# AgentUI - Full LLM Reference

## Service Identity
- service: agentui
- base_url: https://agentui.platphormnews.com
- product_name: PlatPhorm AgentUI
- purpose: human-facing visual workflow orchestration, MCP tool UI rendering, schema-to-form previews, OpenAPI operation interfaces, Phorm design previews, Layout/ASCII block output viewing, CLI workflow commands, approvals, and protected operator cockpit
- phase3_product_truth: AgentUI turns MCP tools, OpenAPI operations, JSON Schemas, webhook payloads, prompts, Sandbox scripts, BrowserOps journeys, Evals scorecards, Docs reports, Sheets matrices, Trace evidence, Monitor checks, Phorm designs, Layout packets, ASCII outputs, Claws toolchains, Fingerprint summaries, OpenContent objects, Podcasts workflows, and platphormctl commands into usable artifacts.
- mcp_endpoint: https://agentui.platphormnews.com/api/mcp
- openapi_json: https://agentui.platphormnews.com/api/docs
- openapi_yaml: https://agentui.platphormnews.com/openapi.yaml
- asyncapi_yaml: https://agentui.platphormnews.com/asyncapi.yaml
- web4_manifest: https://agentui.platphormnews.com/.well-known/web4.json
- provenance_index: https://agentui.platphormnews.com/.well-known/provenance.json
- web4_status: https://agentui.platphormnews.com/api/web4/status
- route_evidence: https://agentui.platphormnews.com/api/web4/route-evidence

## Inventory
- public_workflows: 0
- total_registry_workflows: 37
- protected_or_private_workflows: 37
- runs: 9
- templates: 46
- built_in_templates: 46
- local_agentui_mcp_tools: 87
- discovered_network_mcp_tools: 140
- inventory_source: live
- storage_mode: server_protected
- server_persistence_available: true
- storage_degraded_reason: none

## Template Registry
- mcp-tool-form: MCP Tool Form (platphorm-native)
- openapi-operation-form: OpenAPI Operation Form (platphorm-native)
- json-schema-form: JSON Schema Form (platphorm-native)
- webhook-payload-form: Webhook Payload Form (platphorm-native)
- protected-action-panel: Protected Action Panel (platphorm-native)
- claws-toolchain-planner: Claws Toolchain Planner (platphorm-native)
- spec-validation-workflow: Spec Validation Workflow (platphorm-native)
- sandbox-test-runner: Sandbox Test Runner (platphorm-native)
- browserops-journey-launcher: BrowserOps Journey Launcher (platphorm-native)
- evals-scorecard-review: Evals Scorecard Review (platphorm-native)
- monitor-health-widget: Monitor Health Widget (platphorm-native)
- trace-timeline-viewer: Trace Timeline Viewer (platphorm-native)
- docs-report-publisher: Docs Report Publisher (platphorm-native)
- sheets-matrix-viewer: Sheets Matrix Viewer (platphorm-native)
- phorm-design-handoff: Phorm Design Handoff (platphorm-native)
- layout-score-review: Layout Score Review (platphorm-native)
- ascii-output-viewer: ASCII Output Viewer (platphorm-native)
- fingerprint-redaction-review: Fingerprint Redaction Review (platphorm-native)
- opencontent-submission-flow: OpenContent Submission Flow (platphorm-native)
- podcasts-episode-workflow: Podcasts Episode Workflow (platphorm-native)
- catalog-capability-creator: Catalog Capability Creator (platphorm-native)
- platphormctl-command-recipe: platphormctl Command Recipe (platphorm-native)
- ide-integration-config: IDE Integration Config (platphorm-native)
- local-draft-artifact: Local Draft Artifact (platphorm-native)
- share-packet: Share Packet (platphorm-native)
- mcp-tool-test: MCP Tool Test (mcp)
- openapi-endpoint-test: OpenAPI Endpoint Test (api)
- content-pipeline: Content Pipeline (content)
- podcast-pipeline: Podcast Pipeline (media)
- docs-to-sheets: Docs to Sheets (reporting)
- sandbox-tool-test: Sandbox Tool Test (validation)
- browserops-ui-check: BrowserOps UI Check (validation)
- agentui-tool-render-check: AgentUI Tool Render Check (renderer)
- spec-evals-sandbox-browserops-loop: Spec to Evals to Sandbox to BrowserOps Loop (release-gate)
- dictionary-entry-workflow: Dictionary Entry Workflow (content)
- emoji-proposal-workflow: Emoji Proposal Workflow (content)
- echo-content-monitor: Echo Content Monitor (monitoring)
- raindeer-demo-check: Raindeer Demo Check (demo)
- api-registry-validation: API Registry Validation (registry)
- mcp-gateway-validation: MCP Gateway Validation (mcp)
- calendar-kanban-remediation-flow: Calendar/Kanban Remediation Flow (remediation)
- phorm-prompt-to-interface-flow: Phorm Prompt-to-Interface Flow (phorm)
- phorm-design-to-mcp-tool-ui-flow: Phorm Design-to-MCP Tool UI Flow (phorm)
- cli-publish-and-validate-flow: CLI Publish and Validate Flow (cli)
- human-approval-handoff: Human Approval Handoff (approval)
- route-compliance-report: Route Compliance Report (compliance)

## MCP Tool UI Renderer
AgentUI renders MCP tool inputSchema objects into form fields, validation state, JSON preview, MCP JSON-RPC examples, curl examples, platphormctl examples, protected-action panels, and raw JSON fallback.

Local AgentUI MCP tools:
- get_agentui_info: public - Return AgentUI product identity, public/protected boundary, and integration purpose.
- get_info: public - Alias for get_agentui_info.
- get_health: public - Return public AgentUI health, metrics-source, MCP, discovery, and degraded-state summary.
- get_storage_status: public - Return AgentUI storage truth model for local browser, generated snapshot, delegated, exported, and server persistence modes.
- list_artifacts: public - List server-visible AgentUI artifacts or honest degraded artifact storage status.
- get_artifact: public - Get one AgentUI artifact storage status by id.
- create_local_artifact: public - Create a generated AgentUI artifact snapshot for local/export/delegation use without claiming server persistence.
- clone_artifact: public - Clone an AgentUI artifact packet as a generated snapshot.
- export_artifact: public - Export an AgentUI artifact as JSON, YAML, or Markdown.
- create_share_packet: public - Create a reusable AgentUI share packet with JSON, YAML, Markdown, MCP JSON-RPC, and platphormctl examples.
- list_workflows: public - List persisted AgentUI workflows from the server registry.
- get_workflow: public - Read one persisted AgentUI workflow by id.
- create_workflow: protected - Create a server-persisted AgentUI workflow.
- update_workflow: protected - Update a server-persisted AgentUI workflow.
- validate_workflow: public - Validate a workflow graph by id or supplied nodes and edges.
- plan_workflow: public - Return a public-safe execution plan for a workflow without executing protected tools.
- dry_run_workflow: public - Dry-run a supplied workflow graph locally or, when authorized, persist a trace-linked dry-run record.
- run_workflow: protected - Execute a workflow through the Phase 1 runtime adapter and persist run results.
- get_workflow_run: public - Read a persisted workflow run by id.
- cancel_workflow_run: protected - Cancel a persisted workflow run.
- list_templates: public - List public AgentUI workflow templates, including built-in Phase 1 templates.
- get_template: public - Get one public AgentUI workflow template by id.
- instantiate_template: protected - Create a server workflow from a public AgentUI template.
- list_approvals: public - List workflow approval records with approval notes redacted.
- approve_step: protected - Approve a pending workflow approval step.
- reject_step: protected - Reject a pending workflow approval step.
- list_mcp_tools: public - List AgentUI MCP tools and their public/protected boundaries.
- render_mcp_tool_form: public - Render an MCP tool inputSchema into AgentUI form fields, examples, and raw JSON fallback state.
- render_openapi_operation_form: public - Render an OpenAPI operation into AgentUI form fields and request examples.
- render_json_schema_form: public - Render a JSON Schema into AgentUI form fields, validation preview, and raw JSON fallback state.
- render_webhook_schema_form: public - Render a webhook payload schema into AgentUI form fields without claiming webhook delivery.
- validate_form_input: public - Validate required fields for a rendered form input without executing a tool.
- render_ascii_conversion_form: public - Render the ASCII image conversion request schema as an AgentUI form.
- execute_ascii_conversion: public - Convert a bounded image payload through ascii.platphormnews.com and return the concrete ASCII text output or an honest degraded state.
- preview_layout_artifact: public - Preview a layout document or template through layout.platphormnews.com and return the confirmed preview output.
- score_layout_artifact: public - Score a layout document or template through layout.platphormnews.com and return the confirmed score output.
- export_layout_artifact: public - Export a layout document or template through layout.platphormnews.com and return the confirmed export packet.
- preview_phorm_design: public - Return a deterministic Phorm prompt-to-interface preview or degraded state.
- convert_phorm_design_to_workflow: public - Convert a confirmed or fallback Phorm design reference into a local AgentUI workflow draft plan.
- list_cli_examples: public - List public platphormctl examples for AgentUI workflows, tool rendering, and validation harnesses.
- get_cli_example: public - Read one AgentUI platphormctl example by id.
- list_integrations: public - List AgentUI integration matrix entries with honest configured/degraded state.
- delegate_artifact: public - Create a degraded/dry-run delegation packet for an AgentUI artifact and PlatPhorm target service.
- generate_platphormctl_command: public - Generate a reproducible platphormctl command for AgentUI render, artifact, delegation, or workflow tasks.
- get_integration_status: public - Get one integration status card by slug.
- list_rendered_forms: public - Report rendered form persistence status.
- get_rendered_form: public - Report one rendered form by id when durable form storage is available.
- get_public_capabilities: public - Return public AgentUI capabilities and protected-action boundaries.
- test_mcp_tool: public - Validate that a named AgentUI MCP tool is registered and whether it requires PLATPHORM_API_KEY.
- execute_mcp_tool: protected - Execute a registered AgentUI MCP tool by name.
- get_route_compliance: public - Return AgentUI route compliance status from currently available local registry state.
- get_discovery_compliance: public - Return AgentUI discovery compliance status from currently available local registry state.
- get_agent_policy: public - Return AgentUI public/protected agent policy summary.
- list_agent_platforms: public - List agent platforms covered by generic AgentUI public access policy.
- get_agent_platform: public - Get one agent platform policy by id.
- evaluate_agent_access: public - Evaluate whether an agent action is public-safe or requires PLATPHORM_API_KEY.
- get_robots_policy: public - Return AgentUI robots policy without exact AI crawler strings unless verified.
- get_ai_policy: public - Return AgentUI public-safe AI access and model-assisted output policy.
- get_agentui_web4_manifest: public - Return the AgentUI Web4 manifest without exposing private workflow state or sensitive fingerprints.
- get_agentui_web4_status: public - Return AgentUI Web4 readiness, source-labeled route evidence, storage truth, integration state, and privacy boundaries.
- get_agentui_scorecard: public - Return the AgentUI Web4 scorecard with route, renderer, policy, trace, and provenance readiness checks.
- list_agentui_fingerprints: public - List public-safe AgentUI fingerprints with privacy class and anchor eligibility.
- lookup_agentui_provenance: public - Lookup public-safe AgentUI provenance by url, hash, workflowId, runId, artifactId, delegationId, or formId.
- verify_agentui_provenance: public - Verify a public-safe AgentUI provenance record by recordId or hash.
- get_trust_policy: public - Return AgentUI trust policy and required PLATPhorm public/protected line.
- get_discovery_manifest: public - Return AgentUI discovery manifest links.
- get_public_access_summary: public - Return public read-only and protected action summary for AgentUI.
- publish_workflow: protected - Publish a workflow by marking it public and published.
- save_artifact: protected - Protected durable artifact save request. Returns degraded when no artifact persistence adapter is configured.
- update_artifact: protected - Protected durable artifact update request. Returns degraded when no artifact persistence adapter is configured.
- delete_artifact: protected - Protected durable artifact delete request. Returns degraded when no artifact persistence adapter is configured.
- publish_artifact: protected - Protected artifact publishing request. Returns degraded without target confirmation.
- create_docs_report: protected - Protected Docs report creation request. Returns degraded unless Docs confirms the result.
- create_sheet_export: protected - Protected Sheets export request. Returns degraded unless Sheets confirms the result.
- create_catalog_entry: protected - Protected Catalog registration request. Returns degraded unless Catalog confirms the result.
- trigger_sandbox: protected - Protected Sandbox trigger request. Returns degraded unless Sandbox confirms execution.
- trigger_browserops: protected - Protected BrowserOps trigger request. Returns degraded unless BrowserOps confirms evidence.
- trigger_evals: protected - Protected Evals trigger request. Returns degraded unless Evals confirms a scorecard.
- import_phorm_design: protected - Protected Phorm design import placeholder that does not fake imports without a confirmed artifact.
- trigger_eval: protected - Protected eval trigger placeholder that reports degraded when no eval adapter is configured.
- trigger_sandbox_validation: protected - Protected Sandbox validation trigger placeholder that reports degraded when no sandbox adapter is configured.
- trigger_browserops_check: protected - Protected BrowserOps trigger placeholder that reports degraded when no BrowserOps adapter is configured.
- publish_docs_report: protected - Protected Docs report publish placeholder that reports degraded when no Docs adapter is configured.
- create_sheet_report: protected - Protected Sheets report export placeholder that reports degraded when no Sheets adapter is configured.
- sync_network_tools: protected - Protected network MCP tool registry sync placeholder.
- update_agent_policy: protected - Protected agent policy mutation placeholder.
- refresh_agent_platform_registry: protected - Protected agent platform registry refresh placeholder.

## Block Input And Output Inspector
- view_pane_contract: selected blocks expose run id, workflow id, step id, block id, status, mode, input, output, output type, artifacts, evidence, result reference, trace id, timestamps, public/protected state, and degraded reason.
- ascii_output: returned ASCII text is displayed exactly as monospace output and exported as a text/ascii artifact when confirmed.
- docs_output: Docs publishing is treated as successful only when Docs returns a slug, URL, id, or location that AgentUI can display as a result reference.
- layout_output: Layout preview, score, and export outputs are shown only from confirmed layout.platphormnews.com responses; otherwise the block remains degraded with its packet still viewable.
- no_fake_output: missing target-service results remain unavailable or degraded. AgentUI does not infer URLs, ASCII art, layout scores, evidence, or publish state.

## Public Read-only
- /
- /templates
- /gallery
- /changelog
- /docs
- /integrations
- /clients/cli
- /clients/platphormctl
- /clients/ide
- /clients/vscode
- /clients/cursor
- /clients/codex
- /clients/mcp
- /faq
- /api/health
- /api/v1/health
- /api/docs
- /openapi.yaml
- /openapi.json
- /llms.txt
- /llms-full.txt
- /llms-index.json
- /robots.txt
- /sitemap.xml
- /rss.xml
- /feed.xml
- /manifest.webmanifest
- /.well-known/mcp.json
- /.well-known/web4.json
- /.well-known/provenance.json
- /.well-known/agents.json
- /.well-known/agent-policy.json
- /.well-known/ai-policy.json
- /.well-known/security.txt
- /.well-known/trust.json

## Protected Actions (PLATPHORM_API_KEY required)
- workflow create/update/delete/run/dry-run/publish
- approval decisions
- MCP tool execution
- sync/test/report trigger endpoints
- protected Phorm imports
- BrowserOps/Evals/Sandbox/Claws triggers
- Docs/Sheets/Catalog publishing

## Protected Action Headers
- Authorization: Bearer $PLATPHORM_API_KEY
- X-PlatPhorm-API-Key: $PLATPHORM_API_KEY

## Core Workflow Endpoints
- GET /api/v1/workflows
- POST /api/v1/workflows
- GET /api/v1/workflows/{id}
- PUT /api/v1/workflows/{id}
- DELETE /api/v1/workflows/{id}
- POST /api/v1/workflows/{id}/validate
- POST /api/v1/workflows/{id}/run
- POST /api/v1/workflows/{id}/dry-run
- POST /api/v1/workflows/{id}/blocks/{blockId}/dry-run
- POST /api/v1/workflows/{id}/blocks/{blockId}/run
- POST /api/v1/workflows/{id}/publish
- POST /api/v1/render/mcp-tool
- POST /api/v1/render/openapi-operation
- POST /api/v1/render/json-schema
- POST /api/v1/render/webhook-schema
- POST /api/v1/render/artifact
- POST /api/v1/render/workflow-node
- POST /api/v1/validate/form-input
- GET /api/v1/artifacts
- POST /api/v1/artifacts
- GET /api/v1/artifacts/{id}
- GET /api/v1/artifacts/{id}/view
- POST /api/v1/artifacts/{id}/clone
- POST /api/v1/artifacts/{id}/share
- POST /api/v1/artifacts/{id}/export
- POST /api/v1/artifacts/{id}/delegate/{service}
- GET /api/v1/delegations
- GET /api/v1/delegations/{id}
- POST /api/v1/delegations/{id}/retry
- POST /api/v1/delegate/{service}
- Delegation status contract: dry-run packets return preview_ready, missing protected auth returns protected_required, unknown targets return not_configured, and resultRef remains null until the target service confirms a result.
- GET /api/v1/storage/status
- GET /api/v1/storage/status?probe=true
- GET /api/v1/storage/modes
- POST /api/v1/storage/export
- POST /api/v1/storage/import
- GET /api/v1/rendered-forms/{id}
- GET /api/v1/runs
- GET /api/v1/runs/{id}
- GET /api/v1/runs/{id}/steps
- GET /api/v1/runs/{id}/steps/{stepId}
- GET /api/v1/runs/{id}/blocks/{blockId}
- GET /api/v1/runs/{id}/trace
- GET /api/v1/runs/{id}/artifacts
- POST /api/v1/runs/{id}/cancel
- GET /api/v1/integrations
- GET /api/v1/integrations/status
- GET /api/v1/integrations/{slug}
- GET /api/v1/integrations/phorm/status
- GET /api/v1/integrations/cli/status
- GET /api/v1/phorm/templates
- POST /api/v1/phorm/design-preview
- POST /api/v1/phorm/import-design
- POST /api/v1/phorm/convert-to-workflow
- GET /api/v1/phorm/exports/{id}
- GET /api/v1/cli/examples
- GET /api/v1/cli/examples/{id}
- POST /api/v1/cli/validate-command
- GET /api/v1/agent-policy
- GET /api/v1/agent-policy/platforms
- GET /api/v1/agent-policy/robots
- GET /api/v1/agent-policy/summary
- GET /api/web4/status
- GET /api/web4/manifest
- GET /api/web4/scorecard
- GET /api/web4/fingerprints
- GET /api/web4/route-evidence
- GET /api/provenance/lookup
- GET /api/provenance/verify

## MCP JSON-RPC Methods
- initialize
- ping
- tools/list
- tools/call
- resources/list
- resources/read
- prompts/list
- prompts/get

## Integrations
- network_graph: https://platphormnews.com/api/network/graph
- base_sitemap_index: https://base.platphormnews.com/sitemap-index.xml
- mcp: https://mcp.platphormnews.com
- trace: https://trace.platphormnews.com
- docs: https://docs.platphormnews.com
- mcp: https://mcp.platphormnews.com (configured)
- api: https://api.platphormnews.com (degraded)
- spec: https://spec.platphormnews.com (configured)
- webhooks: https://webhooks.platphormnews.com (configured)
- evals: https://evals.platphormnews.com (configured)
- sandbox: https://sandbox.platphormnews.com (configured)
- browserops: https://browserops.platphormnews.com (configured)
- claws: https://claws.platphormnews.com (degraded)
- monitor: https://monitor.platphormnews.com (configured)
- trace: https://trace.platphormnews.com (configured)
- docs: https://docs.platphormnews.com (configured)
- sheets: https://sheets.platphormnews.com (degraded)
- catalog: https://catalog.platphormnews.com (degraded)
- fingerprint: https://fingerprint.platphormnews.com (degraded)
- opencontent: https://opencontent.platphormnews.com (degraded)
- podcasts: https://podcasts.platphormnews.com (degraded)
- json: https://json.platphormnews.com (degraded)
- xml: https://xml.platphormnews.com (degraded)
- markdown: https://markdown.platphormnews.com (degraded)
- phorm: https://phorm.platphormnews.com (configured)
- layout: https://layout.platphormnews.com (configured)
- ascii: https://ascii.platphormnews.com (configured)
- ai: https://agentui.platphormnews.com (configured)
- cli: https://cli.platphormnews.com (configured)
- root: https://platphormnews.com/api/network/graph (configured)

## Layout And ASCII Integrations
- layout: https://layout.platphormnews.com is an integration for preview, scoring, export, and design evidence. AgentUI keeps its own identity as the interface layer.
- ascii: https://ascii.platphormnews.com is an integration for image-to-ASCII conversion and returned text artifacts. AgentUI renders the confirmed text output in the block view pane.
- docs_handoff: ASCII and Layout outputs can become Docs report packets, but Docs publish URLs are displayed only after a confirmed Docs response.

## Model Assistance Boundary
- configured_when: PLATPHORM_AI, VERCEL_OIDC_TOKEN, or AI Gateway server configuration is present.
- search_context_when: SEARCH_URL is present.
- platform_auth: PLATPHORM_API_KEY remains the only protected-action key.
- policy: model-assisted labels, summaries, search context, and plans must be labeled advisory and cannot replace renderer validation, block output, target-service execution, or persistence truth.

## CLI
- docs: https://cli.platphormnews.com
- platphormctl site inspect agentui --best-effort [dry-run/best-effort]
- platphormctl mcp validate agentui --best-effort [dry-run/best-effort]
- platphormctl policy inspect agentui --best-effort [dry-run/best-effort]
- platphormctl agentui tools [dry-run/best-effort]
- platphormctl agentui render-tool mcp get_health --dry-run [dry-run/best-effort]
- platphormctl agentui form-schema mcp get_health --dry-run [dry-run/best-effort]
- platphormctl agentui validate-tool-ui mcp get_health --dry-run [dry-run/best-effort]
- platphormctl browserops validate-tool-ui mcp get_health --dry-run [dry-run/best-effort]
- platphormctl harness run agentui-tool-render-check --dry-run [dry-run/best-effort]
- platphormctl harness run spec-evals-browserops-loop --dry-run [dry-run/best-effort]
- PLATPHORM_API_KEY=$PLATPHORM_API_KEY platphormctl agentui publish-workflow <workflow-id> --protected --yes [protected]

## Trace and Metadata
- AgentUI accepts traceparent and tracestate.
- AgentUI propagates X-PlatPhorm trace headers for workflow, run, node, rendered form, approval, Phorm design, CLI example, and agent run identifiers where applicable.
- x-vercel-ja4-digest is fingerprint-adjacent metadata. Raw values are never exposed in public UI, feeds, sitemap, llms files, OpenAPI examples, client logs, or public traces. Public display must be hashed or redacted.

## Web4, Fingerprints, And Provenance
- web4_definition: rendered interfaces, workflows, approvals, artifacts, tool forms, schema previews, dry-runs, execution plans, delegation packets, evidence panels, and result views are discoverable, inspectable, trace-linked, machine-readable, human-verifiable, provenance-ready, and safe to hand off.
- manifest: /.well-known/web4.json
- scorecard: /api/web4/scorecard
- fingerprints: /api/web4/fingerprints
- provenance_lookup: /api/provenance/lookup
- provenance_verify: /api/provenance/verify
- route_evidence: /api/web4/route-evidence returns declared evidence by default and fetches the fixed AgentUI route list only with fetch=true.
- storage_probe: /api/v1/storage/status?probe=true verifies configured database accessibility and returns table counts/schema presence only; it does not return row payloads, secrets, raw connection strings, or private workflow inputs.
- public_anchor_eligible: workflow templates, public workflow records, public run summaries, rendered forms, public artifacts, public delegation envelopes, integration registry, route evidence, llms files, sitemap, OpenAPI, AsyncAPI, MCP manifest, trust policy, agent policy, AI policy, release manifest, optional CID.
- never_anchor: raw IP, raw user agent, raw headers, cookies, session IDs, auth tokens, raw PLATPHORM_API_KEY, raw JA4, raw x-vercel-ja4-digest, browser fingerprint, device fingerprint, behavioral identity, private workflow input, protected tool arguments, protected tool results, private approvals, private artifacts, private run logs, private report exports.

## Phorm
- status: degraded deterministic fallback unless a live Phorm response confirms a design artifact.
- prompt policy: public prompts must not contain secrets or private workflow payloads.
- conversion policy: fallback conversion creates a local draft plan only.

## Trust Policy
Web dashboard, public-safe discovery, browser-based operations, trusted-domain discovery, standard route compliance, Vercel metadata capture, trace inspection, and agentic workflow discovery are intentionally supported for public read-only debugging and operator workflows. Mutating, administrative, ingestion, replay, fork, remediation, deployment, sync, test-triggering, reporting, and write actions require PLATPHORM_API_KEY.

AgentUI product-specific boundary:
Web workflow building, public-safe template discovery, browser-based AgentUI operations, local non-sensitive workflow draft persistence, trusted-domain discovery, standard route compliance, Vercel metadata capture, trace-linked workflow execution, and MCP workflow discovery are intentionally supported for public read-only debugging and operator workflows. Mutating, administrative, workflow execution, approval, MCP tool execution, publishing, sync, test-triggering, reporting, and write actions require PLATPHORM_API_KEY.