Decionis Commerce Gate MCP
Commerce preflights, agent run mandates, D365 authorization, and offer checks; no external writes.
Connector by decionis.comOfficial · Domain verified by the MCP Registry
Install
Works with Claude Code, Codex, VS Code, Cursor, Windsurf, Claude Desktop. Read the repository before you give any tool access to your data.
npx agentmarket install connector/decionis-com-commerce-gate
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this server can do and what it asks for, from its tool list and install config. Checked Oct 25, 2026.
- Tools
- 14 read · 4 write · 0 delete
- Secrets
- 1 (API keys, tokens or passwords)
- Install scripts
- None found
- Runs as
- Node.js package
Tools by what they do 18
Read from the source code; the running server may list different tools.
Read 14 · Look things up without changing anything.
commercegate_agent_forecast_mandatecommercegate_agent_get_budgetcommercegate_agent_get_runcommercegate_describe_capabilitiescommercegate_evaluate_actioncommercegate_evaluate_marketplace_offer_submissioncommercegate_get_dossiercommercegate_get_historical_assessmentcommercegate_get_policy_readinesscommercegate_get_proof_packetcommercegate_list_history_sourcescommercegate_list_shadow_reports
Show 2 more tools
commercegate_summarize_shadow_reportscommercegate_validate_erp_transaction
Write 4 · Create, change, send or run things.
commercegate_agent_record_proposalcommercegate_agent_start_runcommercegate_agent_stop_runcommercegate_start_historical_assessment
Secrets and settings it asks for
Secrets: values you paste in that let it act as you.
DECIONIS_API_KEY
Other settings (environment variables or headers).
AGENTOPS_AUTO_PROVISIONAGENTOPS_HOMEDECIONIS_API_BASEDECIONIS_ORG_ID
Files it touches
- Any path the agent passes to its tools
Tools (18)
Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].
commercegate_agent_forecast_mandateForecast a CommerceAgent MandateBefore proposing a mandate, replay the organization's own recorded decisions for the previous three UTC months against it: how many proposals of the mandate's action families the merchant's tools made on its platforms each month, and the declared face value of those that proceeded, compared with the mandate's action and exposure budgets. This describes what was proposed, never predicts what will be; proposals whose kind the ledger did not capture are reported separately, and no history is reported as insufficient history rather than as a fit. It records nothing and executes nothing.
commercegate_agent_get_budgetGet a CommerceAgent BudgetRead an agent's budget for the current UTC month: the envelope an operator set (actions and currency exposure), what the agent's closed runs used, what its active runs still reserve, and what remains. A run whose mandate exceeds the remaining room cannot start. The envelope is set outside the agent; this tool can only read it. With no envelope the view carries usage only.
commercegate_agent_get_runGet a CommerceAgent RunRead the current state, budgets, horizon and recent proposals of one run. A run past its horizon reads as expired. This is a local read of the run ledger; it does not fetch dossiers or evaluate anything.
commercegate_agent_record_proposalRecord a Proposal Under a RunAfter commercegate_evaluate_action or commercegate_validate_erp_transaction has answered, record that exact proposal under the run: its action family, platform, the outcome Commerce Gate returned, the dossier reference and the amount. Every recorded proposal consumes one action; a PROCEED also consumes its amount from the exposure budget. A proposal outside the objective, outside the platform scope, in another currency, after the horizon or beyond a budget is refused and the run closes where a budget is exhausted. A refusal is a stop condition. Recording never executes the proposal.
commercegate_agent_start_runStart a CommerceAgent RunUse this before proposing commerce actions as Decionis CommerceAgent. It opens one bounded run for the configured organization: an objective that fixes which action families may be proposed, the platforms and optional SKUs or orders in scope, an action budget, a currency-exposure budget and a horizon of at most 31 days. The same idempotency key replays the same run. Starting a run evaluates nothing, calls no network and authorizes no execution; it is the mandate every later proposal is checked against.
commercegate_agent_stop_runStop a CommerceAgent RunThe kill switch. Stops the run so no further proposal can be recorded under it, with a stated reason. Stopping is idempotent and cannot be undone; start a new run only with fresh operator intent. It does not reverse any recorded proposal or any action the merchant's tools already performed.
commercegate_describe_capabilitiesDescribe Commerce Gate CapabilitiesUse this first, or when someone asks 'can you check prices, Walmart orders, refunds, or a marketplace offer against our policy?'. It lists the eleven commerce-action preflights (seven published families plus the four finance families of the unreleased commerce-preflight-v2 source candidate), the Walmart connection and order-path mappings, the separate Dynamics 365 guard, and the Microsoft Partner Center and AWS Marketplace SaaS offer-submission preflight. It also explains PROCEED, HOLD, and BLOCK, tenant connection state, and available evidence tools. This local diagnostic never calls the Deci…
commercegate_evaluate_actionEvaluate a Commerce ActionUse this as the default preflight before a price, inventory, order-acceptance, fulfillment, promotion, refund, return, payment-release, vendor-bank-change, early-payment-discount, or credit-hold-release proposal, for questions like 'would repricing this SKU to $89 on Walmart still clear our margin floor after the referral fee?', 'is it safe to accept this 40-unit order with the stock we have left?', or 'can the support bot refund $2,850 on this order?'. It sends the bounded tenant-scoped action and policy signals to Decionis in default SHADOW mode and may persist a signed Decision Dossier. Sha…
commercegate_evaluate_marketplace_offer_submissionPreflight a Marketplace SaaS Offer SubmissionUse this before manually submitting a Commerce Gate SaaS offer in Microsoft Partner Center or AWS Marketplace. It requires a bounded offer, plan and market packet, public HTTPS listing and technical URLs, platform identity, preview or test confirmation, and digest-bound release evidence. Microsoft packets require Microsoft Entra tenant and application IDs; AWS packets require a Login with Amazon subject bound to the seller AWS account and IAM role. This is a non-writing preflight: it evaluates release metadata and preview or test evidence without reading customer commerce records, calling eith…
commercegate_get_dossierGet a Commerce Gate Decision DossierUse this to answer 'why was that order held?' or 'what policy version decided that price change?' by reading the signed Decision Dossier for a known dossier UUID within the environment-bound organization. This is a tenant read; it never executes or modifies a commerce action and it fails closed when credentials are absent or invalid.
commercegate_get_historical_assessmentRead a Historical Assessment and EvidenceRead an immutable assessment by UUID within the buyer's credential-bound workspace. Evidence remains bound to the historical source and 90-day window. PROCEED is a retrospective classification, never authorization to execute. Missing proof references do not imply signed evidence.
commercegate_get_policy_readinessRead Commerce Gate Policy Readinessread-onlyRead the exact configured policy version and its sixteen configuration states before reviewing a proposal. This current-state read is not an action verdict, does not evaluate an action, issue evidence, reserve budget, or activate enforcement. Missing, disabled or unsupported checks are not passes. Stop when the selected policy is unavailable or the report is malformed.
commercegate_get_proof_packetGet a Commerce Gate Proof PacketUse this when finance or an auditor asks for signed evidence by fetching the proof packet for a known dossier UUID in the environment-bound organization. The tool is read-only, performs no marketplace action, and never treats successful verification as permission to execute.
commercegate_list_history_sourcesList Historical Data SourcesList the synthetic dataset and existing connected stores authorized for the buyer's Commerce Gate workspace. Real data requires a verified store connection established outside this tool. No raw transactions or connection credentials are returned.
commercegate_list_shadow_reportsList Commerce Gate Shadow ReportsUse this to answer 'what would Shadow Mode have held recently?' by listing recent SHADOW evaluations in the environment-bound organization across price, inventory, order, fulfillment, promotion, refund, and return proposals. Results describe what policy would have done; they do not report marketplace execution and must not be interpreted as authority to mutate commerce state.
Show 3 more tools
commercegate_start_historical_assessmentAssess the Previous 90 DaysStart and persist an idempotent historical assessment of the previous 90 days using synthetic history or an authorized connected store. The service fixes dates, policy and tenant; callers cannot upload transactions, set policy, or evaluate future actions. A verified AWS grant is required, including during the 14-day synthetic trial. Results include provenance, missing costs, summaries and bounded evidence; a null proof reference is not signed proof.
commercegate_summarize_shadow_reportsSummarize Commerce Gate Shadow ReportsUse this for 'how much would we have caught this month?' style questions: aggregate SHADOW outcomes, near misses, and policy mismatches for the environment-bound organization, so an operator can judge whether to turn enforcement on. It is a read-only operational summary; it neither changes enforcement mode nor executes any commerce action.
commercegate_validate_erp_transactionValidate an ERP TransactionUse this for a fully formed Dynamics 365 Business Central or Finance and Supply Chain Management transaction that needs central enforced authorization. Commerce Gate evaluates line economics first and, only after policy clears, atomically authorizes the agent budget before returning the binary ALLOW or BLOCK decision used by the ERP enforcement shell. The tool can reserve budget and returns a reason-coded validation response; unlike the Shadow evaluator, it does not promise a retrievable dossier. It never writes, posts, releases, or modifies the ERP transaction itself. ALLOW is authorization e…
Signals
- Weekly downloads
- 157
- Last release
- Oct 15, 2026 (13 days ago)
- Last package publish
- Oct 15, 2026 (13 days ago)
From the package registry.
Details
- Registry name
com.decionis/commerce-gate- Version
- 0.2.0
- Auth
- API key
- Packages
- npm @decionis/commerce
- License
- License unknown
- Website
- commerce.decionis.com
- First seen
- Oct 24, 2026
- Last checked
- Oct 25, 2026
- Source
- Official MCP Registry
More from decionis.com
Cite this page
Decionis Commerce Gate MCP. Agent Mart. https://agentmart.dev/connectors/decionis-com-commerce-gate (last updated 2026-10-25)