Scrutica MCP server
io.github.davidgringras/scruticadavidgringrasstreamable-httpv0.2.0Scanned
AI compute infrastructure intelligence: facilities, supply chains, sovereign AI, export controls.
10 tools: 1 change things and 9 only read.
- Reads private data · 0
- Reads text others write · 0
- Can send data out · 0
- 2 findings
Tools listed without sign-in. Remote posture
With Anlyon
1 tool here changes things. Govern it.
- Anlyon lists the tools and adopts none.
- Each tool you adopt gets approvals and receipts.
- Your agent never holds the key.
A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.
What the Scrutica MCP server can do
The Scrutica MCP server is published by the GitHub account davidgringras. It has 10 tools. It can create or change data (1 tool: scrutica_estimate_flops), but cannot delete data. 9 tools only read.
None of its tools reads private data, brings in text other people write, or sends data out.
The fixed checks matched 2 findings (long-description). Each one comes with the text it matched.
Its remote endpoint lists its tools without sign-in, and its tools changed 2 times since Anlyon Radar first saw them.
Last scanned Nov 20, 2026, version 0.2.0.
Tools
Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.
10 tools
Changes things 1Claude Code: ask
scrutica_estimate_flopsEstimate hypothetical peak and sustained throughput for a user-specified hardware configuration and precision.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Estimate hypothetical peak and sustained throughput for a user-specified hardware configuration and precision. Sustained throughput equals dense peak throughput multiplied by accelerator count and utilization. The reported bounds are fixed at 0.7 and 1.15 times that result; they are model assumptions. Retain the bounds and is_estimated flag when quoting the estimate. This tool does not verify a named facility's inventory, availability or regulatory compliance. Named-facility estimates have separate evidence requirements documented at /methodology#flop-estimation.
Annotations the server declares
- title: FLOP estimator
- readOnlyHint: true
Only reads 9Show 9 read-only tools
scrutica_entity_list_changesWhat CHANGED in the BIS Entity List, as bounded change-log summaries.1 finding
Why it is in this group
- The name "scrutica_entity_list_changes" says "list": a word for reading.
Findings
- Info
The description is 2,238 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ignation_date, falling back to modal effective_date) and labeled with its source."
Description, as the server wrote it What CHANGED in the BIS Entity List, as bounded change-log summaries. Backed by the same derivation the public /export-controls/changes page and its RSS feed render. PREFER this over scrutica_query_export_controls for freshness questions ("what changed this month", "latest additions", "recent removals"): designation rows arrive already grouped into per-Federal-Register-notice change events, so you do not re-derive notice grouping from citation strings. Each event includes: canonical FR citation and link, notice title / publication date / document number where the citation resolved to one FR document, a derived event date labeled with the source column it came from, entry / entity / address / removal counts (entry_count counts entries as the Register does: an entity under two destinations is two entries), per-country counts, a bounded sample of entity names, and the entities cross-referenced to compute-infrastructure organizations. Removal actions are returned separately as date-keyed events — a screening list maintained from additions alone over-blocks forever. Also returns a last-N-ISO-week activity rollup counting BOTH additions and removals. Zero-count weeks are a real quiet-Federal-Register signal, not missing data. Results are bounded summaries, never row dumps: at most 24 events per call (default 6, newest first), with any truncation declared in metadata. Cite totals from the count field and per-event entry_count, never by summing a sampled list. Do NOT use this to look up one entity's designation rows — use scrutica_query_export_controls (substring match on the published name). Use company_id here only to scope the log to notices touching an organization you already resolved. Authority tier: Federal-Register-anchored designations are Tier 1 (primary source). Each entry is dated by the effective date of the Federal Register notice that added it, the notice the Entity List's own eCFR citations for the entry resolve to (a removed record: the notice its entry was dated by while the Register printed it, else its own citation's notice), except where no citation resolves to a notice; the event date is derived (modal designation_date, falling back to modal effective_date) and labeled with its source.
Annotations the server declares
- title: Entity List change log
- readOnlyHint: true
scrutica_get_companyFetch complete details for a single organization (company, government entity, holding company) by canonical Scrutica ID.No findings
Why it is in this group
- The name "scrutica_get_company" says "get": a word for reading.
Description, as the server wrote it Fetch complete details for a single organization (company, government entity, holding company) by canonical Scrutica ID. Returns: legal name, country of HQ, organization type, parent / subsidiary references, supply-chain edge counts. US government listings that are not export controls: us_restrictive_listings holds the rows that name the organization, each with its list (dod_1260h: the Defense Department's Section 1260H list of "Chinese military companies"; ofac_ns_cmic: OFAC's Non-SDN Chinese Military-Industrial Complex Companies List; dhs_fascsa_exclusion: a Homeland Security FASCSA exclusion order), whether it is on the current list, the name as printed, its source (Federal Register notice or OFAC list file) and how it was matched (resolution 'name' or 'alias'; 'ambiguous' when the printed name matches several organizations, which are listed). on_1260h and on_ns_cmic are true when a current row resolves to the organization, false when no current row names it, and null when a current row names it among several organizations (unless the stored flag already reads true) or the rows cannot be read: names that match no record are not resolved, so false is not proof of absence. on_ofac_sdn is always null: OFAC's SDN list is not ingested. Use scrutica_query_export_controls for BIS designation details. Use scrutica_get_supply_chain for full edge graphs.
Annotations the server declares
- title: Company record
- readOnlyHint: true
scrutica_get_facilityFetch complete details for a single Scrutica facility by canonical ID.1 finding
Why it is in this group
- The name "scrutica_get_facility" says "get": a word for reading.
Findings
- Info
The description is 2,395 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ute ownership-chain BIS exposure. Resolve facility IDs first via scrutica_search."
Description, as the server wrote it Fetch complete details for a single Scrutica facility by canonical ID. Returns: roles[] — each organisation the record names, under the relationship its own source ESTABLISHES (site owner, site operator, AI hardware owner, or another relationship such as PeeringDB listing holder, lessee, developer), with the reason, source URL, quote and authority tier; withheld_roles[] — recorded attributions the cited source does not establish, with the reason; then country, place (the city, town or census-designated place containing the site, else its county; then state and country) with place_basis saying how it was derived, the source's own location cells verbatim (source_location_cells), power capacity (MW), GPU inventory (where disclosed), location (lat/lng), facility type, status, data_source, source_url, is_estimated flags, and any logged data-quality flags (cross-source divergence investigations). present_inventory is Scrutica's partition-by-partition compute evidence (the verdict the facility page, threshold table and query tools use): each accelerator partition admitted at its cited 16-bit rate (rate_format names FP16 or BF16, as the vendor prints it; rate_sparsity says dense, or unstated where the vendor labels the rate neither way) or set aside with a reason; one such partition withholds the facility figure. Read facility_peak_pflops_16bit, dense_16bit_tflops_per_accelerator and peak_pflops_16bit. The deprecated facility_peak_pflops_fp16, dense_fp16_tflops_per_accelerator and peak_pflops_fp16 keep their earlier meaning, a figure at an FP16 rate: they are null for a BF16 partition, and the facility figure is null unless every partition is admitted at FP16. facility.peak_pflops_fp16 and h100_equivalent_gpus are the record's catalogue scalars, a different basis. Describe an organisation as the site's owner ONLY where roles[] names it "Site owner"; a recorded organisation is not by itself evidence of title or operation. Where a role's evidence covers only part of the record, that role's scope_extent is "part", scope names the part, and the role reads "…, for <the part>"; state the role of that part only, never of the whole site. For BIS Entity List designations tied to an organisation recorded at the facility, look it up via scrutica_query_export_controls — this tool does not compute ownership-chain BIS exposure. Resolve facility IDs first via scrutica_search.
Annotations the server declares
- title: Facility record
- readOnlyHint: true
scrutica_get_methodologyReturn methodology documentation for a Scrutica metric or index.No findings
Why it is in this group
- The name "scrutica_get_methodology" says "get": a word for reading.
Description, as the server wrote it Return methodology documentation for a Scrutica metric or index. Topics: 'flop-estimation', 'cost-index', 'compute-visibility-index', 'supply-chain-weighting', 'chokepoint-cascade', 'sovereign-execution-classification'. Returns the canonical URL + section anchor + summary. Use this when a user asks "how did you calculate X".
Annotations the server declares
- title: Methodology reference
- readOnlyHint: true
scrutica_get_scenarioFetch geopolitical compute risk scenarios.No findings
Why it is in this group
- The name "scrutica_get_scenario" says "get": a word for reading.
Description, as the server wrote it Fetch geopolitical compute risk scenarios. Available: 'taiwan-strait' (4 TSMC disruption scenarios), 'iran-threat' (IRGC missile range vs Gulf compute), 'tokyo-earthquake' (Japan memory-fab exposure), 'abqaiq-2' (Saudi grid). Returns the scenario's summary plus its substrate serialized from the interactive page's own data modules — assumption scenarios / threat systems / affected assets with per-item sources and authority tiers, capped lists shipping their true totals. The propagation SIMULATION itself remains interactive-only at the returned URL.
Annotations the server declares
- title: Risk scenario
- readOnlyHint: true
scrutica_get_sovereign_programFetch detailed data on a national sovereign AI compute program.No findings
Why it is in this group
- The name "scrutica_get_sovereign_program" says "get": a word for reading.
Description, as the server wrote it Fetch detailed data on a national sovereign AI compute program. Returns: announced_usd, announced_govt_only_usd, committed_usd, disbursed_usd, reality_ratio, status, key_partners, NVIDIA/US dependency, source_count, and published_spending (a government's own spending figures, each in its currency against its budget, with the document). 'list_all' returns a summary table of all tracked programs. announced_usd can include private commitments; announced_govt_only_usd records the government-funded portion. Compare the programme sources for scope, dates and currency conversions. A government-only figure can still span a different budget period or initiative from another programme's figure. reality_ratio divides disbursed_usd by announced_govt_only_usd, using announced_usd when the government-only field is null. The ratio is null when disbursement or the denominator is missing, or the denominator is zero. Each row includes reality_ratio_denominator_usd and reality_ratio_denominator_field so the division is reproducible from the payload. Where disbursed_usd converts a government's published spending figure, reality_ratio_basis is native_currency and the ratio divides that figure by its own budget in the source currency (reality_ratio_numerator_native / reality_ratio_denominator_native, in reality_ratio_currency): the two dollar amounts were converted on different dates and do not reproduce it. disbursed_usd derives only from a government's published expenditure or payments; a transfer to an administering body is listed in published_spending as released and is not counted. Missing fields and differing reporting scopes limit comparisons; reality_ratio_unavailable_reason explains an absent ratio.
Annotations the server declares
- title: Sovereign AI program
- readOnlyHint: true
scrutica_get_supply_chainReturn supply-chain relationships for one or more organizations.No findings
Why it is in this group
- The name "scrutica_get_supply_chain" says "get": a word for reading.
Description, as the server wrote it Return supply-chain relationships for one or more organizations. direction = 'upstream' traces suppliers (who feeds this entity); 'downstream' traces customers (who depends on this entity); 'both' returns both. Each edge carries supplier_org_id, customer_org_id and data_source; product_service, category, sole-source status, share and criticality appear only where the edge's cited source states them for that pair, and attribute_withholding gives the reason for every withheld value. Collection size: 20,763 supply-chain records across all sources (snapshot 2026-11-16). Of these records, 1,924 come from a licensed layer limited to relationships between organizations already in Scrutica's records.
Annotations the server declares
- title: Supply-chain edges
- readOnlyHint: true
scrutica_query_export_controlsLook up BIS Entity List designations for companies or countries.No findings
Why it is in this group
- The name "scrutica_query_export_controls" says "query": a word for reading.
Description, as the server wrote it Look up BIS Entity List designations for companies or countries. The queried dataset is the US Commerce Department Entity List, Federal-Register-anchored; it does NOT cover OFAC SDN or Wassenaar CCL (for an organization's Section 1260H, NS-CMIC or FASCSA listing, use scrutica_get_company — its us_restrictive_listings rows and the on_1260h / on_ns_cmic flags derived from them; OFAC's SDN list is not ingested, and on_ofac_sdn is null). entity_name is a case-insensitive substring match against the designated entity's published name (no alias expansion). Returns entity_name, entity_type, country, list_type, designation_date, removal_date, grounds, federal_register_citation, source_url (Federal Register notice URL), data_quality_flags. Provide entity_name or country. entity_id filtering is not implemented: an entity_id-only call is rejected; alongside a supported filter, entity_id is ignored and the response sets entity_id_ignored: true. VEU authorization removals are returned separately in regulatory_events (maximum 50, with count and truncation); they do not contribute to designation membership or counts. Event effective_date and publication_date are distinct. Authority tier: Federal-Register-anchored designations are Tier 1 (primary source). bis_coverage_class discriminates how a row is covered: explicitly_named (FR notice cites the entity by name) vs affiliate_50pct_rule (inferred via BIS's 50% Affiliates Rule). coverage_note is populated only on affiliate_50pct_rule rows and states whether that rule is currently suspended. data_quality_flags is an array of provenance/caveat notes, empty for most rows; where present (derived-country provenance, removal-date sourcing, or the affiliate rows' suspension and substrate-reproducibility caveats) treat it as load-bearing context for the row.
Annotations the server declares
- title: Export-control designations
- readOnlyHint: true
scrutica_searchSearch facilities, companies, or sovereign programs by free-text query.No findings
Why it is in this group
- The name "scrutica_search" says "search": a word for reading.
Description, as the server wrote it Search facilities, companies, or sovereign programs by free-text query. Returns ranked results with id, name, type, one-line summary, and Scrutica URL. Filter by entity_type to scope to a single class. Filter by ISO 3166-1 alpha-2 country code. Do NOT use this for BIS Entity List / export-control lookups — use scrutica_query_export_controls. Do NOT use this for supply-chain traversal — resolve an entity ID first via this tool, then call scrutica_get_supply_chain.
Annotations the server declares
- title: Search Scrutica
- readOnlyHint: true
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Exfil surface
Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.
No private-data, untrusted-input, outbound path in this server alone: no tool is tagged private-data or untrusted-input or outbound.
- Reads private data0
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write0
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out0
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.
Remote posture
What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.
Looked at Nov 20, 2026, 19:03 UTC, unchanged since Oct 22, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 200
- tools/list served without credentials
- yes
- Tools that change things, listed without credentials
- 1
- Protocol version
- 2025-06-18
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- not observed
- Authorization servers
- not observed
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- not observed
- Client ID metadata documents
- not observed
- WarnThe whole server
The endpoint served tools/list without credentials, and the list has 1 tool that changes things (1 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://scrutica.com/api/mcp: tools/list returned 10 tools to an anonymous client, including scrutica_estimate_flops (write)
Findings
Matches of fixed checks, with the text they matched.
The description is 2,395 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ute ownership-chain BIS exposure. Resolve facility IDs first via scrutica_search."
The description is 2,238 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ignation_date, falling back to modal effective_date) and labeled with its source."
History
Changes to its tool definitions and to its remote endpoint, newest first.
Tools2 changed.~2 changedView diff
- Changed:scrutica_get_facilityOnly readsdescription
- Changed:scrutica_get_supply_chainOnly readsdescription
Tools7 changed.~7 changedView diff
- Changed:scrutica_entity_list_changesOnly readsdescription
- Changed:scrutica_get_companyOnly readsdescription
- Changed:scrutica_get_facilityOnly readsdescription
- Changed:scrutica_get_scenarioOnly readsdescriptioninputSchema
- Changed:scrutica_get_sovereign_programOnly readsdescription
- Changed:scrutica_get_supply_chainOnly readsdescription
- Changed:scrutica_query_export_controlsOnly readsdescription
Install
Pinned to version 0.2.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http scrutica https://scrutica.com/api/mcpclaude mcp add --transport http scrutica https://scrutica.com/api/mcpRun in a terminal.
Claude Code permissions
Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [
"mcp__scrutica__scrutica_search",
"mcp__scrutica__scrutica_get_facility",
"mcp__scrutica__scrutica_get_company",
"mcp__scrutica__scrutica_query_export_controls",
"mcp__scrutica__scrutica_entity_list_changes",
"mcp__scrutica__scrutica_get_supply_chain",
"mcp__scrutica__scrutica_get_sovereign_program",
"mcp__scrutica__scrutica_get_scenario",
"mcp__scrutica__scrutica_get_methodology"
],
"ask": [
"mcp__scrutica__scrutica_estimate_flops"
],
"deny": []
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as scrutica. Review it before you use it.
Pin
Keep the decision: fail CI when the tools change, follow every change, or show a badge.
Pin in CI
Fail a build when the tools you reviewed are not the tools the server serves today.
curl -s "https://radar.anlyon.com/api/v1/servers/io.github.davidgringras/scrutica/verify?hash=c69fb6207a28c3c3408808ea8395244836c9e0a6c56245d9d73fa4a2e0d53c64" | jq -e '.changed == false'Pinned hash c69fb6207a28, from Nov 20, 2026, 20:17 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.davidgringras/scrutica/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.davidgringras/scrutica)Source: MCP Registry. Synced Oct 21, 2026, 18:31 UTC.
Scrutica MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the Scrutica MCP server safe?
- No scan can promise that. What its definitions show: 10 tools, 1 of which can change, send or delete something, and 2 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the Scrutica MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the Scrutica MCP server do things that cannot be undone?
- No tool is classed as irreversible: none of them sends, pays, publishes or merges, going by their definitions.
What tools does the Scrutica MCP server have?
- 10 tools: 1 write, 9 read. scrutica_estimate_flops, scrutica_search, scrutica_get_facility, scrutica_get_company, scrutica_query_export_controls, scrutica_entity_list_changes, scrutica_get_supply_chain, scrutica_get_sovereign_program and 2 more.
Does the Scrutica MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the Scrutica MCP server send data to outside services?
- None of its tools sends, posts or publishes, or takes a recipient, channel or URL.
Has the Scrutica MCP server changed its tools?
- Yes, 2 times since Anlyon Radar started tracking it, most recently on Nov 20, 2026. Every change is on this page and in its Atom feed.