Local tool-calling agent engine: a model-agnostic loop over local LLMs (the Revenant harness).
Project description
nerva-agent
The agent engine behind Revenant: a
model-agnostic, tool-calling loop over local LLMs. Provides the loop, the tool registry, a
dual tool-call protocol (native tool_calls + a prompt-based fallback for models without a
tool template), read/write/edit/bash tools, task-based model routing, and hardware-aware
capacity tuning.
pip install nerva-agent
What's in it
agent_loop—AgentLoop.run(goal): call → parse action → (approve) → dispatch → observe → repeat.agent_tools/agent_protocol— tool registry +parse_action(native + prompt-based).agent_fs_tools/agent_edit_tools/agent_bash_tool— path-confined, approval-gated tools.agent_router— route each turn to the best local model per role.agent_capacity/agent_native_tools— hardware detection + native-tool auto-detection.
Depends on nerva-core. Powers
revenant-cli.
License
MIT.
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file nerva_agent-0.1.0.tar.gz.
File metadata
- Download URL: nerva_agent-0.1.0.tar.gz
- Upload date:
- Size: 26.4 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.11.1
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
bfe8b4528d772eb9b72fbfd36f1d24fa012595ad458f72a4f1d12b388ac961ea
|
|
| MD5 |
2e0e264f4723944aa9e27ce2d1783d26
|
|
| BLAKE2b-256 |
ff169523556edd8257636fea37a80f4acf53d66319e15b33505dec6e5db757d1
|
File details
Details for the file nerva_agent-0.1.0-py3-none-any.whl.
File metadata
- Download URL: nerva_agent-0.1.0-py3-none-any.whl
- Upload date:
- Size: 30.7 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.11.1
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
2e8d037e69331eabdde85d4e660785306001a0b674fe2d9c832bdfbd3bb823a6
|
|
| MD5 |
8464477cd4f2e10ce93eaeb9c1081124
|
|
| BLAKE2b-256 |
d168534620d1114695ad2e3e3899fdf42639f1d430c2395933e6d93c71213eee
|
