📡 MCP Radar

ucsandman/declick

Compile an API, an MCP server, a database or a web page into a CLI your agent can run. Ten engines, one output contract, 4.1x less context than raw MCP.

⭐ 17 stars
⑂ 6 forks
JavaScript
momentum ▲ 46.0
created 2026-09-02
on radar since 2026-09-05
star trend 7 → 17 since 2026-09-05
agent-toolsai-agentsai-codingclaude-codeclicodexcursordeveloper-toolsgraphqlllmmcpmodel-context-protocolnodejsopenapiopenapi-generatorpostmansqlitewindows-automation
View on GitHub ↗Homepage ↗

About declick

A compiler for the interface your agent already has: the shell. MCP makes an agent carry every tool's schema on every turn; declick compiles an API, an MCP server, or a database once into named verbs the model loads one at a time, and every verb returns one envelope with five exit codes. Ten engines, zero runtime dependencies, Node 24.

The saving is measured, not claimed: against nine real MCP servers (258 tools), the raw tool listing an MCP client puts in context is 236,818 bytes and declick describe is 58,309, a 4.1x reduction. node scripts/bench-tokens.mjs reproduces it on your own adapters; the method and the caveats are in docs/bench.md.

Requires Node 24 or newer. An older Node exits 1 with one line naming the version it found, from either entry point, instead of a stack trace.

declick setup puts ~/.declick/bin on PATH, adopts your agent's existing MCP servers as adapters, adds a rules block to its instruction file, and (Claude Code only) installs a hook that nudges the model toward the adapter; declick setup --revert puts everything back byte for byte.

The fifth line is a real call: petstore's user endpoints declare no auth, so the envelope comes back with the row. get-pet-by-id declares an API key in the spec, so without PETSTOREAPIKEY set it exits 4 and names the key; --dry-run shows the request it would send instead.

declick add writes three things: ~/.declick/petstore/manifest.json (the compiled surface), ~/.declick/bin/petstore (a two line launcher, with a petstore.cmd twin on Windows), and a petstore/SKILL.md in each agent skills directory it knows about, s

From the project README.

Maintaining this server?

Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:

[![On MCP Radar](https://mcp.liqiwa.com/badge.svg)](https://mcp.liqiwa.com/s/ucsandman--declick.html)

Related MCP servers

00200200/usagetrim

Keep the signal. Cut the noise. Local CLI + MCP — fold verbose tool output for Claude Code, Codex, Cursor & Desktop.

⭐ 11Python
magna-nz/tallybook

Prices your Claude Code and Codex sessions from the transcripts on disk, shows what each agent cost over any period, and tells you in plain English what would have been cheaper. CLI, Browser UI and MC

⭐ 5Go
adityajha2005/decisionkit

Fast, cheap, probability-scored decisions for coding agents, scripts and CI. CLI + MCP server + Claude Code plugin for the OpenAI Decisions API and Jev (TypeSafe AI).

⭐ 4TypeScript
thaildhe172591/pythia

PL/SQL development for AI agents on Oracle Database - expert queries, impact analysis, and a snapshot-verified write path with honest rollback

⭐ 3Python
holetexvn/memgw

Shared long-term memory for AI coding agents — one local store for Claude Code, Codex CLI, opencode and anything that speaks MCP. SQLite + FTS5, no vector DB, $1-3/month.

⭐ 4JavaScript
friday-memory/friday

⚡ The open-source persistent cognitive memory layer for AI coding agents. Stop your Cursor, Claude & Copilot from having amnesia.

⭐ 51HTML

📬 Get the weekly radar in your inbox

The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.