Connect Claude to the Manual for Courts-Martial
This site runs a free, public Model Context Protocol (MCP) server. Add it to Claude as a connector, and Claude can read every rule, article, and appendix here, run the deadline calculator and the statute-of-limitations checker, and cite the page it drew from, with no copying and pasting. It needs no account, no sign-in, and nothing installed on your computer.
Claude (claude.ai, desktop and mobile)
- Open Customize › Connectors at claude.ai/customize/connectors.
- Choose Add custom connector, paste the server URL above, and name it (for example “Manual for Courts-Martial”).
- For authentication choose No sign-in. Leave the OAuth fields empty.
- Click Add. In a chat, open the + menu › Connectors and make sure it is switched on.
On Team and Enterprise plans an Owner adds it under Organization settings › Connectors, and members then connect. Free accounts can add one custom connector.
Claude Code
claude mcp add --transport http mcm https://www.manualforcourtsmartial.com/mcp
Connectors added at claude.ai are also available in Claude Code when you sign in with the same account.
ChatGPT and Grok
- ChatGPT: Settings › Apps & Connectors › Create, paste the URL, authentication None. Outside Developer Mode ChatGPT uses this server’s
searchandfetchtools, which cover the whole Manual. - Grok: grok.com/connectors › New Connector › Custom, paste the URL.
- Anything else that speaks MCP over HTTP (Cursor, VS Code, Gemini CLI, your own agent) works the same way. The server answers both the current stateless MCP revision and the earlier
initialize-based one.
Try asking
- “What does R.C.M. 305(j) require, and who else cites it?”
- “Charges were preferred on 3 September and the accused was confined on 1 September. Build me the deadline calendar for a general court-martial.”
- “Is an Article 120 offense from March 2019 still within the statute of limitations?”
- “Compare the elements of Article 128 and 128a and list the lesser included offenses.”
- “Search the Manual for ‘unlawful command influence’ and summarise what the rules say.”
What Claude can do with it — 24 tools
Every tool is read-only. The server itself generates this list.
| Tool | What it does |
|---|---|
List corpora list_corpora |
List the five bodies of text on the site (Military Rules of Evidence, Rules for Courts-Martial, UCMJ Part IV punitive articles, the UCMJ statute, MCM Appendix 2.1) with their corpus codes, citation prefixes, groupings (Sections / Chapters / Subchapters) and rule counts. Start here to learn the corpus codes other tools take. |
List rules in a corpus list_rules |
List the rules (or articles, or sections) of one corpus — designator, article number where applicable, title and page URL — optionally limited to one grouping (Section / Chapter / Subchapter number from list_corpora). |
Look up a citation lookup_citation |
Resolve free text — "Article 31(b)", "rcm 305", "Rule 412", "larceny", "speedy trial" — to candidate rules across every corpus, one per corpus where the same number exists in several (Article 128 is both a statute article and a Part IV paragraph). Returns designators and URLs to pass to get_rule. |
Search the Manual search |
Full-text search across every corpus at once. Every word must match (AND); wrap a phrase in double quotes to require it verbatim; a leading corpus word ("rcm 707", "article 120") is used as a ranking hint. Results are grouped by rule with short matching-provision snippets and page URLs, and report the total number of matching rules so you can page with `offset` or narrow the query. |
Get a rule's text get_rule |
The full text of one rule, article or section as Markdown: heading, grouping, page URL, then every provision in document order with its citation (e.g. 305(j)(2), 77.c.(2)(a), 1.3(a)), including Discussion and footnotes. Pass `citation` to return only that provision and its sub-provisions. Identify the rule by corpus + designator, or for corpus ucmj by corpus + article. |
Get cross-references get_cross_references |
Cross-references for one rule: every provision elsewhere on the site whose text cites it (grouped by corpus and rule, with the citing words and a deep link), what the rule itself cites, and the site tools built on it (deadline rows, process stages, lesser-included-offense pairs, punishment-chart lines, the statute-of-limitations checker). |
Get case notes get_case_notes |
The case-law annotations contributors have attached to one rule — case name, reporter citation, court and year, a link where given, and the note text — grouped by the provision each note annotates. These are the site's own annotations, not Manual text. |
Get Benchbook instruction get_benchbook_instruction |
The U.S. Army Military Judges' Benchbook (Electronic Benchbook, chapter 3A, offenses on or after 1 January 2019) pattern instruction(s) for a punitive article: elements, definitions, editor NOTES in place, and the source link. Identify the article by number (e.g. "128") or by its Part IV paragraph designator. |
Get a rule's change history get_rule_changes |
How one rule has changed on the site since the Manual was published: each amended, added or removed provision with its original and current text and the authority (order, statute or case) behind each change, plus links to the full change pages. Provisions edited without change tracking are not reported. |
Check the statute of limitations check_statute_of_limitations |
Apply Article 43, UCMJ (10 U.S.C. § 843) to an offense: the limitations period, its statutory basis, and — given an offense date — the date by which sworn charges must be received (or nonjudicial punishment imposed), with the caveats the statute carries. Offense-specific facts (wartime, capital referral, child victim, enlistment term) go in `conditions`. |
Calculate court-martial deadlines calculate_deadlines |
Given the case dates you know, return one chronological calendar of every RCM, MRE and UCMJ deadline those dates trigger — the computed date, what must happen by it, the governing citation and the quoted words of the rule — plus the counting conventions and the caveats the site's calculator shows. Dates are calendar arithmetic only: excludable delay, tolling, weekends and judicial orders are not modelled. |
List deadline rules list_deadline_rules |
The reference chart of every time period in the RCM, MRE and UCMJ the deadline calculator knows: period, direction, trigger, citation, quoted rule text and notes. Four kinds: deadline (a clock), limit (a punishment ceiling or threshold, never a date), exception (a provision that switches other clocks off) and disclosure (a discovery or notice obligation tied to an event with no number of days, e.g. "before arraignment" — never a computed date). Filterable; topic=discovery gathers the Government's discovery obligations, counted and uncounted. |
Get lesser included offenses get_lesser_included_offenses |
Lesser included offenses under Article 79, UCMJ: the President's Appendix 12A list of "reasonably included" offenses (79(b)(2)) and any "necessarily included" pairs (79(b)(1)) contributors have added, for one greater offense by article or for the whole chart. |
Get maximum punishments get_maximum_punishments |
The maximum punishment chart (MCM Appendix 12): for each offense the authorized discharge, confinement and forfeitures, with the chart's notes, and a link to the governing Part IV paragraph. Narrow by article or free text; the whole chart is long. |
Get sentencing parameters and criteria get_sentencing_parameters |
The sentencing parameters and criteria of R.C.M. 1002 (MCM Appendices 12B, 12C and 12D): the confinement range in months for each of the six offense categories, the category assigned to each offense of a punitive article, and — for the offenses that have criteria in place of a range — the criteria the military judge must consider. Pass `article` for one article's offenses (and its criteria, if it has any) or `category` for every offense in one category. With neither, returns the parameter table and the list of criteria offenses. These are ranges within the maximum punishment, not the maximum itself: use get_maximum_punishments for that. |
Get the court-martial process get_court_martial_process |
The court-martial process from alleged offense to appellate review as an ordered sequence of stages, each with its branch conditions, governing citations, which deadline clocks start there, and a plain explanation; plus the clocks that sit outside the sequence. Optionally one stage by id. |
List court-martial forms list_forms |
MCM Appendix 5: the DoD court-martial forms (DD Form 453, 457, 458 …) in the Manual's order, each with a link to the blank form and to its index page on the DoD forms site. |
List recent changes list_recent_changes |
The site's public revision feed: the most recent tracked edits to Manual text, newest first, each with what changed, the authority cited for it, the editor's note and a link. |
Get an authority and its changes get_authority |
The authorities behind the tracked changes to the Manual's text on this site — executive orders, public laws, statutes, cases — and, for one authority, every provision changed under it (what changed, when, with links to the before/after history and the rule). Answers "what did Executive Order 14130 change?". With no arguments, lists every authority with its change count; `query` matches the label or description (a single match returns that authority in full); `corpus` narrows one authority's changes to a single body of text. |
List MCM editions list_mcm_editions |
Every saved edition of the Manual for Courts-Martial, newest first — 1921 through the current 2024 edition and the 1984 Manual's numbered Changes — each with a link to the full PDF on this site. Use it to point to the edition in force at an earlier date; the PDFs are scans and their text is not served by the other tools. |
Search the CAAF Opinion Digest search_caaf_digest |
Search the headnotes of the U.S. Court of Appeals for the Armed Forces' Opinion Digest — the court's own topic-by-topic summaries of its opinions since the 1999 Term of Court, mirrored verbatim on this site. Every word must appear (prefixes match), "quoted phrases" verbatim; results are grouped by digest topic with the matching headnotes, the case name, M.J. citation, decision date and the opinion PDF at the court. An M.J. citation as the query ("86 M.J. 159") returns that opinion's entries across every topic. |
Get a CAAF digest topic get_caaf_digest_topic |
One topic of the CAAF Opinion Digest in full — every digested opinion under it, by subtopic and term year, with its headnotes — by the topic code a search result or the outline gives ("IIIC26"). Or name a rule instead (corpus with designator or article) to get the digest topics that construe that provision: punitive articles by article number, with the Crimes topics for offenses on/after 1 January 2019 (III.AA) and before (III.A) labelled apart, other rules through the site's curated map — each topic with its most recent entries. |
List recent CAAF opinions list_recent_opinions |
The most recent opinions of the U.S. Court of Appeals for the Armed Forces, newest first — case name, docket number(s), decision date, M.J. citation once published, the opinion PDF at the court, and whether and under how many digest topics the court has digested it. Pass `term` for one term of court's complete table; with no arguments the result also lists the terms available. |
Fetch a document fetch |
Retrieve one document by the id a search result gave ("rcm/305", "ucmj/77", "mre/412", or a digest topic "caaf/IIIC26") or by its page URL on this site. Returns the full rule text — or the full digest topic — as a JSON document {id, title, text, url, metadata}. Equivalent to get_rule / get_caaf_digest_topic; provided for clients that use the search/fetch convention. |
What the server records
The server logs each tool call with the tool name, how long it took, whether it succeeded, the first three parts of the calling address, and the name the client gives itself. This site receives nothing you type into Claude except the arguments of the tool call (a rule number, a search phrase, the dates you ask about), and it does not store them. See the privacy notice.
The Model Context Protocol is an open standard that lets an AI assistant call tools on a server you choose. This server exposes the site’s content and calculators as tools.
Problems connecting? Make sure the URL is exactly https://www.manualforcourtsmartial.com/mcp (https, www) and that authentication is set to No sign-in.
curl -s https://www.manualforcourtsmartial.com/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
One JSON object per POST, no sessions, no SSE. Protocol revisions 2026-07-28 and 2025-03-26 through 2025-11-25.