sutura
Health Uyari
- License — License: Apache-2.0
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Low visibility — Only 5 GitHub stars
Code Uyari
- network request — Outbound network request in .agents/skills-sync.py
- network request — Outbound network request in .agents/skills/engineering/ms-rust/generate.py
Permissions Gecti
- Permissions — No dangerous permissions requested
Bu listing icin henuz AI raporu yok.
sutura is being built to answer questions about data as the person or agent asking, using metric definitions somebody certified
sutura
(Italian) sutura is a surgical stitching technique to bring wound edges together.
sutura:
- is an MCP server
- to answer questions about data as the person or agent asking using metadata; strong metadata like blessed metrics are preferred.
- can connect systems across boundaries.
- has a strong focus on security and E2E impersonation
- clear refusal results based on the metadata
Status
A governed semantic compiler and executor is served over HTTP and MCP. A deployment can verify a
caller's own token or use a shared deployment token. The shipped BigQuery adapter can federate a
verified caller's assertion through a declared per-subject account map, but no observed served run
has shown the source accepting that identity. Other data sources use a declared shared identity.
sutura is designed around 2 flavours:
- single player (shared source identity) - supported by the file and other shared-identity adapters
- multiplayer (per-subject source identity) - built for BigQuery, with live source acceptance still
unproven; see where identity is proven
Each source declares its own identity posture; a federated answer names the posture of every leg.
How a federated join key is compared records what the combiner
refuses and what it does not cover.
Vision
- Have pluggable metadata source/s
- Have pluggable data source/s
- Have a focus on security (impersonation, prompt injection, semantic & rust compiler verification)
- Support query federation (at least lightly)
We are standing on the shoulders of giants/ecosystems/role models but sometimes just taking inspiration and combining:
- https://datafusion.apache.org/index.html
- https://github.com/Canner/WrenAI
- https://github.com/spiceai/spiceai
Building around arrow & arrow flight.
Documentation
https://telekom.github.io/sutura/, built from docs/ in this repository.
Licence
Apache-2.0. Third-party material adapted here is recorded in VENDOR.md.
Yorumlar (0)
Yorum birakmak icin giris yap.
Yorum birakSonuc bulunamadi