QMD Markdown Search
ReviewAudited by ClawScan on May 10, 2026.
Overview
This is a coherent local Markdown search skill, but it relies on an external CLI and can index or retrieve private notes, so it should be configured carefully.
Install qmd only if you trust the external CLI. Add only the Markdown folders you want the agent to search, avoid indexing secrets or confidential notes, verify whether LLM reranking sends data to a provider, and run the MCP server only for trusted clients.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The skill itself is instruction-only, but using it requires trusting the external qmd package that will read and index local files.
The skill depends on installing an external GitHub-hosted CLI globally, and that CLI code is not included in the reviewed artifact set.
bun install -g https://github.com/tobi/qmd
Install qmd only from a trusted source, consider pinning or reviewing the repository version, and avoid running it in highly sensitive directories until verified.
Private notes, documentation, or meeting transcripts may become easy for the agent to retrieve into context, including any sensitive or misleading content in those files.
The setup creates a searchable local collection and embeddings from Markdown files, which can persist and be reused in later searches.
qmd collection add ~/notes --name notes --mask "**/*.md" qmd embed
Index only folders intended for agent use, exclude secrets or confidential notes, and understand where qmd stores its index and embeddings.
If qmd uses a remote model provider, queries or document snippets from local Markdown collections could be included in model requests.
The skill documents LLM-assisted reranking, but the artifacts do not specify whether that LLM processing is local or uses an external provider.
query (hybrid with LLM re-ranking - best quality)
Verify qmd's LLM configuration and provider behavior before using `qmd query` on confidential knowledge bases.
A connected client could search or retrieve indexed Markdown content through the MCP tools.
Running the MCP server can expose local search and retrieval tools to connected MCP clients.
qmd can run as an MCP server for direct integration: ```bash qmd mcp ``` Exposes tools: `qmd_search`, `qmd_vsearch`, `qmd_query`, `qmd_get`, `qmd_multi_get`, `qmd_status`
Run the MCP server only for trusted local clients and with collections limited to content you are comfortable exposing to the agent.