QuestLaw Library MCP
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| QUESTLAW_CONFIG_DIR | No | Directory where consent is recorded. | ~/.questlaw-library-mcp |
| QUESTLAW_VAULT_FILE | No | A .qlvault export file, or a folder from which the server takes the newest export. | ~/Downloads |
| QUESTLAW_KEY_ACCOUNT | No | The secret store account holding the account key. Use 1-64 letters, digits, '.', '_', or '-'. | default |
| QUESTLAW_ALLOW_ENV_KEY | No | Set to '1' to allow QUESTLAW_RECOVERY_CODE to supply the account key. Otherwise that variable is ignored. | |
| QUESTLAW_RECOVERY_CODE | No | The 43-character account key itself. Ignored unless QUESTLAW_ALLOW_ENV_KEY is set to '1'. | |
| QUESTLAW_DISCLOSURE_ACK | No | Set to 'i-understand' to grant consent from the environment. | |
| QUESTLAW_MAX_RESPONSE_BYTES | No | Byte limit per tool result, kept between 8 KiB and 1 MiB. | 65536 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| read_library_textA | Read exact stored source text, notes, analysis, quotations, or relationship notes in small pages. Copy cursor from a readText/readMatch/readMore/nextCall object returned by another tool. Follow nextCall when more text is needed; hasMore refers only to this selected target. Ranges are UTF-16 code units, end exclusive. A page or search excerpt is not a complete document. No manual download is needed to continue stored text. stale_cursor means repeat the original discovery call. |
| library_overviewA | What this library contains: how many authorities, which projects, tags and courts they are filed under, how much captured source text there is, and how old the snapshot is. Call this first. It costs one call and tells you which of the other tools can actually reach what you are after. |
| list_casesA | Every authority in the library, with no query and no filter required. This is how you enumerate the library: what is in here, how big is it, summarise all of it. Optionally narrow by project ("none" for the unfiled ones), court, or tag, and sort by title, recency, year, or quote count. Pages through offset, and every row carries a one-line summary drawn from the user's own notes. |
| search_casesA | Find saved authorities, ranked by relevance across title, citation, tags, notes, analysis, and quote text. Multi-word queries work; so do "quoted phrases", which are required rather than preferred. Citations match in any spelling (603 U.S. 369 = 603 US 369). Narrow by project, court, or tag rather than by putting a court name in the query. Each result carries a summary and the passage that matched, so you can usually judge relevance without calling get_case. Omit query to browse; page with offset. To walk the whole library, use list_cases. |
| get_caseA | One authority with bounded previews of the user's notes and analysis, saved quotations with its pincite, the projects it is filed under, how they connected it to other authorities (related), and any captured source text available for it. Search and list rows already carry a summary, so reach for this when you need the whole note, the quotations, or those connections -- not to identify a case. Follow readText to read all stored research, or readMore for clipped fields. Preview limits never require a manual download. |
| search_quotesA | Search the text of every saved quotation, ranked. Use this when the user wants language to quote rather than an authority to cite: each hit carries the passage, the authority it came from, and its pincite. Quote a phrase to require it verbatim. This searches only what the user chose to save; to search the full text of captured opinions and statutes, use search_library. |
| search_libraryA | One ranked search across everything: saved authorities, saved quotations, and the full captured text of statutes, regulations, rules, and opinions. Start here when you do not already know which of those holds the answer -- captured source text is reachable no other way except by a document id you would have to know in advance. One row per authority or captured document: its other matching quotations or sections are listed under the row as alsoMatched. Pages with offset. Omit query to browse what exists of each kind. |
| list_project_casesA | Every authority filed under one project or matter. Project ids come from library_overview. Authorities filed under no project are unreachable here: use list_cases, with project "none" for exactly those. |
| get_source_textA | The captured text of a statute, regulation, court rule, or opinion. Document ids come from get_case (capturedSources) or a search_library source hit. It does not have to arrive whole: outline lists the sections and their sizes, query returns only the sections matching it, maxChars caps how much text comes back, and offset pages through sections. Follow readText for exact complete text or each section's readMore for its clipped tail; nextOffset alone does not cover clipped text. Query uses the same phrase, shorthand, and word-form rules as search_library. Use matchedPassage.readMatch to read around a late match. Stored text does not require a manual download. |
| find_citation_mentionsA | Given draft text, report every place it cites an authority already in the library, in order. Citations match in any spelling (603 U.S. 369 = 603 US 369), with or without a pincite or year, and a citation used twice is reported twice. Use it to check a draft against the research behind it -- and to find the citations in it that the library does not have. |
| reload_snapshotA | Re-read the newest export from disk. Use after the user exports a fresh backup, or to retry after a load failure. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 11 tools
The set has three clusters of near-neighbors: four search tools (search_cases, search_quotes, search_library, find_citation_mentions), three listing tools (list_cases, list_project_cases, and search_cases with an omitted query), and three retrieval tools (get_case, get_source_text, read_library_text). The descriptions do explicitly cross-reference each other and state when to prefer one over another, which prevents outright misselection, but the boundaries require careful reading rather than being self-evident from names.
Nearly all tools follow a snake_case verb_noun pattern (list_cases, search_quotes, get_source_text, reload_snapshot, find_citation_mentions). The only deviation is library_overview, a noun-only name, which is a minor break in an otherwise predictable convention.
Eleven tools is comfortably within a well-scoped range for a research library covering discovery, search, retrieval, and reading. There is mild redundancy from the parallel search and list variants, but each has a stated distinct job, so nothing feels padded.
The read path is thorough: overview, enumeration, three search granularities, previews, full source text, and snapshot reload. However, the surface is read-only — there is no way to add or update authorities, notes, quotations, tags, or project filings, so an agent cannot complete any write-side lifecycle and must fall back to the user.