YOU ARE VIEWING ONE ITEM FROM THE AICRIER FEED

Mortar ships local FIM autocomplete

AICrier tracks AI developer news across Product Hunt, GitHub, Hacker News, YouTube, X, arXiv, and more. This page keeps the article you opened front and center while giving you a path into the live feed.

// WHAT AICRIER DOES

7+

TRACKED FEEDS

24/7

SCRAPED FEED

Short summaries, external links, screenshots, relevance scoring, tags, and featured picks for AI builders.

Mortar ships local FIM autocomplete
OPEN LINK ↗
// 45d agoOPENSOURCE RELEASE

Mortar ships local FIM autocomplete

Mortar is a minimal VS Code extension for ghost-text code autocomplete against llama.cpp or OpenAI-compatible servers. It favors native llama.cpp `/infill`, falls back to `/v1/completions`, and deliberately skips chat, embeddings, and agent features.

// ANALYSIS

Mortar is small by design, which is exactly the point: local-code-completion users often need less product surface, not another full AI IDE.

  • The setup flow is built around endpoint and model selection, making it a cleaner companion for `llama-server` and `llama-swap`.
  • Native `/infill` support matters because FIM-capable coding models can use prefix and suffix context without prompt-template guesswork.
  • The OpenAI-compatible fallback broadens backend support while preserving the same autocomplete-only UX.
  • This is early and tiny, with a v0.1.0 release and minimal community signal, but it targets a real pain point in local AI coding workflows.
// TAGS
mortarideai-codinginferenceopen-sourceself-hostedllm

DISCOVERED

45d ago

2026-04-22

PUBLISHED

45d ago

2026-04-21

RELEVANCE

7/ 10

AUTHOR

xhimaros