Adobe Target is an anchor experimentation platform because its first-party MCP can inspect test configuration and results, author activities, audiences, and offers, and move activities through live lifecycle states under Target's role model and confirmation gates.
Concrete capability record
A field-by-field summary of what the reviewed first-party references actually support. Publisher update dates and BYO-UI review dates are shown separately below.
A/B, Experience Targeting, Automated Personalization, Auto-Allocate, Multivariate Test, and Recommendations activities; experiences and traffic splits; audiences; HTML, JSON, and redirect offers; properties and mboxes; performance, orders, and A4T reports; QA previews; response tokens; at.js settings; and entity revisions.
List and inspect all documented activity types; retrieve configurations, targeting, offers, goals, metrics, performance, orders, and A4T results; inspect audiences, properties, mboxes, response tokens, at.js settings, revisions, and QA preview links.
Create A/B and Experience Targeting activities; update A/B, Experience Targeting, and Automated Personalization activities; change schedules, names, priorities, traffic splits, variants, offers, and lifecycle state; create or update audiences and offers; and create response tokens. Auto-Allocate, Multivariate Test, and Recommendations activity configurations are read-only even though lifecycle edits are documented across activity types.
An active Target license and Adobe Experience Platform organization are prerequisites. The hosted server uses Adobe IMS OAuth 2.0 with no static credentials, scopes operations to the selected organization, and enforces Target workspace permissions. Observer or higher receives reads, Editor or higher receives create and update tools, and Approver is required for activation and deactivation.
Target write tools carry safety annotations and do not execute state changes without explicit user confirmation. Adobe also recommends sandbox testing and review of every MCP-initiated action. Keep organization, property or workspace, activity state, affected experiences, traffic allocation, schedule, priority, and offer content visible at confirmation.
The service is Public Beta for all Target customers and currently names Claude Web, Claude Desktop, Claude Code, Cursor, and ChatGPT as supported clients. Adobe reports 41 tools. Activity lists return at most 200 records per page, and configuration writes differ by activity type.
Agent access
Support: Official MCP server
Read scope: Experiment portfolio and configuration inspection across six activity types; performance, revenue, orders, and A4T reporting; audience, offer, property, and mbox discovery; QA previews; at.js and response-token inspection; and revision history.
Write scope: Editor-level tools create A/B and XT activities and create or update supported activity configurations, audiences, and offers; other tools alter schedules, priority, lifecycle state, variants, traffic, offer assignment, and response tokens. Approver authority is required for activation and deactivation, and not every activity type supports configuration edits.
Authentication: Adobe IMS OAuth 2.0 through a browser flow: the user signs in, selects an organization, and grants requested permissions. The remote endpoint is `https://targetmcp.adobe.io/mcp`; static credentials are not required, tokens are validated against IMS on each request, and Adobe says the MCP server does not persist them.
Approval boundary: Adobe's MCP confirmation gate applies to state-changing tools, while Target roles independently decide whether the user may execute them. Preserve a separate BYO-UI review for content, audience, traffic, schedule, priority, and live-state consequences before accepting the vendor confirmation.
Confirm the current tool catalog, plan, region, scopes, rate limits, terms, and write behavior before implementation.
Editorial assessment
A comparative architecture lens—not a quality score, market ranking, or buying recommendation. Scale: 1–5.
Proposed customer-shaped experiences
The output could be an export, report, graph, artifact, application, agent, workflow, or downstream feed. These proposals are derived from documented access—not claims that Adobe ships them.
Evidence and dates
“Publisher updated” is shown only when the page exposes an update date. “BYO-UI reviewed” records when this research checked the reference. A missing publisher date is reported as missing—not replaced with the review date.
Public Beta availability, top-level Target MCP object and action scope, supported clients, role thresholds, and native state-change confirmation
Targeted first-party MCP/agent-role review; broader API inventory retained
First-party API reference or API overview recorded for this platform.
Full-profile first-party re-verification on 2026-07-12; see the private audit ledger.
First-party developer documentation or platform overview.
Inventory source: structurally normalized; content was not individually reopened in this pass.
First-party product homepage used to confirm product identity and current positioning.
Inventory source: structurally normalized; content was not individually reopened in this pass.
Exact Target MCP tools, object coverage, write asymmetry, pagination, and role prerequisites
Live first-party tool reference reviewed 2026-07-22.
Deployment endpoint, OAuth, prerequisites, roles, workspace enforcement, and token handling
Live first-party setup and security guide reviewed 2026-07-22.
Tied to cited first-party evidence reviewed for this profile.
Official links recorded for deeper research but not necessarily reopened endpoint by endpoint.
System role, maturity interpretation, and architectural boundary.
Interface patterns and compositions—not vendor product claims.
