No red flags found in any of the 11 categories — no credential harvesting, no data exfiltration, no curl-pipe-shell installer. Scanned against the SlowMist agent-security taxonomy, refreshed every 8 hours. Full audit →
by SAGAR-TAMANG · Agent Tool · ★ 58
Last updated: · Indexed by AgentSkillsHub · Auto-synced every 8h
🔒 Is sarvam-jev safe to install? View the security audit →
sarvam-jev Generation-free typed decisions on Sarvam's Indic models. Most decisions software asks a model to make are small: route this ticket, is this claim supported, does this need approval. A chat model can answer, but it spends its time generating text that the caller immediately parses back into an . This project reads the answer straight out of the logits instead. Options are presented as lettered slots, and the probability distribution is taken over the token IDs of those letters. Nothing is sampled, so nothing can be malformed, and one expensive state prefill is amortised across every question asked about that state. The interface pattern is TypeSafe's Jev. The point of doing it on Sarvam is the tokenizer: sarvam-1 encodes Devanagari at roughly 4 characters per token where Qwen-class tokenizers are 2–4× worse. Since the whole economic argument is "pay for the state once, then ask it many cheap questions", a model that encodes Indic states more compactly starts from a structurally better position. Status Phase 0 is done. The mechanism works; the model is the limit. Phase 0 re
| Stars | 58 |
| Forks | 10 |
| Language | Python |
| Category | Agent Tool |
| Quality Score | 61.6248311882382/100 |
| Last Updated | 2026-09-18 |
| Created | 2026-09-18 |
| Platforms | browser, cli, python |
| Est. Tokens | ~15k |
These tools work well together with sarvam-jev for enhanced workflows:
Explore other popular agent tool tools:
sarvam-jev is Generation-free typed decisions on Indic LLMs. An open Jev-style inference engine on sarvam-1: constrained logit readout instead of autoregressive JSON. Runs client-side in the browser.. It is categorized as a Agent Tool with 58 GitHub stars.
sarvam-jev is primarily written in Python.
You can find installation instructions and usage details in the sarvam-jev GitHub repository at github.com/SAGAR-TAMANG/sarvam-jev. The project has 58 stars and 10 forks, indicating an active community.
Grades come from a rule-based scan built on the SlowMist agent-security taxonomy, covering 11 red-flag categories including credential harvesting, data exfiltration, and curl | sh installers. It is a first-layer scan, not a manual audit — we say so rather than overstate it.
The scale of the problem is documented independently: Liu et al. (2026), in a study of 31,132 agent skills, report that 26.1% contain security vulnerabilities. Our own full-catalog census is published as a citable open dataset.
Sources & who's responsible: