AI Potluck
Infrastructure / Core ML frameworks & libraries

Feluda

Tattle

Feluda is a configurable engine by Tattle for analysing multilingual, multimodal content (images, text, video), built around a modular 'operators' plugin architecture. It supports vector embeddings, hash-based search, text extraction from images, and entity extraction. It is open infrastructure for analysing misinformation and social-media content, particularly in Indian-language contexts.

Verified 2026-08-13 via the tattle-made/feluda repository and the GitHub API.

Openness

5 high confidence
5.0
license
GPL-3.0(OSI copyleft)
source
public
core-gated
ungated(all nine analysis operators ship in operators/ under the one GPL-3.0 license. Tattle offers no hosted edition and no paid tier)

GPL-3.0 licensed open source project with public source. The core is ungated, from a direct read of the repository. Feluda is Tattle's civic-tech content-analysis library and there is no commercial side to gate with: no hosted edition, no pricing page and no enterprise path in the tree. Every operator the engine offers is published - classify_video_zero_shot, cluster_embeddings, detect_lewd_images, detect_text_in_image, dimension_reduction, image_vec_rep, vid_vec_rep and video_hash - under the one GPL-3.0 license that the GitHub API also classifies as GPL-3.0. The README's only mention of a paid service runs the other way: an OpenCV-based operator exists so that users who "don't want to use a google product" can avoid the Cloud Vision API.

Adoption

1 high confidence
1.0

95 GitHub stars on tattle-made/feluda, which falls in the under-1K band, level 1 on the stars fallback scale; that scale caps at 3 because a star is not a use. The project distributes no package, and no download, install or customer figure is published for it, so stars are the only honest signal available and the level is directional.

Capability

3 high confidence
3.0

Flexible modular engine focused on multimodal content analysis.

Verified 2026-08-12