xbrlkit
PYPI · XBRLKIT
Read SEC EDGAR and XBRL filings with an AI: 18 tools over the filing itself. Local, free, no account
Available components
Last scanned (9 hours ago)
Scores higher than 60% of PyPI servers
Category Finance and Payments
How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →
Supply Chain Security50
- Malware scan not yet available for this package.Unverified
- No known CVEs affecting this package version or its production dependencies.Pass
- Runs hatchling.build at install time, a recognised build step with no custom scripting around it. View diagnostics → Pass
- 4 of 45 dependencies flagged as unhealthy. View diagnostics → Partial
Provenance & Transparency97
- Source repository is publicly reachable at the declared URL. View diagnostics → Pass
- Cryptographically verified build provenance (signed, bound to RoboFinSystems/xbrlkit). View diagnostics → Pass
- Clear OSI-approved license (MIT).Pass
- Actively maintained (last published 2 days ago).Pass
- Disclosure check failed: no security disclosure policy was found in the source repository. See how to fix → Fail
Schema Quality & AI Usability60
- AI-judged instruction clarity (excellent).Pass
- Context-footprint check failed: tool/resource definitions use about 5087 tokens (~282/item across 18 items; 18 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management40
- Stability observed for 12 of 30 days with no destabilising changes; credit accrues until the full window elapses.Partial
Tool Coverage100
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 100% of tool parameters carry a description.Pass
Tool Safety75
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- 0 of 1 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation; "unload_filing" implies "drop" and declares no destructiveHint at all, which the MCP spec reads as destructive by default. See how to fix → Fail
- An AI judge read all 19 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
- Implements a current MCP spec version (2026-07-28).Pass
Checks we ran (15)
- Known CVEs
- Install scripts
- Dependency health
- Source repository
- Provenance
- License
- Maintenance
- Security disclosure
- Schema quality
- Stability
- Tool coverage
- Injection markers
- Destructive annotations
- Judged manipulation
- MCP protocol
Checks we could not complete (1)
These are gaps on our side, not findings about this package. We only credit what we can confirm, so a category made up entirely of them scores 0.
- Malware scan in Supply Chain SecurityMalware scan not yet available for this package.
Does not apply to this package (2)
- Endpoint SecurityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
- Transport & ReachabilityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
The score does not measure everything. What the score does not measure →
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 20 Nov 26 −14
- Malware scan: pass → unverified ▼ security
- 19 Nov 26 +26
- Malware scan: unverified → pass ▲ security
- Known CVEs: unverified → pass ▲ security
- Dependency health: unverified → 0.97 ▲ functional
- Package version: 0.21.7 → 0.23.1 functional
- 18 Nov 26 −25
- Malware scan: pass → unverified ▼ security
- Known CVEs: pass → unverified ▼ security
- Dependency health: 0.97 → unverified ▼ functional
- 17 Nov 26 +15
- Malware scan: unverified → pass ▲ security
- 16 Nov 26 0
- Package version: 0.21.6 → 0.21.7 functional
- 15 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 17 to 20. That category is still filling its 30-day observation window: 5 days of observed history at the previous scan, 6 at this one. The score rises as the window fills, whether or not the server changes.
- 13 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 10 to 13. That category is still filling its 30-day observation window: 3 days of observed history at the previous scan, 4 at this one. The score rises as the window fills, whether or not the server changes.
- 12 Nov 26 −4
- Malware scan: pass → unverified ▼ security
- Known CVEs: unverified → pass ▲ security
- Dependency health: unverified → 0.97 ▲ functional
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 21 Nov 2026 · Analysed pypi/xbrlkit@0.23.1
Provenance Verified
A signed build attestation was found and verified, binding this exact artifact to the source repository it claims to come from.
| Result | Verified |
|---|---|
| Ecosystem | pypi |
| Reason | Verified |
| Discovered via | Registry attestation endpoint |
| Source repo | RoboFinSystems/xbrlkit |
| Certificate issuer | https://token.actions.githubusercontent.com |
| Certificate SAN | https://github.com/RoboFinSystems/xbrlkit/.github/workflows/create-release.yml@refs/heads/main |
| Rekor log index | 3987836022 |
| Predicate type | PyPI publish attestation https://docs.pypi.org/attestations/publish/v1 |
| Subject digest | sha256:2691adc786eee405123fa958a328e58bc4b71332af16b391be49b2ba150c9ed0 |
Background: How many MCP packages publish verified provenance →
Install scripts 1 script
| Hook | Tier | Command |
|---|---|---|
| build_backend | Known build tool | hatchling.build |
Background: Why install scripts are a supply-chain risk →
Dependencies 45 packages
| Packages resolved | 45 |
|---|---|
| Stale | 3 |
| No linked repository | 1 |
| Tree resolution | Complete |
Background: SBOMs and build attestations, explained →
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
calculation ~168
What sums to a total: the calculation-linkbase children of a concept with their weights, and per period the reported total, the sum computed from the children's consolidated facts, and the difference. Use for 'which line items make up operating expenses' or to check a subtotal. A concept that is a child rather than a parent reports what it contributes to.
| Name | Type | Req | Description |
|---|---|---|---|
| concept | string | yes | The total's concept qname or name, e.g. us-gaap:OperatingExpenses. |
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| period_end | – | – | A period end date (YYYY-MM-DD) or a year (YYYY). |
| role | – | – | Restrict to calculation networks whose name contains this. |
No output schema declared.
No examples provided.
describe_filing ~168
How a loaded filing is laid out: entity, form and fiscal context; fact counts; the reporting periods with the `key` fact_grid and statement use; units; the primary statements by role and the notes by name (`disclosures` lists a note's blocks; under `--pure`, every presentation network by role); the dimensional axes present; and the text sections with their character offsets. For an 8-K it also returns the SEC's own item codes and what they mean (2.02 = an earnings release) with a note saying where that filing's substance actually is. Call this first — never guess names or keys.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
No output schema declared.
No examples provided.
disclosures ~172
The filing's disclosures — its information blocks as families, read from the filer's own role titles: each note with its policies, its tables and every details table; each statement with its parenthetical; the cover page. With no topic, the list — one row per family with block and fact counts. With a topic ('leases', 'income taxes', 'cover'), that family's index: each block's id, level, fact count, axes and text blocks. The cheap call: use it to find the block a question is about, then information_block for that block alone.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| topic | – | – | A family's title or part of one; omit to list them all. |
No output schema declared.
No examples provided.
documents ~174
What else was filed with this filing — the exhibits, and any second document the content actually lives in (an 8-K's EX-99.1 press release, a 13F's INFORMATION TABLE of holdings). An 8-K's EX-99 exhibits are also in its text while the text tools read whole documents; this lists them with the rest. Costs one small fetch the first time and nothing after. Every document carries its URL and says whether read_document can read it: a PDF or an image is listed with its address so a caller that can open one may fetch it directly. The XBRL package and the SEC's own rendered copies are not listed.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
No output schema declared.
No examples provided.
export_filing ~340
Write the loaded filing as one of xbrlkit's projections into /tmp/xbrlkit/output (the server's --out-dir) and return the path: `clawdog` (ClawDog JSON-LD), `holon` (RDF / JSON-LD, opens in the RoboSystems holon viewer), `tavi` (the Project TAVI compiled model, JSON), `oim` (xBRL-JSON), `lpg` (the property graph as a LadybugDB database, the one to query; needs the lpg extra), `icebug` (experimental: the same graph as an icebug-disk directory any LadybugDB reads in place; portable, but slow for multi-relationship queries on a large stack; needs pyarrow), or `model` (the parse itself as JSON — load_filing reloads it without Arelle). For `lpg` and `icebug`, `filings` stacks several loaded filings into one graph: what they share (the entity, periods, units, labels) is stored once and every fact stays on its own report.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| filings | – | – | Several loaded filings' ids (tickers or accessions also work) to stack into one graph; lpg and icebug only. Replaces `filing`. |
| format | string | yes | The projection to write. |
| name | – | – | The stacked graph's file name, without extension; defaults to `<ticker>-stack`. |
No output schema declared.
No examples provided.
fact_grid ~301
Values for one or more concepts across the filing's periods. By default the consolidated total per concept and period (facts with no dimensional qualifier), keeping the most precise of duplicate tags, newest period first. `period_end` (YYYY-MM-DD or YYYY) narrows the periods; `period_type` takes instant or duration (and, under the product profile, annual / quarterly buckets). Set include_dimensions, or name an axis / member (substring match), for segment and member breakdowns. Concepts may be qnames or bare names (`Revenues`).
| Name | Type | Req | Description |
|---|---|---|---|
| axis | – | – | Keep facts on an axis whose qname contains this (implies dimensions). |
| concepts | – | – | The same as elements; either name is taken. |
| elements | – | – | Concept qnames or names, e.g. ['us-gaap:Revenues', 'NetIncomeLoss']. |
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| include_dimensions | boolean | – | Return member breakdowns beside the consolidated totals. |
| limit | integer | – | Rows to return (max 500). |
| member | – | – | Keep facts whose member qname contains this (implies dimensions). |
| period_end | – | – | A period end date (YYYY-MM-DD) or a year (YYYY). |
| period_type | – | – | instant | duration; product profile also annual | quarterly | semi_annual | nine_months |
No output schema declared.
No examples provided.
information_block ~336
One information block read whole — a primary statement, the cover page, or any details table: rows in the filing's order with the consolidated value per period, the same rows broken out by the block's own axes (a segment table's segments, a maturity table's years), its calculation arcs with a footing check per period, and its tagged text blocks with the offset read_text pages from. THIS IS THE EXPENSIVE CALL: a details table with many members runs 15,000–30,000 characters. Call disclosures first for the family index, and narrow this one with `member` or `periods` when part of the block answers the question. `block` is an id from disclosures or describe_filing, a name or part of one, or a role URI. A `truncated` response continues from its `next_offset` as `offset`.
| Name | Type | Req | Description |
|---|---|---|---|
| block | string | yes | Block id, name (or part of one), or role URI. |
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| max_members | – | – | Member breakdowns to keep, most facts first (max 200). By default as many as fit the response budget; a row is never left blank. |
| max_rows | integer | – | Rows to return (max 400). |
| member | – | – | Keep only member breakdowns whose key contains this. |
| offset | integer | – | Rows to skip: the `next_offset` a truncated response returned. |
| periods | – | – | Period keys, end dates or years to keep as columns. |
No output schema declared.
No examples provided.
list_filings ~20
The filings this server has loaded, with their ids.
Input schema present but exposes no named parameters.
No output schema declared.
No examples provided.
load_filing ~422
Load a filing into the server and return a receipt — the entity, the form and fiscal context, and the fact counts. The layout is describe_filing's job: call it next for the period keys, the networks and the text sections. `source` is a local path (an inline XBRL .htm, an XBRL instance .xml, a filing directory, a .zip or .xbri package, or a JSON report: a ClawDog report, a TAVI compiled model, a holon, or a model.json written by export_filing), an http(s) URL — an XBRL document Arelle can load, or a JSON report published as an artifact — an EDGAR `cik:accession` (e.g. `1045810:0001045810-26-000021`), a ticker with an optional form (`NVDA`, `NVDA 10-Q`) for the latest filing of that form, or — outside the SEC — `lei:<LEI>` for that filer's latest filing on filings.xbrl.org, or one of that index's filing ids (e.g. `213800H2PQMIF3OVZY47-2022-03-31-ESEF-GB-0`). Any XBRL taxonomy loads: US GAAP, IFRS, ESEF, ACFR. A zip or directory with no report loads as a taxonomy; the receipt names the entry point, and `entry_point` picks another. Takes seconds to a minute; the taxonomy cache makes repeat loads fast, and a JSON report loads at once, with no Arelle and no taxonomy fetch.
| Name | Type | Req | Description |
|---|---|---|---|
| entry_point | – | – | A taxonomy's entry point, by path, file name or name from the receipt's `taxonomy`; omit for the default. |
| filing_id | – | – | An id to refer to the filing by; defaults to its accession or file name. |
| source | string | yes | Path, URL, cik:accession, or ticker. |
No output schema declared.
No examples provided.
read_document ~135
Read up to 8000 characters of one of the filing's other documents, from a character offset — the name comes from `documents`. An XML document (a 13F's holdings) comes back as its record tables rendered to text. Returns the next offset when more follows.
| Name | Type | Req | Description |
|---|---|---|---|
| document | string | yes | The document's name, from `documents`. |
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| length | integer | – | Characters to read (max 8000). |
| offset | integer | – | Character offset. |
No output schema declared.
No examples provided.
read_text ~129
Read up to 8000 characters of the filing's primary document as plain text — every Item, note, table, the cover and the signatures, tagged or not, from a character offset (from search_text, or a section offset in describe_filing). Returns the text and the next offset when more follows.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| length | integer | – | Characters to read (max 8000). |
| offset | integer | – | Character offset to start at. |
No output schema declared.
No examples provided.
records ~155
The record tables of an XML filing — a Form 4's transactions and holdings, a 13F's positions — as rows, with the document's header fields beside them. A 13F's positions are filed in its INFORMATION TABLE, not the cover page; records fetches that document and returns its `infoTable` first. Omit `table` for every table; describe_filing's `sections.records` lists the primary document's tables.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| limit | integer | – | Rows per table. |
| table | – | – | One table's name; omit for all of them. |
No output schema declared.
No examples provided.
resolve_element ~160
Find the XBRL concepts a filing reports for a phrase: 'revenue', 'lease liability', 'us-gaap:Assets'. Matches qnames, names and labels, ranked; each match carries its label, type, period type, balance, how many facts the filing reports for it, and the networks it appears in. Use the returned qnames in fact_grid and calculation.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| limit | integer | – | Matches to return (max 100). |
| phrase | – | – | The same as query; either name is taken. |
| query | – | – | A concept name, qname, or label phrase. |
No output schema declared.
No examples provided.
search_filings ~334
Find filings ACROSS EDGAR by phrase, form, date and filer — the discovery step before load_filing. `text_query` searches the filing documents themselves (quote a phrase for an exact match: '"material weakness in internal control"'), `forms` narrows by type (['10-K', '8-K']), `start_date` / `end_date` are YYYY-MM-DD, `ciks` restricts to given filers. Returns a page of hits and the total that matched; each hit's `source` is the `cik:accession` load_filing takes, so a result reads directly. EDGAR's full-text index begins in 2001 — an earlier date range finds nothing, however the filing reads. This is the corpus-wide search; `search_text` searches inside one loaded filing.
| Name | Type | Req | Description |
|---|---|---|---|
| ciks | – | – | Restrict to filings EDGAR associates with these CIKs — which for an ownership form (3, 4, 5) means the issuer as well as the reporting owner, so a company's CIK also returns the insider forms filed a… |
| end_date | – | – | Latest filing date, YYYY-MM-DD. |
| forms | – | – | Form types, e.g. ['10-K', '20-F']. |
| limit | integer | – | Hits to return; the total matched is reported regardless. |
| start_date | – | – | Earliest filing date, YYYY-MM-DD (EDGAR indexes from 2001). |
| text_query | – | – | A phrase to find in the filing text; quote it for an exact match. |
No output schema declared.
No examples provided.
search_text ~232
Case-insensitive regular-expression search over the filing's primary document as plain text — every Item, note, table, the cover and the signatures, tagged or not. Returns up to max_hits matches (max 25), each with its character offset and a window of text centred on it (default 300 characters, max 1500), plus the total match count. Follow up with read_text at an offset. When more matched than came back, `sections` counts where the matches fall, the busiest sections first — search or read there rather than narrowing blind — and `sections_omitted` how many further sections hold the rest. When nothing matched, `terms` counts the pattern's own words separately, so a phrase the filer words differently says which word to try.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| max_hits | integer | – | Matches to return. |
| pattern | string | yes | A regular expression (Python syntax). |
| window | integer | – | Characters of context around each match. |
No output schema declared.
No examples provided.
statement ~232
Render one presentation network as a table: rows in the filing's own order with depth, label and the consolidated value per period column. `statement` is a network id, name (or part of one) or role URI from describe_filing; under the product profile a kind also works (balance_sheet, income_statement, cash_flow_statement, equity_statement, or a phrase like 'balance sheet'). `periods` limits the columns to those keys, end dates, or years; otherwise the most recent eight. A `truncated` response continues from its `next_offset` as `offset`.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| max_rows | integer | – | Rows to return (max 400). |
| offset | integer | – | Rows to skip: the `next_offset` a truncated response returned. |
| periods | – | – | Period keys, end dates or years to keep as columns. |
| statement | string | yes | Network id, name (or part of one), role URI, or a kind. |
No output schema declared.
No examples provided.
unload_filing ~47
Drop a loaded filing from the server (its id from list_filings). Load another with load_filing; nothing else changes.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | string | yes | The loaded filing's id. |
No output schema declared.
No examples provided.
view_filing ~127
Open a loaded filing in the browser-based report viewer and return the link. The document is serialized and served from an unguessable path on this machine's loopback interface, readable only by the viewer's own origin, for as long as this server runs. Give the returned `viewer_url` to the user; it renders the statements, notes and cover as a report.
| Name | Type | Req | Description |
|---|---|---|---|
| filing | – | – | The loaded filing's id (from list_filings; a ticker or accession also works). Omit when only one filing is loaded. |
| format | string | – | The serialization the viewer reads. |
No output schema declared.
No examples provided.
How do I install the xbrlkit MCP server?
xbrlkit runs locally as a PyPI package, launched with uvx --with 'xbrlkit[mcp]' xbrlkit serve --transport stdio. Ready-made configuration for Claude, Cursor, VS Code, Codex and 3 more is on this page, copied from each client's own documentation.
pypi · xbrlkit
claude mcp add ai-robosystems-xbrlkit -- uvx --with 'xbrlkit[mcp]' xbrlkit serve --transport stdio
What is the xbrlkit MCP server?
xbrlkit is an MCP server listed in the public MCP registry as ai.robosystems/xbrlkit. Read SEC EDGAR and XBRL filings with an AI: 18 tools over the filing itself. Local, free, no account. This page covers its PyPI package (xbrlkit).
Is the xbrlkit MCP server safe to use?
xbrlkit scores 67 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 21 November 2026. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the xbrlkit MCP server expose?
xbrlkit exposes 18 tools: list_filings, search_filings, load_filing, unload_filing, describe_filing, and 13 more. Their descriptions and schemas cost roughly 3,652 tokens of context every time the server is loaded.
Is the xbrlkit MCP server still maintained?
xbrlkit is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.
What licence is the xbrlkit MCP server under?
xbrlkit declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.
What did VerifyMCP test on the xbrlkit MCP server?
VerifyMCP ran 15 checks on xbrlkit, covering Known CVEs, Install scripts, Dependency health, Source repository and 11 more. One item could not be completed, a gap on our side rather than a finding about xbrlkit: Malware scan. Endpoint Security and Transport & Reachability do not apply to a PyPI package, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.