{"full_name":"zylon-ai/private-gpt","owner":"zylon-ai","name":"private-gpt","shard":"1","url":"https://github.com/zylon-ai/private-gpt","description":"Complete API layer for private AI applications on local models: RAG, skills, tools, MCP, text-to-sql, and more. Works with any OpenAI-compatible inference server.","homepage":"https://www.zylon.ai/private-gpt","topics":["ai","ai-tools","on-premise"],"tag_pages":["ai","ai-tools","on-premise"],"language":"Python","license":"Apache-2.0","avatar":"https://avatars.githubusercontent.com/u/143802295?v=4","created_at":"2023-05-02T09:15:31Z","pushed_at":"2026-11-13T17:13:10Z","metrics":{"stars":57635,"forks":7627,"watchers":447,"issues":1251,"prs":698,"releases":14,"issues_open":3,"issues_closed":1248,"prs_open":12,"prs_closed":313,"prs_merged":373,"discussions":465},"growth":{"day":{"stars":1,"forks":1,"base_day":"2026-11-15"},"week":{"stars":2,"forks":0,"base_day":"2026-11-09"}},"history":{"dates":["2026-11-09","2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[57633,57633,57634,57634,57635,57635,57634,57635],"forks":[7627,7629,7629,7626,7626,7625,7626,7627],"watchers":[448,448,448,447,447,447,447,447],"issues_open":[null,null,2,2,2,2,3,3],"prs_open":[null,null,13,13,13,9,11,12]},"boards":{},"same_owner":[],"issues":[{"number":2449,"title":"[FEATURE] 3x fewer recomputed tokens when chunks are re-retrieved, by keeping the first occurrence in context dedup","comments":1,"created_at":"2026-11-06T16:58:19Z"},{"number":2463,"title":"Private Gpt MCP is listed on MCP ServerList","comments":0,"created_at":"2026-11-14T01:40:04Z"},{"number":2441,"title":"CI: update-claude-specs failing: openapi_spec_url not found in anthropic-sdk-typescript .stats.yml","comments":0,"created_at":"2026-11-01T19:13:46Z"}],"discussions":[{"number":2460,"title":"Keeping a private, offline model up to date without retraining","comments":1,"category":"Ideas","created_at":"2026-11-12T19:21:16Z"},{"number":2458,"title":"A revocation-boundary experiment for private-gpt workflows","comments":0,"category":"Show and tell","created_at":"2026-11-12T00:31:33Z"},{"number":2433,"title":"Show and tell: Hardening Private Text-to-SQL & MCP Tool Backends — Read-Only Guardrails & Stdio Isolation for Local Models","comments":0,"category":"Show and tell","created_at":"2026-10-26T23:28:39Z"},{"number":2406,"title":"already accessed by another instance of Qdrant client","comments":4,"category":"Q&A","created_at":"2026-10-10T04:30:42Z"},{"number":2395,"title":"Two traps for local RAG: context_window vs the loaded n_ctx, and why abliterated models undermine grounding specifically","comments":0,"category":"General","created_at":"2026-09-26T12:31:34Z"},{"number":2381,"title":"Optional retrieval rerank (off by default): generic hook, Jev as one possible backend","comments":0,"category":"Ideas","created_at":"2026-09-18T17:09:49Z"},{"number":2376,"title":"Private data, local models, tools and MCP: what stops an agent doing more than the task?","comments":0,"category":"General","created_at":"2026-09-18T13:42:38Z"},{"number":2351,"title":"[BUG] ImportError: No module named 'fcntl' on Windows in model_discovery.py","comments":1,"category":"General","created_at":"2026-08-30T20:09:33Z"}],"latest_release":{"name":"v1.0.1","tag":"v1.0.1","published_at":"2026-06-18T13:56:00Z","body":"## [1.0.1](https://github.com/zylon-ai/private-gpt/compare/v1.0.0...v1.0.1) (2026-06-16)\n\n\n### Bug Fixes\n\n* add missing docker database deps ([#2264](https://github.com/zylon-ai/private-gpt/issues/2264)) ([4302bd6](https://github.com/zylon-ai/private-gpt/commit/4302bd629a4ca5fcee23e9064a0448cd110eb523))\n* ensure that the publisher is ready to publish new messages ([#2266](https://github.com/zylon-ai/private-gpt/issues/2266)) ([4c0c500](https://github.com/zylon-ai/private-gpt/commit/4c0c500bbd34a06feedb52e1d267d9b7b918554b))\n* make embedding API key optional for local OpenAI-compatible engines ([#2261](https://github.com/zylon-ai/private-gpt/issues/2261)) ([c120a90](https://github.com/zylon-ai/private-gpt/commit/c120a90603f78bb28b19f758a9ba4b40c37480ff)), closes [#2260](https://github.com/zylon-ai/private-gpt/issues/2260)\n* use PGPT_HOME for local data, caches, and cleanup paths ([#2267](https://github.com/zylon-ai/private-gpt/issues/2267)) ([4021cf4](https://github.com/zylon-ai/private-gpt/commit/4021cf4e203d4c309f496a8fbf95271f6084834c))\n\n\n### Performance Improvements\n\n* **prompts:** use date-level current_date to preserve prompt-prefix caching ([#2271](https://github.com/zylon-ai/private-gpt/issues/2271)) ([a64f1c9](https://github.com/zylon-ai/private-gpt/commit/a64f1c9cf088af5d5047126c9099ea4d674cc70c))","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}