LatentGate

Install

Works with Claude Code, Codex, VS Code, Cursor, Windsurf, Claude Desktop. Read the repository before you give any tool access to your data.

npx agentmarket install connector/kathanmodh259-latent-gate

About

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list, install config and code. Checked Oct 25, 2026.

Tools

5 read · 2 write · 0 delete · 2 not yet classified

Secrets

None asked for

Install scripts

None found

Runs as

Python package

Tools (9)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• compress_conversation

  Compress a multi-turn conversation history into a compact summary.

• compress_documents

  Compress retrieved RAG documents into key facts.

• compress_image

  Describe an image with a local vision model (Ollama llava) as a compact ~150-token structured scene payload, instead of sending the raw image (typically 1000+ tokens). Requires Ollama with a vision model pulled.

• compress_text

  Compress a long prompt with the optimizer plus a fact-checked local LLM rewrite (Ollama). Prefer optimize_text when Ollama isn't running.

• count_tokens

  Count tokens in text (tiktoken o200k_base when installed, else an estimate).

• fetch_url_optimized

  Fetch a public web page, JSON API response, raw log or doc by URL and return a token-optimized version: HTML is reduced to its visible text, JSON is compacted losslessly, runs of near-identical lines are folded (first, last and value ranges kept). Deterministic, no summarizing model; code, URLs and quoted strings come back byte-for-byte. Use level=lossless when every line must survive. Pass `question` to focus on what you need. Private/internal addresses are refused.

• get_stats

  Get cumulative token savings statistics.

• optimize_text

  Deterministically shrink text before passing it on (e.g. to another tool or model): lossless JSON/whitespace/duplicate folding, log-run folding, filler removal, optional question-aware selection under a token budget. Same input always gives the same output. No Ollama or API key required.

• read_file_optimized

  Read a local text file (logs, JSON/API dumps, docs, config, source) and return a token-optimized version instead of the raw content. Use this INSTEAD of reading large files directly: pretty-printed JSON shrinks ~30% losslessly and repetitive logs ~90% (runs are folded with value ranges kept). Code, URLs and quoted strings are returned byte-for-byte. Pass `question` to focus selection on what you need. Works offline, no Ollama or API key required.

Signals

Stars

25

Forks

2

Open issues

0

Contributors

2

Weekly downloads

57

Last commit

Oct 21, 2026 (8 days ago)

Last release

Oct 8, 2026 (21 days ago)

Last package publish

Oct 8, 2026 (21 days ago)

From GitHub (fetched Oct 24, 2026) and the package registry.

Details

Registry name

io.github.KathanModh259/latent-gate

Version

1.4.2

Auth

No auth

Packages

pypi latent-gate

License

Proprietary

Repository

github.com/KathanModh259/latent-gate

Website

github.com/KathanModh259/latent-gate#use-it-in-claude

First seen

Oct 24, 2026

Last checked

Oct 24, 2026

Source

Official MCP Registrygithub-graphql