Skip to main content
Glama
Sirenic MCP server

Sirenic MCP server

Official

Compare French companies

compare_french_companies
Read-only

Assess 2–5 French companies side by side using identity, risk, filed accounts, legal alerts, and sanctions data, with rankings and explicit comparability warnings; no overall winner is declared.

Instructions

Use when asked to compare, rank or choose between 2 to 5 French companies (customers, prospects, suppliers, competitors). Returns a cross-table (identity, deterministic default-risk score, latest filed accounts with their accounting scope, BODACC legal alerts, sanctions screening of the legal name), per-axis rankings and, importantly, an explicit comparabilite block stating when the companies are NOT comparable (different sectors, different sizes, a holding in the batch). NEVER returns an overall winner: a holding with no debt outranks a large industrial group on default risk, which would be misread as a verdict on quality. Since 2026-09-08 nine rankings (risk, revenue, net result, EBITDA margin, lowest debt ratio, liquidity, revenue growth, cash vs short-term debt, seniority) each carry a status in eligibilite_classements (statut autorise/interdit with raison, type evaluatif/descriptif, donnee, perimetre, exercices, couverture): the evaluative risque_le_plus_faible is served ONLY when the batch is comparable for a score and the ranked scores rest on exactly the same axes (axes_communs); amounts are forbidden across different fiscal years, ratios also across different sectors or sizes; a forbidden ranking is absent from classements, never rank companies from solidite.score_risque or the ratios yourself when it is interdit. solidite.classement_risque repeats per company whether it is eligible, ranked, and why not (score_non_concluant, axes_differents); the score stays served with its couverture_axes. Billed per company at $0.12 via x402, the amount is the unit price times the number of SIREN. A batch holding such a SIREN (sole proprietorship, restricted diffusion, group without legal personality) is refused whole: 400, never charged, sirens_refuses gives each refused SIREN and its code; replay without them. When you report a result, cite its citation verbatim.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "sirens": "<sirens>"
    },
    "name": "compare_french_companies"
  }
}

Input Schema

NameRequiredDescriptionDefault
sirensYesList of 2 to 5 nine-digit SIRENs

Output Schema

NameRequiredDescriptionDefault
hintNoHow to settle the quote, present when payment_required is true.
quoteNoThe signable x402 payment requirements when payment_required is true: {x402Version, accepts[]} where each entry carries scheme, network, amount, asset and payTo (USDC and EURC options at the same numeric amount). Sign one entry and call again with `x_payment`.
resultatNoThe endpoint's JSON response when payment_required is false. Paid responses carry `source`, `disclaimer` and an Ed25519 signature, and every paid JSON response carries a `provenance` array (all tools since 2026-09-06): one entry per block served, with the official register, licence, version, `empreinte_sha256` (full snapshots), `as_of` date and `precision_as_of` (what that date means), plus the common per-block envelope: `etat` in a closed list (servi, absence_mesuree, absence_non_conclusive, partiel, perime, indisponible, sans_objet), `motif` when unavailable, `age_jours` (only for publication or ingestion dates), `couverture` (complete / partielle / non_mesurable with its measure) and `confiance` in bands (faible / moyenne / forte, only where a measure grounds it, never averaged across blocks). Read the states before the numbers: an absence_non_conclusive, indisponible or partiel block is never « nothing to report »; only absence_mesuree asserts an absence. Register codes: GET /v1/provenance/registres ; states, closed lists and reading rules: GET /v1/lecture (both free). The PDF tools carry no JSON envelope (the PDF report prints it).
free_tierNoFree API key offer for this tool, present with a quote.
payment_requiredYesTrue when this response is an x402 payment quote instead of data: settle one of the quote's `accepts` options and call the tool again with `x_payment`.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

A4.5/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only supply readOnlyHint and openWorldHint, so the description carries most of the behavioral load and does so well: it discloses the refusal path (400, never charged, sirens_refuses per SIREN, replay without them), the billing model ($0.12 per company via x402, unit price times SIREN count), and the deliberate design constraint that no overall winner is returned and why. It also states an output obligation ("cite its `citation` verbatim"), which is behavior an agent cannot get from annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

It is front-loaded with the when-to-use clause, which is good, but the body is a dense block of long semicolon-chained sentences mixing eligibility rules, error handling, pricing and citation policy. There is some redundancy, e.g. the `eligibilite_classements` explanation is partly restated in the `solidite.classement_risque` sentence, and much of the structural detail would be better served by the existing output schema.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a tool with billing, refusal paths, and non-trivial eligibility logic, the description covers cost, failure modes, what is and is not returned, and how to report results. Even with an output schema present, the added behavioral context is what an agent needs to call and interpret this tool correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The single `sirens` parameter is fully described in the schema (2-5 nine-digit SIRENs), so the baseline would be 3. The description adds genuine input semantics beyond the schema by enumerating which SIREN categories are unacceptable (sole proprietorship, restricted diffusion, group without legal personality) and what happens when such a SIREN appears (whole batch refused, sirens_refuses lists each with a code).

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a specific verb set ("compare, rank or choose") and a precise resource and scope ("2 to 5 French companies (customers, prospects, suppliers, competitors)"), then names the concrete artifacts produced: cross-table, per-axis rankings, and a comparabilite block. This is clearly distinguishable from single-company siblings such as get_french_company_default_risk or get_french_company_file.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It gives an explicit trigger condition ("Use when asked to compare, rank or choose between 2 to 5 French companies") and an explicit prohibition on inferring a winner, plus rules about when a ranking must not be used ("never rank companies from solidite.score_risque or the ratios yourself when it is `interdit`"). It does not name a specific alternative tool for the >5 or single-company case, leaving that routing to inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools