All servers / Research / ResearchZosho MCP server
ResearchZosho MCP server
Deep research MCP server with citations and memory. Starts ResearchZosho for Claude Code, Codex, Gemini CLI or any MCP client that can run npx, and fetches the
Tested by ConnectATDScore 90 out of 100. Tested 2026-11-18 on package version 0.7.8. Package: @wyrdsekai/researchzosho-mcp · source
Score breakdown
| Runs | 18 / 20 | |
|---|---|---|
| Tool descriptions | 20 / 20 | |
| Package safety | 20 / 20 | |
| Maintenance | 20 / 20 | |
| Adoption | 12 / 20 |
What we saw
- Starts and answers the MCP handshake: yes (16.6 seconds including download)
- Tools offered: 69
- Last published: 2026-11-17
- Weekly downloads (npm): 1,613
- License: Apache-2.0 · Direct dependencies: 0
Want this server set up for your team?
ConnectATD builds AI automation and wires up MCP servers like this one. Tell us what you are trying to do.
Talk to ConnectATDTools and how they are described
Agents choose tools from these descriptions. Under each name you can see how many of its parameters are explained.
library_ask4 of 4 parameters describedAsk The Librarian ? the long-term research library. Returns the answer PACKAGE: full relevant entries (findings, shelf articles, captured documents) with states, claim types, sources and disputes, open frontier threads, and holds_nothing=true with NO entries when the library has nothing. Reviewed background: it never overrides direct evidence you gather yourself.
library_search5 of 5 parameters describedSearch The Librarian's catalog (hybrid BM25 + dense, CJK-aware). Returns hits {id, kind, title, snippet, score, state, subjects[]} and next_cursor; use library_get for one entry in full.
library_get6 of 6 parameters describedOne entry in full by id: body, every source with locator and edition, supersedes / superseded_by, subjects, related entries, and the review record {round, reviewer, decision, stale}. An investigation also carries sections[] {heading, chars}, sources[] as rows {n, locator, title, edition, published, fetched, language, same_as}, claims[] (its findings with bodies) and open_questions[]. A write-up can run past what one call may carry: pass section (a heading; "answer" = the write-up without the harness's sections; "workers", "references", "evidence") to get one section, and offset / max_chars for a window of the body.
library_read3 of 3 parameters describedThe captured raw text behind a source locator ? the verbatim path: read the evidence, not only the claim. Returns title, edition, captured_at and the text (truncated at max_chars).
library_established2 of 2 parameters describedWhat the shelves hold on a claim: accepted entries bearing on it, open disputes touching it, and a verdict computed from their states ? established | disputed | not_established. No model composes it.
library_submit7 of 7 parameters describedSubmit one finding as a DRAFT. It enters the librarian's review like any research run and never touches canon directly. A claim with no sources is refused (error no_sources). Requires a patron with write access.
library_add6 of 6 parameters describedRead material into the library: a file, a folder (a mounted network share is a folder), or a url. mode=keep (default) copies the extracted text onto the shelves; the file itself is never copied. mode=link reads the files where they are and keeps only a pointer and a hash: nothing is copied, and if the drive is not mounted the library says so when it reads. mode=survey counts a folder and says what keeping it would take (files, text size, free disk) without shelving anything. For a folder, survey first, then ask keep or link, unless the person has already said. Paths are only accepted from the keeper (the terminal, the pages); a url from any patron with write access. No model runs; to ask a question of what was added, library_research with sources=shelves and the collection.
library_translate20 of 20 parameters describedTranslate a document the person has (txt, md, pdf, docx or epub) into a language and write it out in the same format or another, as a job. The document is cut into its units (headings, paragraphs, list items, table cells, captions) and each is translated in order with the library's glossary; the structure is kept, the layout is not. quality=checked (default) is one pass with the program's checks (every number, address, citation and glossary term on both sides, a length band, no unit dropped), the scores and a revision of what was flagged; quick is the one pass alone; reviewed and thorough add a second reading. What the checks still doubt after the revision is listed in the translation note, and the document holds the translation alone; notes=end numbers those passages and lists them at the end of the document too. Pass notes=end when the person asks for the doubts (the notes) at the end of the document and notes=off when they ask for none; without it the library's setting translate_notes decides, and the answer's notes says which the run was filed with. The answer is the plan: the units, the characters and the time, and translation: who translates, who reads and where the scores are made, with a sentence for the person in translation.said; library_job follows the run, whose result names the file, the translation note beside it and both documents on the shelves. op=glossary lists the glossary, or with term and rendering adds to it. A comic is translated too (manga, manhwa, manhua, a web comic's strips, a comic book, a graphic novel: a CBZ, a CBR, a ZIP or a folder of page pictures by itself; a PDF with comic=on): its pages are read by the comic helper on the library's machine, or on the machine that serves its heavy helpers, and the translation is set into the speech balloons. It is read from, and lettered in, Japanese, Chinese, Korean and the languages of the Latin, Cyrillic and Greek alphabets; from names the language of its lettering, and without it the helper finds the language on a sample of the pages (the answer's comic.from and comic.from_how say which language and how it is known; pass from when the person says it is another). A Japanese comic is read from the right, a Chinese one from the right when its lettering runs in columns, any other from the left: pass reads=right-to-left when the person says the comic reads from the right, as an English, French or Korean edition of a manga does (the answer's comic.reads and comic.reads_how say which way and whether it was said or found). The answer's comic.second_reading says whether the translator reads each balloon's lettering a second time (twice, for a comic that is not Japanese at checked and above when the translator sees pictures), or why the comic helper alone reads it (quick, blind, japanese), or at-start while that is to be found. Where the comic helper is not set up the call is answered with the error confirm, whose message says what a set-up fetches (about a gigabyte) and whose comic_helper lists each piece with its size and licence: show the person the message, and on their yes send the same call with setup=true. Paths are only accepted from the keeper (the terminal, the pages); raw names a document on the shelves.
library_images9 of 9 parameters describedPictures: find candidates on the web for words, keep them on the shelves, and have the model that sees pictures describe or judge them. op=find with words returns found[] (each picture's address, the page it stands on, its title, the engine, its size, a small copy's address and when it was found) from Brave when its key is kept and from the library's SearXNG, merged and told apart; with criteria each is judged (verdicts[]: yes, no or unsure and one sentence why); with keep=true (write access) each is fetched onto the shelves under catalog/images/, given an entry on the shelves with its page and shown at /images/<file>. op=keep with images (addresses, or the found[] objects) keeps them. op=describe with images returns the seeing model's words about each (what it shows, the people, the place, the time, any writing), and its answer to question. op=judge with criteria and images returns one typed verdict per picture. An image is a web address, a kept picture's file name or shelf entry, or (from the keeper) a path on this machine. When no model sees pictures the answer is unavailable, with the sentence that says what the library has and how one is added (researchzosho models test <name> probes an entry).
library_absorb11 of 11 parameters describedAbsorb a conversation the person had with another assistant (ChatGPT, Claude, any chat) as a starting point for research. Give a path (a saved transcript or an export such as conversations.json), a url, or the pasted text. The transcript is shelved as it is; the person's questions join the open questions; the claims the assistant made come back as a list to check and are filed as nothing, since an assistant's say-so is not a source. verify=true files one research run that checks those claims against sources. An export with many conversations takes the first `limit` (default 25) and reports how many remain. Paths from the keeper only (the terminal, the pages).
library_survey8 of 8 parameters describedA thing the person already has, read as a starting point for research: a code repository (a folder, or a git url cloned when git is installed ? it says so when git is missing), a paper (a PDF or document file, a DOI, an arXiv page), a website or product page (any other url), or an issue tracker (a GitHub issues page, or an export file). The kind is detected from the thing, or given. op=survey (the default) reads it, shelves the text, files one draft claim saying what it is, what it says or does, what it claims and what it rests on, and returns numbered directions: research questions the library could answer about it. Nothing runs at that point. op=pick with picks="1,3" files a research run for each direction named; op=do with question files the person's own direction and its run. pick and do take the thing by name (as surveyed).
library_repo7 of 7 parameters describedA code repository as a starting point for research: library_survey with kind=repo. A folder on this machine (keeper only) or a git url, cloned when git is installed; op=survey (the default) reads it, shelves the text, files one draft claim saying what it is and does, and returns numbered directions; op=pick with picks="1,3" files a research run for each direction named; op=do with question files the person's own direction.
library_items15 of 15 parameters describedA list of things ? books, tools, places, an inventory of line items ? as a starting point for research. Give a path (one item per line, a markdown table, or a CSV), a url, or the list as text. The list is shelved as it is so the library knows what the person has; each item is checked against the shelves (held or not); then, through the lens ("{item}: what it is, who made or wrote it, what it is for, and what is known about it" unless the person says what they want to know), as=frontier (default) files a question per item on the open questions, as=runs sends them out as research runs with a lane per item (batches of 8), as=none only reports what is held. Paths from the keeper only.
library_db8 of 8 parameters describedThe databases the library owner gave read access to (SQLite, PostgreSQL, MySQL or MariaDB, SQL Server, MongoDB, DuckDB and data files). op=list names them. op=schema with `database` gives tables, columns, row counts and sample rows: read it before writing a query. op=query runs ONE reading statement: `sql` (a single SELECT or WITH), or for MongoDB a `collection` with a `filter` or a `pipeline` as JSON. Rows are capped, so count and group in the query. The result is saved as a page; cite its locator. Only the owner may use this, and connections are added from the command line.
library_holdings3 of 3 parameters describedWhat the person's own lists hold: a Calibre library, an inventory, a reading list ? every list shelved through library_items. An exact lookup: every word of the query must appear in an entry (title words, an author, a year). Ask it for each candidate to tell owned from not owned; a search snippet is a guess, this is not.
library_who7 of 7 parameters describedFamily history: who, of the people the web shows under a name, is the person in the family. A name is shared, and only somebody in the family can say which entry is their relative. op=list: who waits for an answer. op=show: one person's entries, numbered, each with who they are, their pages, and what those pages say that the family's facts also say. op=answer: the family's word. op=tell: something said of the person, kept in those words for their search. op=find: look one person up now (about half a minute). The answer is the family's own word: show the entries and ask.
library_remove4 of 4 parameters describedTake a report or a claim out of the library for good. id is a report (I-?) or a claim (F-?). For a report, what=all (default) removes it and the claims that are its alone, what=report the report only, what=claims its claims only; a claim another report also cites stays and is named. dry=true shows the plan and removes nothing. Files are deleted, the index and the changes log updated; what the run fetched from the web stays. There is no undo ? retire is the gentler thing for a claim.
library_check11 of 11 parameters describedCheck the person's own draft ? a memo, a chapter, notes (text, markdown, Word, PDF) ? as a starting point. One voice: its definite statements come back as claims to check, its questions join the open questions, and the citations it carries (urls, DOIs, arXiv ids) are fetched onto the shelves; one that cannot be read becomes a source request. Nothing becomes a finding. verify=true files one research run that checks the claims against sources, the draft's own citations included. Paths from the keeper only.
library_reading8 of 8 parameters describedA reading list as a starting point: BibTeX, RIS (Zotero, EndNote), a CSV export, or lines of DOIs, urls and titles. Every entry with a locator is fetched onto the shelves as a collection; one that cannot be read becomes a source request; an entry with a title only is listed so the person can give it a locator or research it by name (library_items). watch=true has the housekeeping re-read the pages. Then library_research with sources=shelves and the collection reads from them. Paths from the keeper only.
library_questions10 of 10 parameters describedA list of questions ? a syllabus, an exam, what the person needs answered ? onto the open questions in order (as=frontier, the default; the housekeeping's explorer works through them at night), or as research runs one each up to 10 (as=runs; the rest are filed as open questions). One question per line. Paths from the keeper only.
library_mcp16 of 16 parameters describedThe OTHER MCP servers whose tools this library's research may use as sources, the keeper's alone: list them, with whether each answers and which of its tools research may call; show one's tools; add one (a command this computer starts, an address, or exa or tavily by name alone); remove one; allow or forbid a tool; let readers use a server; cap its calls a month; test one; call one of its tools as research would. Research calls the tools that read; a tool that may change things stays off unless allowed. A key is never returned. Since 0.7.0.
library_sources10 of 10 parameters describedThe places the library mines for a field ? registers, directories, boards, list pages, feeds, saved searches ? each with what it yields and for whom. op=list (field, for, off=true to include the ones taken off); op=show with id; op=add with address (and for, field, kind, yields, note), after the person said so; op=remove with id takes one off (a saved search is parked); op=restore with id brings it back. Read access for list and show, write for the rest.
library_models11 of 11 parameters describedThe models this library can run on, by name: list them; add one (a name, a host address or word ? anthropic, openai, openrouter, gemini, nebius, huggingface ? a model id, prices per million tokens as in/out/cache, an effort, a monthly cap in dollars); remove one; keep a host's key; test one (whether its server answers now: answers, starting, or down, with a note when no key is kept for its host). The list states the default outright: default (name, model, host, drive) is the entry runs and chats that name no model use, summary says it in one sentence, and each entry's key is a word, kept, missing or not needed (a local server). A key is never returned. Write access.
library_settings4 of 4 parameters describedThe library's settings (write access): op=list shows every setting with what it is for and its value, a key only as set or not; op=set with name and value writes one (name as the key, or without its RESEARCHZOSHO_ prefix; a blank value, or op=unset, unsets it). A program sets a key or an address only when the person gave it in this conversation. op=read_in with value sets the language the person reads reports in (a blank unsets it); read access is enough. op=check asks every outside service whose key is set whether it is accepted and what is left, keeps the answers and returns checks[].
library_share3 of 3 parameters describedThe library reached from outside the house (write access): op=status says how it stands; op=tailscale, op=public (Tailscale Funnel: anyone with the link) or op=cloudflare (a quick tunnel; mail=? lets only listed addresses through) turns the relay on when its program is on the machine, only when the person asked for it in this conversation; op=off turns it off. A missing program is installed by the person on their own screen: researchzosho share, or /share in the chat.
library_my_shelf8 of 8 parameters describedThe person's own shelf: their Calibre library (the calibre setting), the podcasts they follow, what they bought. op=profile: what they went for ? the books they added one by one (a day of fifty or more additions is an import: holdings, not taste) and any they rated themselves ? the subject mix among the liked books with its tilt against the whole shelf, the authors rated well more than once, decades, languages, sample titles, podcasts followed, shops bought from; an imported rating is set aside. op=have with title (and author): whether a book is already on the shelf, in the purchase ledger or on a holdings list. op=liked with words: the rated books on a subject or by an author, best first. op=dna with title: a shelf book's DNA ? the text counted (length, sentences, dialogue, reading grade, the feeling's arc) and the model's reading of it over twelve windows (kind, themes, moods against an ordinary book of its kind, pace, darkness, scale, humour, style, drive, topics), made now when missing; without a title (the owner) the night's reading of the shelf runs now.
library_dna7 of 7 parameters describedThe DNA store beyond the shelf. op=status: how many documents and how many labelled, per kind, and whether the person shares. op=pack-export (the owner) with path writes a pack of the labelled documents (the shared pool's format: the works' layers and provenance, nothing of the person); op=pack-import (the owner) with path reads one in; op=pack-refresh fetches the pack researchzosho.org announces, when there is one. op=podcast-like with words: the podcast episodes like one, among the transcripts the library holds, the same show set aside; op=podcast-read (the owner) reads the transcripts for their DNA now.
library_follow5 of 5 parameters describedThe podcasts the person follows: a show's feed kept as a search, so its new episodes land on the arrival table and the housekeeping transcribes them at night. op=list the shows followed; op=follow with feed (the feed's address; library_search or the podcast search gives it) and every (days, 1 unless set), only when the person asked for it in this conversation; op=unfollow with which (the show's name as listed, or its feed address). The transcripts already in the library stay. Read access for list, write for the rest.
library_prefs5 of 5 parameters describedThe person's preferences file for books: what they want more and less of, each line with its source (drafted from the shelf, learned from the reactions, or said by the person). op=show (default) the lines; op=draft (the owner) writes one from the shelf; op=set with name and value changes a line (weight.subject 2 | roam 0.4 | popularity one|none|known | words "?" | exclude.author|subject|kind|series <value> | prefer.mood|subject|kind|pace|author <value> | avoid.<the same> <value>); op=unset drops one; op=export and op=import with path carry the file; op=learn reads the reactions into it. "Like X" and browsing read it. A program sets a line only when the person said so in this conversation.
library_react5 of 5 parameters describedWrites the person's reaction to a book down: like, dislike, seen or own, by title (and author); about=<title> when the book was a row of a "like" list for that title. A disliked book leaves later lists, a seen or owned one is marked, and the preferences file learns from the likes and dislikes. The owner, and only when the person said so in this conversation.
library_owned12 of 12 parameters describedWhat the person owns: their purchases, imported from their own records (the zip Amazon's data request returns, a mailbox of order confirmations, a bank or card statement as CSV). op=status says what the ledger holds and where each file's records begin; op=find with words (and year, shop) looks something up: what was bought, when, where, for how much; op=import with path reads a file on this machine (the owner only); op=forget with file drops what one file added. The ledger is a reference, never a requirement.
library_bookmarks8 of 8 parameters describedA browser's bookmarks as a starting point: the exported bookmarks file, Chrome's Bookmarks JSON, or lines of urls. Each page is fetched onto the shelves as a collection (folder=? takes one folder); one that cannot be read becomes a source request. watch=true has the housekeeping re-read the pages and keep a new copy when one changes. Paths from the keeper only.
library_meeting10 of 10 parameters describedA meeting transcript as a starting point: WebVTT (Zoom, Teams, Otter) or "Name: words" lines. The transcript goes on the shelves with its decisions; the questions raised join the open questions; the claims made come back to check, with who said them. Nothing becomes a finding. verify=true files one research run that checks the claims. Paths from the keeper only.
library_seeks4 of 4 parameters describedSeeks: the same question researched several times apart (library_research with seek=<group>, each on its own copy of the shelves, on any model), and one report merged from them. op=list the groups and their seeks; op=read a seek's report (group, job_id) or the merged one (group); op=merge files the merge run: the judge reads the seeks' claims in pairs, agreed claims are filed as findings, contradictions as disputed findings with the question on the frontier, what only one seek found is listed; op=drop removes a group's seeks and their shelves (the merged report stays).
library_dossier7 of 7 parameters describedEverything the library holds on a person or an organisation, as one document: the relations the graph has, the claims by the class of their source (public record, own publication, told, supplied, third-party), a timeline, what was told in conversation, the reports to dive into, the open questions, what was looked for and not found, and the subjects it links to. A view over the shelves, written fresh each time; a research run adds to it.
library_export5 of 5 parameters describedA REPORT OR A DOSSIER AS A FILE: a PDF or a Word file holding what its page shows (a report's title, question, the answer's sections, how each point was checked, the open questions and sub-goals, the checks, the claims with their marks and the sources as links; a dossier's sections), written under the library's catalog/exports/ and served by its web pages at /exports/<file>. Returns path, url, file, bytes. A report comes in the language in names, else in the caller's own language to read in, when the library keeps that translation; otherwise as written, and note says so. format=md is the same document as Markdown text. Read access. Since 0.7.0.
library_suggest5 of 5 parameters describedWhat the library suggests the person do next, from the shelves' own events: a forecast whose date passed, a story the shelves have grown past, a question of theirs a claim now answers, a question asked again and again, a report that landed, a dossier not researched for months. Each is classed now, next or hold with the probability they take it up; their marks are the outcomes a Brier score is kept over. op=list (default): the open suggestions, now first. op=mark (write): id, outcome (taken, dismissed, later), note.
library_tell3 of 3 parameters describedWhat a person tells the library about a subject from their own knowledge: what it was like, who was there, an incident, a date. Filed as a claim of class told, in their words, dated, with them as the writer; the questions it raises that the public record could answer join the open questions, so an account can open a search. Answer with 'noted:' and what it raised.
library_bridges15 of 15 parameters describedDiscovery by combination: two areas of the library that no source read together, joined by specific terms both areas' claims share (Swanson's fish oil and Raynaud's). op=run computes the candidate pairs from an area (or the hottest ones) and files up to a few proposals as open questions of type bridge; dry=true only shows the pairs. op=list the open proposals; op=accept files the research run that tests one (its verified claims are the test); op=dismiss drops it; op=settings reads or sets an area's dials (sources, reach, strict, toward, away, since, per_night); op=measure reads the ledger: proposed, kept, dismissed, corroborated.
library_frontier13 of 13 parameters describedThe queue of open questions the housekeeping's explorer researches a few of each night ? op=list returns them in queue order with type, parked, position, tonight, the report that left each (report, report_title, report_fate: kept | waiting | disputed | retired | none), perspective, subjects, language, and similar (the head of a group that reads alike); list takes filters: type, show (queued | parked | all, default all), report (an id or its prefix), fate, who (perspective text), subject, language, q (words); hints=true adds answered {id, title, state} where a claim on the shelves already answers a question. op=add queues one attributed to the patron; op=next moves one to the head; op=later to the tail; op=park keeps one out of the explorer's reach, with why (the reason, listed back as parked_why with the date); op=unpark returns it; op=drop closes one without researching it; op=tidy removes duplicate lines (write access). A report's leftover questions are filed parked.
library_inbox14 of 14 parameters describedThe claims waiting for the keeper's decision ? drafts, and accepted claims whose review went stale ? and the decisions. op=list returns items[] {id, title, state, stale, kind, tier, confidence, writer, date, subjects, language, sources, report, report_title}, oldest first, with filters report (an investigation id or its prefix), subject, kind (extraction | synthesis | interpretation | speculation), tier, confidence, writer, state (draft | stale), language, q (words in the title). op=accept puts claims into every answer from now on; op=dispute files why; op=retire keeps them on disk and out of every answer ? each takes ids[] (or id), or report for every waiting claim of one investigation (write access). op=list also returns reports[] {id, title, state, held_by, date, topic, reasons}: the reports that wait for the keeper's eye because one of their own checks stands against them (the write-up did not answer the question, or part of it; flagged sentences never sent for revision; most cited sources not read); op=accept with such a report's id (I-?) in id or ids accepts it as it stands.
library_serials5 of 5 parameters describedThe searches the housekeeping keeps running on a cadence and reports what is new from ? op=list (each with parked and due); op=add {name, query, every_days} keeps one; op=every {name, every_days} changes its cadence; op=park {name} keeps it without running it, op=unpark {name} puts it back in the rotation; op=remove {name} stops keeping it (write access).
library_subjects1 of 1 parameter describedThe controlled vocabulary: {id, label, broader, narrower[], count}. Facets (the part before '--') are the broader terms. Use an id with library_search's subject filter.
library_changes3 of 3 parameters describedRecall notices: everything that happened to findings and investigations after a cursor ? added, state changes (retired, disputed, accepted), edits, supersessions. Re-check what you cited with one call. Pass the previous next_cursor as since; 0 = from the beginning.
library_request_access4 of 4 parameters describedAsk to be let into this library. No access is needed to ask. Returns a request id and a claim secret (shown once); the owner approves or denies; collect the outcome with library_access.
library_access10 of 10 parameters describedWhat became of an access request (op=collect, the default, with request_id and claim): pending, denied (with the reason), or approved, in which case the token, once, when the claim secret is right. The keeper's side, for the person who keeps the library alone (anyone else is refused): op=readers lists who may read or write and the default; op=allow with did, level (deny, read or write) and name sets a reader's level; op=remove with did removes one; op=invite with name and level makes a link that signs a new person in once, shown once; op=requests lists who asked to be let in; op=approve with request_id and level, or op=deny with request_id and reason (or did), answers one; op=keys lists the keeper's keys for programs; op=key with label and level makes one, shown once; op=revoke with key_id revokes one. Nothing here is done unless the person asked for it in this conversation.
library_subscribe4 of 4 parameters describedRegister a webhook: every change on this library's feed (retired, disputed, revised, supplied, ?) is POSTed to your address, signed with a secret (X-ResearchZosho-Signature: sha256=hmac). The changes feed stays the source of truth. Needs a named patron.
library_unsubscribe2 of 2 parameters describedRemove a webhook you registered.
library_research19 of 19 parameters describedA RESEARCH RUN: send a question to The Librarian's workers. The library's own runner plans sub-questions (yours, if you pass them), researches each in parallel with web search + fetch noting every fact with its source, has a critic decide on a second round, and writes the investigation in sections; the result enters the library as a draft investigation attributed to you and the review crew extracts findings from it. Returns a job_id to poll with library_job. Write access. Takes minutes to hours ? do not wait inline. A field that acts only when asked for, such as family history, joins the run only when field names it; a question that looks like one of its questions is researched as ordinary research, and the result then carries suggestion {field, why, how}: what the field's mode does and how to ask for it.
library_candidates10 of 10 parameters describedCANDIDATES FOR A VIDEO YOU REMEMBER but cannot name: give the name of the thing made or shown, two to four words, in each language the video may be in, and what you remember of the video itself as short phrases in your own words (remembered), never a search, and the library enumerates for itself: deep YouTube searches of its own plain phrasings, the pages where other people describe a video in the words remembered, every upload of the channels those surface, the YouTube Data API in date order when a key is set, other search indexes, and your own watch history first. It drops what a machine can check (rejected ids, uploads outside the dates, lengths outside the minutes) and answers with the pool best fit first, a video whose title, description or linking page holds more of what is remembered before one that holds less, forty a page, with no popularity in it. About ten minutes; the pool is kept, so page 2 fetches nothing. Then give the first five to library_look. Since 0.7.0.
library_sift7 of 7 parameters describedSIFT THE CANDIDATES for a video you remember, the funnel between the pool and the five a look can afford: every candidate's thumbnail, then the first storyboard sheet of those that pass, each judged yes, maybe or no by the model that reads pictures, keeping yes and maybe; the thumbnail for what one still can show (the thing, its colours, a tag or label, a person shown or not), the first sheet for what you remember about the start; what no picture shows is left to the look. Give the pool's key from library_candidates (from) and what you remember (what), or the videos themselves (videos). The thumbnails go nine to a call as one numbered grid, under a second each, and a sheet takes about ten seconds on a local model; it says how many were not reached, and the same call with page=2 goes on; nothing is judged twice; one sift runs in a library at a time, and a second call while one runs comes back at once saying so. Then give the first five survivors to library_look. Since 0.7.0.
library_patents7 of 7 parameters describedPRIOR ART, THE POOL: every patent publication of a classification in a span of years, enumerated from the European Patent Office's own records page by page to the end, never a search engine's first page. Give the classification symbols the invention sits in (each a CPC group with its slash, like A47G19/22), the years, and the invention's features, each a short phrase with the words and synonyms that would name it in a title or an abstract, in every language you can give; never a query. The library says the office's own count, keeps one line a patent family with the other members on it, counts which features each family's titles and abstracts name, and answers with the pool ordered by more features named, then the rarer feature, then the older, forty a page, with no citation count, family size or other popularity in the order. A class of a thousand publications takes two to five minutes and about ten megabytes of the office's free four gigabytes a week; every page is kept, so page 2 and the reading fetch nothing. It needs the office's free key (researchzosho records key epo-query <key> <secret>). Then give the key to library_patent_abstracts. Since 0.7.0.
library_patent_abstracts4 of 4 parameters describedPRIOR ART, THE ABSTRACTS: the cheap reading between a pool of patents and their claims. The library's model reads the abstract of every family in the pool, best fit first, twenty to a call, and says of each feature yes, maybe or no; the families that pass come back best first with the counts. Give the pool's key from library_patents (from); the features are the pool's. About ten seconds for twenty abstracts on a local model; it says how many were not read, and the same call with page=2 goes on; nothing is read twice; one reading runs in a library at a time. Then give the key to library_patent_claims. Since 0.7.0.
library_patent_claims6 of 6 parameters describedPRIOR ART, THE CLAIMS: the claims of patents from the patent office's own full text. For a pool (from), the claims of the families that pass by their abstracts, best first, and for each feature the claim that comes closest, quoted with its number and the office's link, or that none that was read meets it. For publications given by number (publications), each one's record, its classes and its claims. Where the office holds no text for a publication (it holds none for the United States, China, Japan or Korea) the family member whose text it does hold is read, and the answer says which; where no member has a text there, the claims are read from Google Patents' copy of the office's record at the publication's own address, never by a search there, and the answer says so; a publication with claims at neither is read by its abstract, said to be an abstract. A quotation is kept only when its words stand in the text as read; a status worded on Google's page is given as Google's own reading. About twenty seconds a family; the same call with page=2 goes on; nothing is read twice. Since 0.7.0.
library_look4 of 4 parameters describedLOOK AT A VIDEO at any address: the library reads its frames and transcript into three tracks, what was said, the words shown on screen and what is seen, keeps them, and answers what you ask of it with found, maybe or no and the second. About seven minutes per ten minutes of video on a local model; sparse=true reads a third of the frames. Since 0.7.0.
library_records2 of 2 parameters describedTHE PUBLIC RECORDS OF A HOME by its address: one line saying which source answered (the city's or county's own portal, a state's parcel layer, a written reader) or that none is known and what was looked at; the sales or values and the permits on the street; the Census tract's figures when a key is set. Seconds to a minute; once a county, the library researches where it publishes its records. Since 0.7.0.
library_board4 of 4 parameters describedAN EMPLOYER'S OWN CAREERS BOARD and its openings: the hiring system the employer uses, found from its careers page or from the organisation it belongs to, and the openings read from it, cut to the towns you name. Seconds to a minute. Since 0.7.0.
library_video6 of 6 parameters describedTHE VIDEOS THE LIBRARY HAS READ, beyond the three tools that find one (library_candidates, library_sift, library_look): op=status (default) says how many videos are held and with which tracks; op=saw with query finds the moments where words were said or shown across every video held, with the second and a link; op=picture with description files the picture in the person's head (what they remember seeing) as a run that reads the held videos' frames for it (write). The ops look, candidates and sift are taken too, as those tools take them.
library_home17 of 17 parameters describedTHE HOME SEARCH beyond one address's records: a person looking for a home to buy or rent. op=list the searches; op=start with name; op=read (search) the topics decided, guessed and open; op=tell with words, read by the model for the topics they settle; op=want with dimension, state (decided, guess, open) and want; op=keep_out with words (drop=true takes it back); op=search files the search run (more=true: the next page of listings); op=listings (show) the kept listings through the cut, in tiers, with the ones set aside and why; op=react with id and reaction (like, pass, or blank) and why; op=card with id reads the records and researches one listing; op=add with words (url, tenure) keeps a listing the person found; op=records with address is library_records. Read access for list, read, listings and records; write for the rest.
library_career21 of 21 parameters describedTHE CAREER SEARCH beyond one employer's board: a person looking for work. op=list the people kept; op=provide with what (a resume, a LinkedIn file, a path for the keeper, or words) and name; op=evidence the facts kept; op=reading the model's reading of the working life (write); op=shape (words) how seats are filled for them; op=wants what they want, drafted again; op=questions the interview, op=answers with answers (a number to its answer) and op=resettle; op=company and op=keep_out with words (drop=true takes it back); op=search files the search run and op=more reads more; op=openings (show) and op=companies (show) what was found; op=react with id or company, reaction (like, pass, or blank) and why; op=company_state with name and state (kept, dropped, or blank); op=watch (on) the standing search; op=cut (titles, titles_also, titles_not, places, remote, remote_where, prefer_titles, prefer_places, again) the cut; op=want with dimension, state and want; op=board with employer is library_board. Read access for list, evidence, questions, openings, companies, cut and watch without changes, and board; write for the rest.
library_job6 of 6 parameters describedOne job from the ledger ({job}: state queued|running|done|failed|stopped|offered, elapsed_s, result when finished), or your jobs (active[], finished[], paused) when job_id is omitted. op=stop {job_id} stops one run: queued, it never starts; running, it ends within seconds. op=pause holds the runner (queued runs wait, a running one holds at its next turn); op=resume lets it go (write access). op=choose {job_id, choice} answers a run held for one of its named choices, by number or by name; op=allow and op=decline answer for the person on a run the service held for their yes about help: a program sends any of the three only after it asked the person (write access).
library_crews2 of 2 parameters describedTHE HOUSEKEEPING: op=plan (default) says what its next run will do, from the same files and plan it uses ? the kept searches due, the open questions the explorer takes and in what bundles, the counts, the weekly and monthly extras ? as runs[] and as text; op=run files its tasks as a job now, which the service runs behind any run in progress (write).
library_spend1 of 1 parameter describedWHAT THE MODEL CALLS COST, at each model's list prices: the library's arithmetic on the servers' usage counts, not a bill. This month and in all, by host and by named model, with the cap where one is set, as hosts[] and entries[], and by what the calls were for (chat, nightly, research, translation, picture, other) as uses[]; and as text. Write access, as the settings.
library_status1 of 1 parameter describedlibrary_id, library_name, contract, counts by kind and state, last_updated.
library_help2 of 2 parameters describedThe library on ITSELF: what ResearchZosho is, which version is running, what it can and cannot do, how to do something in it, and why something does not work. Returns its live state (version, what this library holds, the model and whether it answers, the web search, the fields that are on, the chat's commands, the pages) and the sections of the program's own documentation that answer the question, with where to ask what neither answers. Use it for any question about the library or the program rather than answering from memory.
library_sharpen2 of 2 parameters describedSHARPEN a rough research question BEFORE running it: the library looks at what it already holds, asks who studies the question and what each would ask, and answers with the question rewritten, the assumptions it made (read them: a model sharpens by narrowing), a brief (scope, sub_questions, sources, deliverables), what is already held, at most three questions_for_you, a mode and a size (quick | full), and research_question ? the text to pass to library_research as it is or after the person edits it. Proposes only; never runs research. Needs a model drive; two or three calls.
library_explain6 of 6 parameters describedA READING AID, never a record. id + rung re-explains an entry for a reader at that rung: beginner (new to the field), familiar (knows the field but not this work) or written (the sources' own register). term (+ in, the entry it appears in) explains a term as it is used there. Shelves only: written from the entry, its findings and its captured sources; every paragraph is read back against what it cites and marked when the shelves do not support it. grounding is shelves | thin | none; when none, offer carries the library_research call that would fill the gap ? pass it as it is (quick: true runs it now, ahead of the night's work), then call library_explain again. terms[] are words the reader may need next; each is a further library_explain. Needs a model drive the first time; cached after.
library_perspectives3 of 3 parameters describedBefore researching: WHO studies this question and what would each of them insist on asking (STORM's move). Returns perspectives [{perspective, why, questions[]}] and sub_questions[] ready to pass to library_research. One or two searches and one judge call; needs a model drive.
library_map4 of 4 parameters describedThe GRAPH around a focus: a person, place, event, document, organisation, work or concept by name, or an entry id. Returns the neighbourhood ? nodes {id, kind, label, also[], wikidata} and edges, where every edge IS a finding {from, predicate, to, finding, state, confidence, disputed} ? plus the open questions touching it. This is how a family shelf and a computer-science shelf meet at one person.
Tested in a sandbox with an empty environment: no API keys, a throwaway folder, nothing from a real account. A passing test shows the server runs and responds. It is not a security audit.