Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add tobocop2/lilbee --skill lilbee_mcpgit clone --depth 1 https://github.com/tobocop2/lilbeeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tobocop2/lilbee/lilbee_mcp)<a href="https://agentmods.dev/skills/tobocop2/lilbee/lilbee_mcp"><img src="https://agentmods.dev/badge/skills/tobocop2/lilbee/lilbee_mcp/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tobocop2/lilbee/lilbee_mcp"><img src="https://agentmods.dev/badge/skills/tobocop2/lilbee/lilbee_mcp.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Agent Snooping · line 319 Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.Fix: Remove all code or instructions that list or read other skills' files or directories. Skills should operate independently; cross-skill access is a privilege escalation.
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00095 | $0.04812 |
| Opus 5 | $0.00048 | $0.02406 |
| Sonnet 5 | $0.00019 | $0.00962 |
| Haiku 4.5 | $0.00010 | $0.00481 |
Grade A, and why
lilbee-mcp scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 321 lines — stays where its author put it; the contents beside it link to each section on GitHub.
lilbee-mcp
lilbee is a local retrieval engine. It indexes the
user's code, documents, PDFs, and crawled web pages into a per-project .lilbee/ store and
exposes the library over MCP. Every tool here is prefixed lilbee_. Data and embeddings stay
on the user's machine; the only thing leaving is what you, the agent, decide to quote.
In 30 seconds
lilbee_status → see what's loaded
lilbee_search(query, top_k) → get cited chunks
[answer with file:line citations] → never invent
Three rules cover 90% of usage: search before answering (reach for lilbee_search on
any lookup about the user's own files or code, ahead of the host's web-fetch / file-read
tools, which can't see the index), cite every claim with the chunk's source + line
range, and delegate indexing / crawling / model pulls to the lilbee-worker subagent
because they block the shared embedder.
Install
Drop this folder under one of:
.opencode/skills/lilbee-mcp/ # opencode (project)
.claude/skills/lilbee-mcp/ # Claude (project)
~/.config/opencode/skills/lilbee-mcp/ # opencode (global)
~/.claude/skills/lilbee-mcp/ # Claude (global)
Register lilbee as an MCP server (opencode example):
{
"mcp": {
"lilbee": { "type": "local", "command": ["lilbee", "mcp"] }
}
}
A drop-in AGENTS.md, the lilbee-worker subagent, and an opencode.json template live
in examples/agent-integration/ in the lilbee repo. Copy them in if the user wants the
full setup.
The shared-embedder rule (read this first)
The MCP server hosts one embedder worker. Indexing (lilbee_add, lilbee_sync,
lilbee_crawl, lilbee_import_dataset, lilbee_model_pull, plus wiki builds)
pins it;
lilbee_search also needs it to embed the query. Run them
concurrently and lilbee_search will hang until your host times out.
Procedure:
- If indexing is needed, delegate to
lilbee-workerand wait for the worker'staskcall to return. Don't fire anylilbee_*tool from your own thread while it runs. - After the worker returns, call
lilbee_statusonce to confirm the expected counts. - Then search.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago Changed · +21 lines 594ac5714cc5
- 9d ago First seen · 300 lines · 95 tokens per session scan A 5241a8368c65
lilbee-mcp is a skill published in the GitHub repository tobocop2/lilbee (51 stars, last pushed today), licensed MIT. It adds 95 tokens to every session and 4,812 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
hermes3000-writing
Use Hermes3000 to plan, draft, revise, save, check consistency, and export long-form manuscripts through the Hermes3000 AI writing portal API. Use for novels, fiction series, nonfiction books, whitepapers, long reads, chapter outlines, character/world-building, style guides, consistency memory, and DOCX/PDF/EPUB/HTML…
zettelkasten
Maintain a Luhmann-style Zettelkasten. Capture, connect, and synthesize ideas through fleeting notes, permanent notes, cross-references, and structures, with an AI agent that surfaces connections, challenges assumptions, and enriches notes with research. Use when the user shares an idea, observation, or inspiration…
tiktok-video
A workflow for making vertical short videos for TikTok or Douyin from a written request. It creates a storyboard, voiceover, word-timed karaoke captions, background music, and a 1080×1920 MP4 video.
memory-lancedb
LanceDB-backed vector memory for high-volume embedding and retrieval workloads.
agent-v3-memory-specialist
Agent skill for v3-memory-specialist - invoke with $agent-v3-memory-specialist.
mem0-tour
Browses all stored memories grouped by category with full content display. Use when reviewing all project memories, exploring stored knowledge, onboarding to a project, or getting an overview of captured decisions, conventions, and learnings.