media MCP Server
LeScout — sandboxed scouting + caveman-compress context for every agent. Part of LeSearch AI.
Discovered via github-topic:mcp and last synced 3mo ago.
1. Install the package
npm install lescout
Role
See [`Plans/BRAND.md`](Plans/BRAND.md) for the full Personal Software Factory thesis. --- ## Why Every developer today spends 30–60% of their time **searching**: docs, Stack Overflow, X, old chats, their own old code, GitHub. LeSearch flips that: **search once, agents act forever.** The brain compounds; ingest is deterministic; retrieval is hybrid; agents run the loops. LeScout is the first primitive of that stack: 1. **Sandbox** — foreign code (cloned repos) never touches your host. Only structured text crosses the container boundary. 2. **Context-loader** — `lescout context <target>` collapses everything the brain knows about a project into ONE dense file. A fresh agent loads that file and skips the re-derivation cost. --- ## Install **Prerequisites:** macOS or Linux · [Bun](https://bun.sh) ≥ 1.1 · Docker (or Colima) · [GBrain](https://github.com/garrytan/gbrain) on `$PATH`. ```bash # Clone git clone https://github.com/aryateja2106/lescout.git ~/Projects/lescout cd ~/Projects/lescout # Install deps bun install # Build the sandbox image (one time; 13 MB Alpine) docker build -t lescout/sandbox:latest docker/sandbox/ # Add the CLI to your PATH (one of): ln -s ~/Projects/lescout/bin/lescout ~/.bun/bin/lescout # if ~/.bun/bin is on PATH ln -s ~/Projects/lescout/bin/lescout /usr/local/bin/lescout # system-wide # …or just: alias lescout="$HOME/Projects/lescout/bin/lescout" # Verify lescout --version # 0.0.5 lescout help # the man-page ``` LeScout shells out to `gbrain` from `$PATH`. Override with `GBRAIN_BIN=/abs/path/to/gbrain`. --- ## Quickstart ### The killer flow — give a fresh agent dense project context in one line ```bash # 1. Compress everything in the brain about a project into one file: lescout context lockshell # 30K-token bundle # 2. Open a fresh chat in any agent (Claude Code, Cursor, Codex, Gemini, pi) # and paste: > Read ~/.lescout/context/lockshell-2026-05-15.md before answering anything. > What's the next blocking thing on this project? ``` That's it. The agent now has dense, ranked context: architecture, threat model, roadmap, contributing notes, related session summaries — without re-deriving anything. ### Sandbox-scout a repository ```bash # Read-only Docker, --cap-drop ALL, --no-new-privileges, no postinstall scripts lescout repo https://github.com/safishamsi/graphify lescout repo https://github.com/rohitg00/agentmemory --dry-run ``` ### Sandbox-scout official docs from a GitHub repo ```bash lescout docs https://github.com/colinhacks/zod ``` ### Discover and resume sessions across every agent harness ```bash lescout session list --limit 10 lescout session list --agent codex --project mconnect lescout session resume c1a443ce ``` ### Self-discovery for any agent ```bash lescout help # tight man-page lescout help context # per-command detail lescout context --help # same ``` ### Progressive disclosure for skills Agent skill files (`SKILL.md` with YAML front matter) live in multiple directories. Loading every body for every agent burns context. LeScout flips this: read front matter first, fetch bodies only when actually needed. ```bash lescout skills list # ~100 tokens per skill (cheap index) lescout skills list --scope pi # filter by source: pi
The cheapest AI media API on the market. Generate images (Flux), music (AceStep), speech with voice cloning, transcribe video/audio, OCR, video generation, background removal, upscale, style transfer, and prompt enhancement — all through one unified API. Free $5 credit on signup.
Write professional press releases that get media attention and coverage
Learn how to use the deAPI AI Media Suite (Community) Claude skill. Complete guide with installation instructions and examples.
Learn how to use the Press Release Writer Claude skill. Complete guide with installation instructions and examples.