servers / conduit-agentic-commerce
Conduit Agentic Commerce MCP server
communitystreamable_httpremotewrite capablehealthy
Search multi-merchant supply, checkout, and track orders via MCP.
01Tools · 31
How to read this: tool names here are observed from a live tools/list handshake. The Risk label is a heuristic inferred from the tool name (write/destructive verbs), not from executing the tool — a conservative guess, not a verified capability. We never escalate risk from a description. Found one that's wrong? Tell us — we fix on report.
| Tool | Risk | Side effects | Approval |
|---|---|---|---|
| payment_mandate AP2 spend mandates. action=request (needs scope) → approval_url; list; update (mandate_id+scope); revoke (mandate_id). Widening needs human re-approval; narrowing applies immediately. | read | false | unknown |
| agent_create Register an agent (ES256 P-256 public JWK JSON string + optional payment rails + destination). BEFORE: read ~/.conduit/credentials.json — if it exists, reuse agent_id (do NOT register again). AFTER: write credentials (version, agent_id, public_key, private_key as JWKs; recommend handle; chmod 0600). Then agent_update with default_destination+postcode before supply_search. Optional friendly_name / human_description / role_description only if the human stated them. Omit handle to auto-generate (e.g. parcel-watcher-12). | write | true | unknown |
| agent_update Patch handle, rails, default_destination (postcode required for ships-to), friendly_name, human_description, role_description, business profile, or preferences. | write | true | unknown |
| agent_report_issue Report unexpected tool errors or confusing Conduit outcomes for AX review (agent_report_issue — not order_feedback). Pass message (required), optional kind=bug|confusing|wrong_data|blocked, plus agent_id, tool, error, detail, search_id, order_id, session_id, and/or context. Dedupes open reports with the same tool+error+correlation. Does not change reputation. | unknown | unknown | unknown |
| order_execute Handoff or autonomous checkout. Autonomous needs badge payable_now + active mandate. Use idempotency_key. Prefer supply_delivery when only comparing delivery. | write | true | unknown |
| order_update_status Agent manual correction or degraded handoff recovery. note required except cancelled. Owning agent_id required. | write | true | unknown |
| order_feedback Attest delivery outcome (outcome=on_time|late|never_arrived|damaged|wrong_item). Requires owning agent_id. Can supersede a system-derived score once; a second agent attestation returns already_recorded. Poor outcomes next→order_dispute. Omit quality/carrier_rating if unknown. | write | true | unknown |
| agent_authenticate Two-step re-auth. Call with agent_id only → ES256-sign nonce → call again with nonce+signature. Prefer keys from ~/.conduit/credentials.json. Do not agent_create if credentials already exist. | unknown | unknown | unknown |
| agent_organization Read the linked organization (name, slug, country, default address) for an agent. Returns org_not_linked when organization_id is unset — dashboard claim/join is not released yet; do not invent join_org. | read | false | unknown |
| supply_search Multi-provider discovery (live REAL merchants by default). Returns a ranked page (default limit=30, max 100) with total/has_more/next_offset. Follow next.args (search_id+offset+limit) to page without re-fanout. Optional fetch_limit (max 300) deepens the upstream pull on new searches. Pass include_sandbox=true only to append DEMO/SANDBOX test merchants at the bottom (testOffer=true). Always pass agent_id for mandate-aware badges. Carry search_id through supply_details / order_execute / order_feedback. | read | false | unknown |
| supply_details Full normalized product detail from search cache. For shipping ETA/cost use supply_delivery. Re-run supply_search if offer_not_in_cache. | read | false | unknown |
| supply_delivery Non-committing checkout probe for shipping ETA/cost. Uses agent default_destination, else linked org default address, or country override. | unknown | unknown | unknown |
| order_list List orders for an agent ({ orders }). Optional status or open_only filter. | write | true | unknown |
| order_track Honest deal-type-aware status (never fabricates carrier scans). Requires owning agent_id. | write | true | unknown |
| order_events Lifecycle timeline — transitions only. Pair with order_track. Requires owning agent_id. | write | true | unknown |
| order_dispute Refund/chargeback/report paths for an order. Requires owning agent_id. | write | true | unknown |
| create_agent_profile Register an agent (public key + optional payment rails seed + default destination). BEFORE calling: read ~/.conduit/credentials.json (or CONDUIT_CREDENTIALS_PATH) — if it exists, reuse that agent_id and do NOT register again. Hosts without filesystem (Claude app / remote MCP): do not invent a new profile — give the human `jq -r .agent_id ~/.conduit/credentials.json` and continue from their paste. AFTER success: immediately write that file with version, agent_id, public_key, and the private key you generated (chmod 0600); if you cannot write disk, give a mkdir/cat/chmod terminal command instead. Then set default_destination with postcode via update_agent_profile before search_supply. Returns agent_id + persist guide. | write | true | unknown |
| authenticate_agent Two-step re-auth. Prefer loading agent_id + private_key from ~/.conduit/credentials.json (or CONDUIT_CREDENTIALS_PATH): call with agent_id only → sign nonce with private_key → call again with agent_id + nonce + signature. Hosts WITHOUT filesystem (Claude app / remote MCP): do NOT invent ES256/openssl — (1) ask human `jq -r .agent_id ~/.conduit/credentials.json`; (2) call with agent_id only; (3) forward response `terminal` verbatim; (4) paste their JSON into the second call. Never paste private_key into chat. Do not create_agent_profile if credentials already exist. | unknown | unknown | unknown |
| update_agent_profile Patch handle, payment rails, default destination, business profile, or preferences. Same fields as create_agent_profile; only pass what changes. | write | true | unknown |
| payment_methods List or enable agent payment methods (action=list|enable) with friendly labels. Defaults: cod + x402. invoice (B2B) is coming_soon. bank_card is handoff-only. | read | false | unknown |
| search_supply Multi-provider discovery. Always pass agent_id for mandate-aware badges. Decision fields: badge + action (and recommended). Ignore autonomy if present. Carry search_id through get_supply_details / execute_order / submit_feedback. Read badgeCounts, hint, and partial.merchantErrors. | read | false | unknown |
| get_supply_details Full normalized product detail for one offer from the search cache. Pass agent_id to refresh badge/action for current mandate state. Catalog deliveryOptions are often empty — for shipping ETA/cost call get_delivery_estimate (non-committing checkout probe). Re-run search_supply if offer_not_in_cache; pass search_id for analytics correlation. | read | false | unknown |
| get_delivery_estimate Non-committing checkout probe for authoritative shipping options (etaText, options[], shippingCost) to a destination. Prefer this over execute_order when comparing delivery/price — does not create a Conduit order. Requires agent_id (uses default_destination with postcode) or country. Shopify Global Catalog offers only in v0. | read | false | unknown |
| execute_order Complete an order only when offer badge is payable_now (autonomous, needs active AP2 mandate) or handoff (returns continue_url). Handoff responses include billBreakdown (shipping/tax/total) from the merchant checkout — useful after you have chosen an offer; for delivery comparison without creating an order, use get_delivery_estimate first. Do not call for coming_soon, discovery_only, or needs_mandate (request_payment_mandate first). Prefer search recommended / action. Use idempotency_key. handoffQuality full = auto-poll on track_order; degraded = recover via update_order_status. | write | true | unknown |
| request_payment_mandate Create a standing AP2 spend mandate; returns approval_url. | read | false | unknown |
| manage_mandate List, update, or revoke AP2 mandates. Update with wider caps returns approval_url (pending human re-approval); narrowing applies immediately. | read | false | unknown |
| get_dispute_path Resolve refund/chargeback/report paths for an order. | read | false | unknown |
| update_order_status Agent manual status correction or recovery. note required except cancelled. Attach tracking_id, carrier, external_checkout_id, handoff_endpoint to recover degraded handoffs when auto-poll did not wire checkout. Locks agent status when not PENDING; merchant poll will not demote locked status. | write | true | unknown |
| submit_feedback Report delivery/quality outcome; updates supplier + agent reputation. | unknown | unknown | unknown |
| get_order_events Lifecycle timeline for an order — transitions only (coalesced merchant poll periods). Use track_order for current status and merchant_observed snapshot. When merchant_observed.conflicts_with_agent is true, read events then decide via update_order_status or get_dispute_path. | write | true | unknown |
| track_order Honest deal-type-aware order status (never fabricates carrier scans). Triggers handoff poll when checkout ids are wired (~30s throttle). Returns merchant_observed, status_source, agent_status_locked, and recovery hints for degraded handoffs. | write | true | unknown |
02Install & source
https://mcp.aiconduit.app/mcp
remote_url03Access granted
Manage e-commerce · write
The access this server can exercise, inferred from its verified tools — not a declared OAuth scope.
05Provenance & freshness
sourcesOfficial MCP Registry [p1]
last_checked2026-08-15 01:20Z
next_check2026-08-17 01:12Z
cadenceevery 48h
verifiedtools_list:passed handshake:passed metadata:failed tools_list:passed handshake:passed metadata:failed tools_list:passed handshake:passed metadata:failed tools_list:passed
index_statusindex — 6 unique facts >= 5
06Badge
Add the “as seen on MCPExplorer” badge to your README.
[](https://mcpexplorer.com/servers/conduit-agentic-commerce)
Next step
This is one server. A loadout combines the right servers, governance, and proven plays for a whole job — assembled deliberately, not tool-dumped.
Explore loadouts →