limelit-co avatar

limelit-open

Connect a self-hosted Limelit Open install: your own AI visibility tracking, read from Claude over the local limelit binary.

version
0.1.0
GitHub stars
447

Install

/plugin install limelit-open@limelit

Run it inside Claude Code after adding the marketplace with /plugin marketplace add limelit-co/open.

Contents

Discovered from source

Skills 1

AI visibility report (Limelit Open)

Build a short, evidence-backed report from the limelit-open MCP server, which runs the local limelit binary.

If its tools are missing, the binary is not on PATH or the data directory is wrong. Tell the user to install Limelit Open (https://github.com/limelit-co/open), confirm limelit mcp starts from a terminal, and set the data directory in the plugin's settings. Then stop.

Rules that apply to every number

  • Quote the n behind every figure. Under 20, say it is still settling.
  • Report only metrics a tool returned.
  • Name the targets each figure covers.
  • Headline visibility excludes prompts tagged branded.
  • A Google query with no AI Overview is excluded, not a miss.

Runs

Only start a run when the user asks. Call reevaluate_all_prompts (or reevaluate_prompt) with dry_run=true, show the plan, and call again with dry_run=false after they confirm. Then check get_run_activity until it is done.

Steps

  1. get_active_property. If has_answers is false, give the user its next_step and link, list once_set_up_you_can_ask, report no numbers, and stop. If it returns suggested_days, pass that as days to the metric tools.
  2. Where they stand: get_overview_kpis.
  3. What they lose: get_matrix, then list_chats with show=missed. Quote one short line of evidence with get_chat.
  4. What gets cited instead: list_top_sources, then list_source_urls with value=<host> for the top one or two.
  5. Freshness: get_run_activity.

Not in Limelit Open

Sentiment, prompt generation, competitor discovery, segments, portfolios, Search Console, plans and drafts are Limelit Cloud features with no tool here. Say so rather than approximating them. upgrade_to_cloud moves the data across when the user wants those.

Output

  • Headline: share of voice, rank, n, targets covered.
  • By engine: one row per engine.
  • Lost questions: up to five, each with the competitor that won it.
  • Cited instead: up to five sources.
  • Data note: window, n, anything still settling.

MCP servers 1

Sponsors

Details

Marketplace
limelit-co/open
Source
plugins/limelit-open
Category
Marketing
Keywords
geo, aeo, ai-visibility, self-hosted, open-source, mcp

More from limelit-co/open

All 3