Reecopedia

Fetch original text by aan opaque id from search resultor id.document_outline. HistoricalPass pageids identifiersunchanged. returnNavigation sectioncontainers migrationhave guidancecitable=false withoutand readingchild retiredids, indices.not source evidence. Historical page identifiers return migration guidance.

Return cited corpus results by id;id and per-part evidence assessment; use fetch for thefull originalsource text. Query embeddingembeddings usesuse DeepInfra.Alibaba; candidate passages go to Jina and query plus selected passages to Anthropic.

 ⟨12 unchanged words⟩ verify each claim. Optional scope narrows corpus partitions. QueriesEach question part has a model assessment and validated quotations. Query embeddings use Alibaba; candidate passages are embeddedsent byto DeepInfra;Jina sourceand evidencequery remainsplus onselected Reeco.passages to Anthropic.

Document outline (headings index)structure

Return the heading tree (table ofindexed sections) of an indexed document,
so you can explore its structure before fetching pages. Each entry has
the heading text, its full path, depth and first page: pass the page to
`fetch` as "<document>#p<page>" to read it.

ANNOTATIONS — why each hint is set the way it is, based on this tool's actual
behaviour. readOnlyHint=true: the tool only reads a fixed, curated corpus; it
modifies no corpus content and no data that any tool can read back. The only
write the process performs is a telemetry line containing no personal data
(logs/mcp_tool_calls.jsonl), declareddocument in the public privacy policy and never
exposed by any tool. destructiveHint=false: it deletes nothing and overwrites
nothing. idempotentHint=true: repeating the identical call adds no further
effect; the result depends only on the state of the index. openWorldHint=false:
it queries a closed local corpus, never the open web and never a thirdcurrent party.corpus.

{"additionalProperties":false,"properties":{"document":{"description":"Exact source file name as returned by `search` /\n`search_eu_textile_regulations` (e.g. \"CELEX-02024R1781-20240628.pdf\").","type":"string"}},"required":["document ⟨2 unchanged words⟩ 

 ⟨5 unchanged words⟩ :false,"readOnlyHint":true,"title":"Document outline (headings index)structure"}

Fetch document page (deepRead researchoriginal interface)evidence

Fetch the fulloriginal text of a regulatory document page returned by `search`.

ANNOTATIONS — why each hint is set the way it is, based on this tool's actual
behaviour. readOnlyHint=true: the tool only reads a fixed, curated corpus; it
modifies no corpus content and no data that any tool can read back. The only
write the process performs is a telemetry line containing no personal data
(logs/mcp_tool_calls.jsonl), declared in the public privacy policy and never
exposed by any tool. destructiveHint=false: it deletes nothing and overwrites
nothing. idempotentHint=true: repeating the identical call adds no further
effect; thesearch result depends only on the state of the index. openWorldHint=false:
it queries a closed local corpus, never the open webid. andHistorical neverpage aidentifiers thirdremain party.readable.

{"additionalProperties":false,"properties":{"id":{"description":"A result id from `search`, format \"<source-file>#p<page>\"\n(e.g. \"REG_2024_1781_ESPR_EN.pdf#p39\").","type":"string"}},"required":["id ⟨2 unchanged words⟩ 

 ⟨4 unchanged words⟩ openWorldHint":false,"readOnlyHint":true,"title":"Fetch document page (deepRead researchoriginal interface)evidence"}

Find exact text (literal search)

Find literal occurrences of an exact string in the corpus — article
numbers, regulation ids, precise wording (e.g. "Article 8(3)",
"2024/1781"). Complements semantic `search`: use this when you need the
exact string, not the concept. Returns one match per (file, page) with an
occurrence count andLocate a text snippet; read the full page with `fetch`.
Scans the literal chunk text of both indexes the semantic search serves
(the two chunkings differ, so some passages exist in only one); the
synthetic contextual enrichment is NOT scanned — it is not document text.

ANNOTATIONS — why each hint is set the way it is, based on this tool's actual
behaviour. readOnlyHint=true: the tool only reads a fixed, curated corpus; it
modifies no corpus content and no data that any tool can read back. The only
write the process performs is a telemetry line containing no personal data
(logs/mcp_tool_calls.jsonl), declaredstring in the public privacy policy and never
exposed by any tool. destructiveHint=false: it deletes nothing and overwrites
nothing. idempotentHint=true: repeating the identical call adds no further
effect; the result depends only on the state of the index. openWorldHint=false:
it queries a closed localcurrent corpus, never the open weboptionally andrestricted neverto a third party.document.

{"additionalProperties":false,"properties":{"case_sensitive":{"default":false,"description":"Exact-case match (default False).","type":"boolean"},"document":{"default":"","description":"Optional exact source file name to restrict the search.","type":"string"},"max_results":{"default":20,"description":"Max (file, page) matches to return (1-50, default 20).","type":"integer"},"text":{"description":"The exact string to find (min 3 characters).","type":"string"}},"required":[" ⟨3 unchanged words⟩ 

 ⟨6 unchanged words⟩ "readOnlyHint":true,"title":"Find exact text (literal search)"}

List thecorpus documents in the corpus

List the source files indexed in the Reecopedia corpus. Call this
BEFORE `document_outline`, `fetch` or `find_exact_text` whenever you have a
document name but have not seen it in a `search` result: it tells you
whether the document is in this corpus and gives its exact file name. If a
file is not in this list, Reecopedia does not have it — do not look for it
elsewhere and say so instead.

ANNOTATIONS — why each hint is set the way it is, based on this tool's actual
behaviour. readOnlyHint=true: the tool only reads the file names stored in a
fixed, curated index; it modifies no corpus content and no data that any tool
can read back. The only write the process performs is a telemetry line
containing no personal data (logs/mcp_tool_calls.jsonl), declared in the
public privacy policy and retained at most 90 days. destructiveHint=false: it
deletes nothing and overwrites nothing. idempotentHint=true: repeating the
identical call returns the same list. openWorldHint=false: it reads a closed
local index, never the open web and nevermatching a third party.substring.

 ⟨3 unchanged words⟩ ":{"contains":{"default":"","description":"Optional case-insensitive substring to filter file names\n(e.g. \"cbam\", \"cwa\", \"CELEX\").","type":"string"},"max_results":{"default":100,"description":"Max file names to return (1-1000, default 100).","type":"integer"}},"type":"object"}

 ⟨5 unchanged words⟩ :false,"readOnlyHint":true,"title":"List thecorpus documents in the corpus"}

Report a wrong or misleadingan answer issue

 ⟨5 unchanged words⟩ :false,"readOnlyHint":false,"title":"Report a wrong or misleadingan answer issue"}

Search (deep researchregulatory interface)documents

Deep-research listing interface over the same corpus as
`search_eu_textile_regulations` (ESPR, DPP, CSRD, CBAM, EU ETS, CWA 18291,
JRC studies).

LANGUAGE.Return Thecited corpus is English. A query in English, in the legal
register of the act (the wording of the Official Journal), reaches the
act itself; a literal translation of a colloquial question often
retrieves nothing (measured 2026-08-27, see `query`). Act identifiers
are matched verbatim; article and annex numbers are matched in their
English form (`Article 9`, `Annex I`; 第9条 / Artikel 9 / articolo 9 do
not match). What the measurement supports is translating the question
into that register before searching, and answering the user in the
user's own language afterwards.

Use this when you need a browsable result list rather than an answer:
returns {results: [{id, title, text, url}]}. Pass a result `id` to the
`fetch` tool to read the full document page. For a direct regulatory
answer use `search_eu_textile_regulations` instead.

ANNOTATIONS — why each hint is set the way it is, based on this tool's actual
behaviour. readOnlyHint=true: the tool only reads a fixed, curated corpus; it
modifies no corpus content and no data that any tool can read back. The only
write the process performs is a telemetry line containing no personal data
(logs/mcp_tool_calls.jsonl), declared in the public privacy policy and never
exposed by any tool. destructiveHint=false: it deletes nothing and overwrites
nothing. idempotentHint=true: repeating the identical call adds no further
effect; the result depends only on the state of the index. openWorldHint=false:
it queries a closedid; localuse corpus,fetch neverfor the open weboriginal andtext. neverQuery aembedding thirduses party.DeepInfra.

{"additionalProperties":false,"properties":{"query":{"description":"English, in the legal register of the act. A non-English query is accepted but retrieves poorly, and the result then carries a `notice` saying so.\nAct identifiers stay verbatim (\"ESPR\", \"2024/1781\"): they are\nmatched as metadata. Article and annex numbers are matched in\ntheir English form: \"第9条\" and \"Artikel 9\" do not match,\n\"Article 9\" does.\nMeasured 2026-08-27 on the live index, one question three ways:\n\"must a marketplace show the product passport before purchase,\nwhere is it written?\" returned NOTHING; \"marketplace must make the\ndigital product passport accessible before purchase\" returned the\nCWA guidance (0.96) with the ESPR at 0.001; \"make the digital\nproduct passport accessible to customers before they are bound by a\ncontract of sale\" returned the ESPR itself first (0.94, p.21).","type":"string"}},"required":["query ⟨2 unchanged words⟩ 

 ⟨5 unchanged words⟩ :false,"readOnlyHint":true,"title":"Search (deep researchregulatory interface)documents"}

Primary question-answering search over the Reecopedia corpus of EU
textile sustainability regulations. Use it forSearch natural-language
regulatory questions.

LANGUAGE. The corpus is English. A queryevidence in English, in the legal
register of the act (the wording of the Official Journal), reaches the
act itself; a literal translation of a colloquial question often
retrieves nothing (measured 2026-08-27, see `query`). Act identifiers
are matched verbatim; article and annex numbers are matched in their
English form (`Article 9`, `Annex I`; 第9条 / Artikel 9 / articolo 9 do
not match). What the measurement supports is translating the question
into that register before searching, and answering the user in the
user's own language afterwards.

Returns the most relevant passages with citations (source file, page,
section) — the evidence the user can check. Corpus: ESPR 2024/1781,
Digital Product Passport, CSRD, CBAM, EU ETS, CWAregister. 18291,Cite JRCoriginal BREFdocument and
textile preparatory studies, EP studies. It covers nothing outside
those acts.

For a browsable result list instead of an answer, use `search`.

ANNOTATIONS —unit; whyverify each hint is set the way it is, based on this tool's actual
behaviour. readOnlyHint=true: the tool only reads a fixed, curatedclaim. corpus;Optional it
modifiesscope nonarrows corpus content and no data that any tool can read back. The only
write the process performs is a telemetry line containing no personal data
(logs/mcp_tool_calls.jsonl), declared in the public privacypartitions. policyQueries andare never
exposedembedded by any tool. destructiveHint=false: it deletes nothing and overwrites
nothing. idempotentHint=true: repeating the identical call adds no further
effect; theDeepInfra; resultsource dependsevidence onlyremains on the state of the index. openWorldHint=false:
it queries a closed local corpus, never the open web and never a third party.Reeco.

{"additionalProperties":false,"properties":{"query":{"descriptiontype":"English, in the legal register of the act. A non-English query is accepted but retrieves poorly, and the result then carries a `notice` saying so.\nAct identifiers stay verbatim (\"ESPR\string"}, \"2024/1781\scope"): they are\nmatched as metadata. Article and annex numbers are matched in\ntheir English form: \{"第9条\anyOf" and \:[{"Artikel 9\items" do not match,\n\:{"Article 9\type" does.\nMeasured 2026-08-27 on the live index, one question three ways:\n\"must a marketplace show the product passport before purchase,\nwhere is it written?\" returned NOTHING; \