi533stars97forks

View on GitHub

How to install concierge?

concierge-hq/concierge

$git clone https://github.com/concierge-hq/concierge

Installs into the current project.

Files · 1

View on GitHub

README.md

1	<div align="center">
2
3	<!-- omit in toc -->
4
5	<picture>
6	<img width="900" alt="Concierge Banner" src="assets/concierge-banner.png" />
7	</picture>
8
9	# Concierge AI 🚀
10
11	<strong>The fabric for reliable MCP servers and AI applications.</strong>
12
13	[![Docs](https://img.shields.io/badge/docs-getconcierge.app-blue)](https://docs.getconcierge.app)
14	[![Discord](https://img.shields.io/badge/community-discord-5865F2?logo=discord&logoColor=white)](https://discord.gg/Y3ayRa33Pg)
15	[![PyPI - Version](https://img.shields.io/pypi/v/concierge-sdk.svg)](https://pypi.org/project/concierge-sdk)
16	[![Python](https://img.shields.io/badge/python-3.9+-8B5CF6?logo=python&logoColor=white)](https://pypi.org/project/concierge-sdk)
17
18	</div>
19
20	The [Model Context Protocol](https://modelcontextprotocol.io) (MCP) is a standardized way to connect AI agents to tools. Instead of exposing a flat list of every tool on every request, Concierge progressively discloses only what's relevant. Concierge guarantees deterministic results and reliable tool invocation.
21
22	## Getting Started
23
24	> [!NOTE]
25	> Concierge requires Python 3.9+. We recommend installing with [uv](https://docs.astral.sh/uv/) for faster dependency resolution, but pip works just as well.
26
27	```bash
28	pip install concierge-sdk
29	```
30
31	Scaffold a new project:
32
33	```bash
34	concierge init my-store # Generate a ready to run project
35	cd my-store # Enter project
36	python main.py # Start the MCP server
37	```
38
39	Or wrap an existing MCP server two lines, nothing else changes:
40
41	```python
42	# Before
43	from mcp.server.fastmcp import FastMCP
44	app = FastMCP("my-server")
45
46	# After: just wrap it
47	from concierge import Concierge
48	app = Concierge(FastMCP("my-server"))
49	```
50
51	> [!TIP]
52	> Concierge works at the MCP protocol level. It dynamically changes which tools are returned by `tools/list` based on the current workflow step. The agent and client don't need to know Concierge exists, they just see fewer, more relevant tools at each point.
53
54	<br />
55
56	```python
57	from concierge import Concierge
58	from mcp.server.fastmcp import FastMCP
59
60	app = Concierge(FastMCP("my-server"))
61
62	# Your @app.tool() decorators stay exactly the same.
63	# You can additionally add app.stages and app.transitions.
64	```
65
66	> [!NOTE]
67	> The wrap and go gives you progressive tool disclosure immediately. Add `app.stages` and `app.transitions` when you want full workflow control, no code changes required.
68
69	<br />
70
71	## Usage
72
73	### Group tools into steps
74
75	Instead of exposing everything at once, group related tools together. Only the current step's tools are visible to the agent:
76
77	```python
78	app.stages = {
79	"browse": ["search_products", "view_product"],
80	"cart": ["add_to_cart", "remove_from_cart", "view_cart"],
81	"checkout": ["apply_coupon", "complete_purchase"],
82	}
83	```
84
85	### Define transitions
86
87	Control which steps can follow which. The agent moves forward (or backward) only along paths you allow:
88
89	```python
90	app.transitions = {
91	"browse": ["cart"], # Can only move to cart
92	"cart": ["browse", "checkout"], # Can go back or proceed
93	"checkout": [], # Terminal step
94	}
95	```
96
97	<details>
98	<summary><b>Share state between steps</b></summary>
99	<br>
100
101	Pass data between workflow steps without round-tripping through the LLM. State is session-scoped and works across distributed replicas:
102
103	```python
104	# In the "browse" step - save a selection
105	app.set_state("selected_product", {"id": "p1", "name": "Laptop"})
106
107	# In the "cart" step retrieve it directly
108	product = app.get_state("selected_product")
109	```
110	</details>
111
112	<details>
113	<summary><b>Scale with semantic search</b></summary>
114	<br>
115
116	When you have hundreds of tools, enable semantic search to collapse your entire API behind two meta-tools:
117
118	```python
119	from concierge import Concierge, Config, ProviderType
120
121	app = Concierge("large-api", config=Config(
122	provider_type=ProviderType.SEARCH,
123	max_results=5,
124	))
125	```
126
127	No matter how many tools you register, the agent only ever sees:
128
129	```
130	search_tools(query: st

People who liked this also viewed

6 picks

Type