rp-engine

Name: rp-engine
Author: jieyefriic

1.2kstars9forks

How to install rp-engine?

jieyefriic/rp-engine

$git clone https://github.com/jieyefriic/rp-engine

Installs into the current project.

Summary

YAML-native agent workflow execution engine, written in Rust.

Files · 1

View on GitHub

README.md

1	# riceprompt-engine
2
3	YAML-native agent workflow execution engine, written in Rust.
4
5	You describe an agent workflow as a YAML file — nodes, edges, prompts, data
6	sources, MCP tools — and the engine parses it, resolves dependencies, and
7	executes the graph: making LLM calls, running scripts, querying databases,
8	calling MCP tools, iterating over data, and orchestrating multi-agent plans.
9
10	This engine powers [RicePrompt](https://riceprompt.app) — the visual
11	agent IDE where you build and run these workflows without writing YAML by hand.
12
13	```toml
14	[dependencies]
15	riceprompt-engine = "0.1"
16	```
17
18	## Features
19
20	- YAML-native — entire workflow (graph, prompts, data sources, providers)
21	in a single declarative file. See [`docs/FLOW_SPEC.md`](docs/FLOW_SPEC.md)
22	for the authoritative spec.
23	- Multi-provider LLM support — OpenAI, Anthropic, Gemini, DeepSeek, Qwen,
24	Zhipu, Moonshot, MiniMax, xAI, Huoshan, and any OpenAI-compatible endpoint.
25	- Streaming, tool calling, structured output — first-class across providers.
26	- Rich node types — `generate`, `transform` (Rhai scripting), `iterator`,
27	`supervisor` (multi-agent routing), `subgraph`, `data_connector`,
28	`skill_set` (progressive-disclosure knowledge bundles), `mcp` /
29	`mcp_tools` (Model Context Protocol).
30	- Built-in data connectors — PostgreSQL, MySQL, MongoDB, Redis, Qdrant,
31	S3-compatible object storage, REST APIs.
32	- Harness layer — workflow-level instructions (CLAUDE.md-style) injected
33	into every generate node, with persistent memory support.
34	- Self-describing results — `ExecutionResult` can include the source YAML
35	so downstream tooling renders the topology + per-node results from one file.
36	- Checkpoint / resume — pause and resume long-running workflows.
37
38	## Quick start
39
40	A minimal three-node workflow:
41
42	```yaml
43	version: "1.0"
44	name: "hello_world"
45
46	providers:
47	openai:
48	api_key: "${OPENAI_API_KEY}"
49
50	nodes:
51	- id: start
52	type: start
53
54	- id: greet
55	type: generate
56	config:
57	provider: openai
58	model: gpt-4o-mini
59	template: tpl_greet
60	variables:
61	name: "start.name"
62
63	- id: response
64	type: response
65	config:
66	output:
67	greeting: "greet.output"
68
69	edges:
70	- from: start
71	to: greet
72	- from: greet
73	to: response
74
75	templates:
76	tpl_greet:
77	user_prompt: "Greet {{name}} warmly in one sentence."
78	```
79
80	Run it:
81
82	```rust
83	use riceprompt_engine::Engine;
84	use serde_json::json;
85
86	#[tokio::main]
87	async fn main() -> anyhow::Result<()> {
88	let yaml = std::fs::read_to_string("hello.yaml")?;
89	let engine = Engine::builder().build()?;
90	let result = engine.run_yaml(&yaml, json!({ "name": "Ada" })).await?;
91	println!("{}", serde_json::to_string_pretty(&result)?);
92	Ok(())
93	}
94	```
95
96	More runnable examples live under [`examples/`](examples/).
97
98	## Documentation
99
100	- [`docs/FLOW_SPEC.md`](docs/FLOW_SPEC.md) — authoritative YAML workflow spec
101	(node types, fields, providers, data sources, harness, skills, MCP).
102	- A user-facing usage guide ("skill guide") will be published separately.
103
104	## Related
105
106	- [RicePrompt](https://riceprompt.app) — visual agent IDE built on top
107	of this engine. Design workflows in a graph editor, run them in-browser,
108	and export the same YAML this engine consumes.
109
110	## Project status
111
112	`0.1.x` — the API may change between minor versions while the spec
113	stabilizes. Pin an exact version if you need stability.
114
115	## Contributing
116
117	Issues and PRs welcome. Please:
118
119	- Run `cargo fmt` and `cargo clippy --all-targets` before submitting.
120	- Add tests for new node types or provider behaviors.
121	- For changes that touch the YAML surface, update `docs/FLOW_SPEC.md` in
122	the same PR.
123
124	## License
125
126	Licensed under either of
127
128	- Apache License, Version 2.0 ([LICENSE-APACHE](LICENSE-APACHE) or
129	<http://www.apache.org/licenses/LICENSE-2.0>)
130	- MIT license ([LICENSE-MIT](LICENSE-MIT) or
131	<http://opensource.org/licenses/MIT>)
132
133	at your option.
134
135	Unless you explicitly state otherwise, any contribution intentionally
136	submitt

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

rp-engine

1.2kstars9forks

View on GitHub

How to install rp-engine?

jieyefriic/rp-engine

$git clone https://github.com/jieyefriic/rp-engine

Installs into the current project.

Summary

YAML-native agent workflow execution engine, written in Rust.

Files · 1

View on GitHub

README.md

1	# riceprompt-engine
2
3	YAML-native agent workflow execution engine, written in Rust.
4
5	You describe an agent workflow as a YAML file — nodes, edges, prompts, data
6	sources, MCP tools — and the engine parses it, resolves dependencies, and
7	executes the graph: making LLM calls, running scripts, querying databases,
8	calling MCP tools, iterating over data, and orchestrating multi-agent plans.
9
10	This engine powers [RicePrompt](https://riceprompt.app) — the visual
11	agent IDE where you build and run these workflows without writing YAML by hand.
12
13	```toml
14	[dependencies]
15	riceprompt-engine = "0.1"
16	```
17
18	## Features
19
20	- YAML-native — entire workflow (graph, prompts, data sources, providers)
21	in a single declarative file. See [`docs/FLOW_SPEC.md`](docs/FLOW_SPEC.md)
22	for the authoritative spec.
23	- Multi-provider LLM support — OpenAI, Anthropic, Gemini, DeepSeek, Qwen,
24	Zhipu, Moonshot, MiniMax, xAI, Huoshan, and any OpenAI-compatible endpoint.
25	- Streaming, tool calling, structured output — first-class across providers.
26	- Rich node types — `generate`, `transform` (Rhai scripting), `iterator`,
27	`supervisor` (multi-agent routing), `subgraph`, `data_connector`,
28	`skill_set` (progressive-disclosure knowledge bundles), `mcp` /
29	`mcp_tools` (Model Context Protocol).
30	- Built-in data connectors — PostgreSQL, MySQL, MongoDB, Redis, Qdrant,
31	S3-compatible object storage, REST APIs.
32	- Harness layer — workflow-level instructions (CLAUDE.md-style) injected
33	into every generate node, with persistent memory support.
34	- Self-describing results — `ExecutionResult` can include the source YAML
35	so downstream tooling renders the topology + per-node results from one file.
36	- Checkpoint / resume — pause and resume long-running workflows.
37
38	## Quick start
39
40	A minimal three-node workflow:
41
42	```yaml
43	version: "1.0"
44	name: "hello_world"
45
46	providers:
47	openai:
48	api_key: "${OPENAI_API_KEY}"
49
50	nodes:
51	- id: start
52	type: start
53
54	- id: greet
55	type: generate
56	config:
57	provider: openai
58	model: gpt-4o-mini
59	template: tpl_greet
60	variables:
61	name: "start.name"
62
63	- id: response
64	type: response
65	config:
66	output:
67	greeting: "greet.output"
68
69	edges:
70	- from: start
71	to: greet
72	- from: greet
73	to: response
74
75	templates:
76	tpl_greet:
77	user_prompt: "Greet {{name}} warmly in one sentence."
78	```
79
80	Run it:
81
82	```rust
83	use riceprompt_engine::Engine;
84	use serde_json::json;
85
86	#[tokio::main]
87	async fn main() -> anyhow::Result<()> {
88	let yaml = std::fs::read_to_string("hello.yaml")?;
89	let engine = Engine::builder().build()?;
90	let result = engine.run_yaml(&yaml, json!({ "name": "Ada" })).await?;
91	println!("{}", serde_json::to_string_pretty(&result)?);
92	Ok(())
93	}
94	```
95
96	More runnable examples live under [`examples/`](examples/).
97
98	## Documentation
99
100	- [`docs/FLOW_SPEC.md`](docs/FLOW_SPEC.md) — authoritative YAML workflow spec
101	(node types, fields, providers, data sources, harness, skills, MCP).
102	- A user-facing usage guide ("skill guide") will be published separately.
103
104	## Related
105
106	- [RicePrompt](https://riceprompt.app) — visual agent IDE built on top
107	of this engine. Design workflows in a graph editor, run them in-browser,
108	and export the same YAML this engine consumes.
109
110	## Project status
111
112	`0.1.x` — the API may change between minor versions while the spec
113	stabilizes. Pin an exact version if you need stability.
114
115	## Contributing
116
117	Issues and PRs welcome. Please:
118
119	- Run `cargo fmt` and `cargo clippy --all-targets` before submitting.
120	- Add tests for new node types or provider behaviors.
121	- For changes that touch the YAML surface, update `docs/FLOW_SPEC.md` in
122	the same PR.
123
124	## License
125
126	Licensed under either of
127
128	- Apache License, Version 2.0 ([LICENSE-APACHE](LICENSE-APACHE) or
129	<http://www.apache.org/licenses/LICENSE-2.0>)
130	- MIT license ([LICENSE-MIT](LICENSE-MIT) or
131	<http://opensource.org/licenses/MIT>)
132
133	at your option.
134
135	Unless you explicitly state otherwise, any contribution intentionally
136	submitt

People who liked this also viewed

6 picks

Type