bug-ops/zeph

Rust AI agent where skills evolve from real usage and every context token earns its place. Hybrid inference: Ollama, Claude, OpenAI, GGUF. Graph memory, DAG planner, MCP client, ACP for Zed/VS Code/Helix. One 15 MB binary, zero runtime deps.

41
/ 100
Emerging

Implements semantic skill retrieval with Bayesian re-ranking and self-learning evolution from real usage feedback, plus a three-tier context compression pipeline (deferred summaries, pruning, LLM compaction) with subgoal-aware protection and SQLite-backed tool output archiving. Supports multi-provider routing via EMA latency balancing, complexity triage, and LinUCB bandit algorithms; integrates with Qdrant for graph memory with typed entity relationships, temporal versioning, and point-in-time queries across PostgreSQL or SQLite backends.

No Package No Dependents
Maintenance 13 / 25
Adoption 9 / 25
Maturity 9 / 25
Community 10 / 25

How are scores calculated?

Stars

6

Forks

1

Language

Rust

License

MIT

Last pushed

Mar 13, 2026

Monthly downloads

105

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/rag/bug-ops/zeph"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.