Query Markup Documents - On-device hybrid search for markdown files with BM25, vector search, and LLM reranking
Caprock is installed from its publisher's own source and answers where it runs, so this marketplace is not in the path of a single call. There is no address here to send one to, and a panel that pretended otherwise would be showing you an answer we made up. Install it and call it from your own client — the Installation tab has the entry for each one.
What it does
Caprock records every coding-agent session on your computer: Claude Code, Codex, OpenCode, Gemini CLI and others. This server lets the agent you are working with ask it, read-only:
- "Did we decide this before?" searches the decisions and findings kept from earlier sessions (Second Brain).
- "Which session explained the SSO change?" searches the prose agents wrote, across every session.
- "What did the last sessions here do?" gives titles, branches, files touched, pull requests and cost.
- "What has this repo cost this week?" gives spend and tokens per day and per model.
- "Is CI green?" reads GitHub Actions on the default branch and the open pull request.
- "Is anything stuck?" lists running sessions that repeat the same tool call.
12 tools, nothing sent anywhere: it reads the Caprock app running on the same machine. Caprock is free and open source for macOS, Linux and Windows: https://caprock.dev/install
Quickstart
Install it the way its author documents:
The full guide: https://github.com/dspv/caprock/blob/master/docs/mcp.md
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one runs on your machine: your client starts Caprock as a process under your own user, with your files and your network, so the tool surface below is what it can do to your computer rather than to a server somewhere else. Its scan result is on the Security tab — read the surface before you approve it.
Install it the way its author documents:
The full guide: https://github.com/dspv/caprock/blob/master/docs/mcp.md
Tool surface
What the model actually sees. Descriptions are diffed on every release — see version history.
Search the user's Second Brain: decisions, findings and notes kept from earlier coding sessions, across every project or one. Returns up to 30 ranked candidates (any word matches; rarer words, and hits in a title or its English keywords, rank higher; a few synonyms such as PR/pull request, cost/price/spend and decision/rule are matched; typos are tolerated), each with kind, date, project, keywords
| Parameter | Type | Required | Notes |
|---|---|---|---|
| limit | integer | no | How many to return (default 30, at most 50). |
| project | string | no | Keep one project's memories (its name, e.g. the repository folder's name). Omit for all. |
| query | string | no | Words to look for. Empty lists the newest memories. |
The decisions in force for a project, newest first: every decision in Second Brain that no newer one supersedes. Read this before changing direction on something the user already decided.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| limit | integer | no | How many (default 50, at most 100). |
| project | string | no | The project's name (the repository folder's name). Omit for the repository this server was started in. |
One Second Brain memory in full, with what it is linked to (topics, people, projects, related memories).
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | integer | yes | The memory's id, from search_memories. |
The topics Second Brain has memories about, most-linked first, and how many memories each project holds.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| project | string | no | Keep one project's topics. Omit for all. |
What Caprock hands a new session started in a repository: the Second Brain memories it carries, whether the handoff is switched on, and when the passage it would quote was written.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| repo | string | no | A directory inside the repository; defaults to the directory this server was started in (normally the project the agent is working in). |
Recent coding-agent sessions (Claude Code, Codex, OpenCode, Gemini and others) in a repository, newest first: title, status, branch, model, what it is doing, cost and tokens, and a loop alert when one is active.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| all_repos | boolean | no | Every repository instead of one. |
| limit | integer | no | How many to return (default 10, at most 50). |
| repo | string | no | A directory inside the repository; defaults to the directory this server was started in (normally the project the agent is working in). |
One session: its title and description, status, branch, cost and tokens, turns and tool calls, the files it touched and the pull requests it opened.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | The session's id, from recent_sessions. |
The prose the agent wrote in a session (its answers and explanations, not tool output), newest first. Notes marked fragment read as mid-thought.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| limit | integer | no | How many passages (default 5, at most 20). |
| session_id | string | yes | The session's id, from recent_sessions. |
Search the prose agents wrote across every session on this computer — "which session explained the SSO change?" Newest first.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| limit | integer | no | How many to return (default 10, at most 50). |
| query | string | yes | Words to look for. |
Agent spend and tokens per day for a repository (or all of them), with totals by model. Figures are Caprock's, priced from its model table.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| all_repos | boolean | no | Every repository, with a total per repository. |
| days | integer | no | Days back, today included (default 7, at most 90). |
| repo | string | no | A directory inside the repository; defaults to the directory this server was started in (normally the project the agent is working in). |
GitHub Actions for a repository: the newest run of each workflow on its default branch, and the checks of the newest session's open pull request, with the failing jobs named. Needs GitHub connected in Caprock's Settings; read through Caprock, which already holds that connection.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| repo | string | no | A directory inside the repository; defaults to the directory this server was started in (normally the project the agent is working in). |
Running sessions that are repeating the same tool call — a likely loop — with the tool, how many times, the repeated input and what the repeats paid to re-read the conversation.
Takes no parameters.
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under Apache-2.0
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because mcprush bills nothing
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable: mcprush charges nothing to install Caprock or to call it, at any volume.
Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | high | capability blast radius (high) — client exposure if the model is manipulated | −6 |
| verification-discount | repo | publisher verification (public source) — no provenance, but the source is public and inspectable | −1 |
| coverage-honesty | source | inspection depth (source) — how much of the target the scan could see | −0 |
What the scan could actually read
A grade is only as meaningful as its coverage, so the scanner publishes its own depth before it publishes its result.
No tools were enumerated, so prompt-injection, capability and toxic-flow analysis had no tool surface to inspect. To grade a package’s real runtime tools, scan the running server: --command "npx -y <package>".
ui/src/components/PipelineView.test.tsx (33 KB) was not read: over the file-count budget.
Capability — what it could do if the model were manipulated
Tags derived from each tool’s schema and the implementation, not from what the tool calls itself. high is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| No per-tool rows for this release — see the note below. | ||
Toxic-flow graph
The lethal trifecta, checked as a graph rather than as a checklist: untrusted input, a sensitive source and an external sink have to meet before there is a path worth worrying about.
The public result for this release does not print the flow graph, so there is nothing to show here. That is not the same as "no paths were found": what the scan did read is above, under coverage.
Supply chain and provenance
This is the first scan of this surface here, so there is nothing yet to compare it against.
Every result on this tab comes from one deterministic pass over the published package — offline, rule by rule, and auditable line by line above. Same methodology version, same bytes, same score.
Release history
It installs from its repository rather than from a versioned package, so the command cannot be pinned to a release. The history below is for reading; the current release on file is 0.184.2.
No release note was published with this version.
No release note was published with this version.
No release note was published with this version.
- The version this listing was on when mcprush began following its releases. The registry was not asked when it shipped, so this row carries no date.
One review per account, from a signed-in account that does not publish this listing. Nothing else is asked — you can say what you think before you install it. Publishers can reply once per review, and a review can be edited or deleted by whoever wrote it.
Any signed-in account that does not publish Caprock can review it, once — before installing it or after. A review from an account that has installed it is marked as one.
Nobody has reviewed this listing. The rating on the card is the mean of the reviews written here and nothing else, so there is no rating until somebody writes the first — which takes a signed-in account that does not publish it, and nothing else.