Validate token and fetch basic account context.

Run TestNeo PR validation planning from diff or git refs. Returns shared workflow context, impacted tests, planned verification stages, and Claude-ready findings using the new orchestration contracts.

Returns whether a TestNeo self-hosted agent is registered and recently heartbeating (local runner). Includes setup_url on the same origin as TESTNEO_BASE_URL (for example …/web/agent). Read-only.

List web test case ids for a project matching one or more tags. Tags may include a leading @; each tag is queried separately (union with tag_match=any, intersection with all). Backend: GET /api/web/v1/test-cases/?tag_filter= (one tag per request).

List web test cases in a project using the backend text search (?search= on GET /api/web/v1/test-cases/). Matches name/description per API rules. Read-only — use returned id + name with testneo_run_generated_test_pipeline (test_case_id) or pass project_id + name_query to run by name.

Create and execute a multi-test run for tests matching tags (write: requires TESTNEO_MCP_ALLOW_WRITE + confirm=true). Resolves tags like testneo_list_tests_by_tags, then POST /api/web/v1/multi-test-runs/create and POST …/execute. When TESTNEO_MCP_DEFAULT_EXECUTION_MODE=local and TESTNEO_MCP_PREFER_LOCAL_AGENT=true, sets use_agent for the local TestNeo agent. While use_agent: polls GET /agents/my-agent until connected or until TESTNEO_MCP_WAIT_FOR_AGENT_MS / wait_for_agent_seconds elapses (avoids failing if you start the agent seconds after invoking the tool). Optional TESTNEO_MCP_OPEN_AGENT_SETUP_ON_AGENT_FAILURE opens setup_url on hard agent failure.

Scan Swagger/NLP-derived API-style web tests for a project and return suggested business-flow chains (ordering, phases, suite summaries). Read-only: GET /api/web/v1/projects/{id}/api-test-chains/suggest. Use before recommending which suite to run or before testneo_save_api_test_chain / testneo_run_api_test_chain.

List user-saved API test chain suites for a web project (ordered test_case_ids). Read-only: GET /api/web/v1/projects/{id}/api-test-chains.

Persist a named API test chain (ordered NLP/API web tests) for a project. Guarded: TESTNEO_MCP_ALLOW_WRITE + confirm=true. Backend: POST /api/web/v1/projects/{id}/api-test-chains.

Delete a saved API test chain by id. Guarded: TESTNEO_MCP_ALLOW_WRITE + confirm=true. Backend: DELETE /api/web/v1/projects/{id}/api-test-chains/{chain_id}.

Create and execute a multi-test run using an ordered API chain: either explicit test_case_ids (preserves order) or saved_chain_id from testneo_list_saved_api_test_chains. Same execution routing as testneo_run_batch_by_tags (local agent wait, use_agent, etc.). Guarded: TESTNEO_MCP_ALLOW_WRITE + confirm=true.

Get project-level MCP route-hardening map/profile from project_settings.mcp_route_hardening.

Persist project-level route-hardening map/profile in project_settings.mcp_route_hardening (guarded write action).

Rewrite vague Navigate-to NLP lines into {{base_url}}/path using server env (TESTNEO_ROUTE_PROFILE, TESTNEO_ROUTE_MAP_JSON) plus optional per-call overrides. Read-only; does not call the TestNeo API.

Parse Swagger/OpenAPI (JSON or YAML) from base64 and return spec format, tags, and endpoint counts. Read-only; no DB writes. Backend: POST /api/web/v1/ai-test-gen/preview.

Upload Swagger + optional business rules → unified context indexing + NLP web test cases (multipart). Guarded: TESTNEO_MCP_ALLOW_WRITE + confirm=true. Respects project execution preconditions. Large payloads: set TESTNEO_MCP_SWAGGER_TIMEOUT_MS. Backend: POST /api/web/v1/ai-test-gen/upload-and-generate.

Compare an uploaded Swagger revision against the last snapshot for a web project, diff endpoints, and list impacted swagger-sourced NLP tests. Persists the modified spec bytes for the next baseline. Guarded: allow-write + confirm=true. Backend: POST /api/web/v1/ai-test-gen/impact-analysis.

Bulk apply impact triage on web test cases: mark_stale | archive | keep, then promote modified spec snapshot when possible. Guarded: allow-write + confirm=true. Backend: POST /api/web/v1/ai-test-gen/impact-actions.

Upload OpenAPI JSON/YAML to a classic API project (stores spec on Project.openapi_spec). Use before testneo_api_project_openapi_impact. Guarded: allow-write + confirm=true. Backend: POST /api/v1/projects/{id}/upload-openapi (multipart field: file).

Run OpenAPI impact analysis for API (non-web) test cases against a new or stored spec. Pass openapi_spec to diff an inline revision, or omit to analyze using the spec already saved on the project. Guarded: allow-write + confirm=true (service may flag tests). Backend: POST /api/v1/projects/{id}/openapi-impact.

List projects available to the current API key user.

Create a new web automation project (stored under your API key account). Mirrors POST /api/web/v1/projects. Guarded: TESTNEO_MCP_ALLOW_WRITE=true and confirm=true. By default creates a default web environment with base_url (and optional username/password variables) in the same request; set create_default_environment=false for project-only. New projects get Lighthouse performance audits enabled unless project_settings overrides.

Create a named web project environment with optional variables (e.g. base_url for {{base_url}} in NLP). Backend: POST /api/web/v1/projects/{project_id}/environments. Guarded: allow-write + confirm=true.

One-shot onboarding: validate → create web project with optional default environment (base_url + credentials) in a single POST when add_base_url_variable=true. Returns a trace + recommended_next_tools. Guarded like other writes.

List recent executions, optionally filtered by project/status/release/build.

Fetch primary execution status, steps and summary metadata for an execution ID.

Get analytics summary for an execution (status, pass/fail, duration, video metadata).

Get execution logs for an execution ID.

Search failed executions for a project by test name or execution id fragment.

Summarize pass/fail trend for a project over a date range.

Poll execution summary until terminal status or timeout window.

Get a compact failure triage bundle: summary, event sample, logs, inferred theme, next actions, plus a concrete suggested NLP patch (diff + testneo_update_test_case_nlp payload) when execution summary includes test_case_id.

Run an agentic multi-step QA workflow (triage_failure, rerun_decision, qa_intelligence) over TestNeo data.

Ingest Figma metadata via ETL, optionally wait for completion, then create a linked unified context for test generation.

List unified contexts for a project with id + human-readable names. Use before testneo_generate_tests_from_context so agents do not have to scrape context_id from the UI.

Resolve unified context_id from a natural-language name against this project’s contexts (calls list internally). Great for onboarding and demos.

Same Web AI Assistant as the product UI (/web/ai-assistant): natural-language Q&A over a project, optionally scoped to a unified context (PDF/Figma/requirements ingest). POST /api/web/v1/etl/ai-assistant/query. Pass context_id or context_name_query; omit both for project-wide analytics-style questions. Uses your Web AI chat quota. Optional recommend_context / rag_context match the web request body for AI-Q and document-aware answers.

Generate NLP test cases from an existing unified context (Figma, requirements, etc.). Resolve context via testneo_list_unified_contexts or testneo_get_unified_context_by_name (name_query, not scraped UI ids). Omit auth_preamble for public / no-login apps (default: no SauceDemo login injected, no SauceDemo route auto-align). Pass auth_preamble { enabled:true, preset:'saucedemo' } only for demos against saucedemo.com. Custom maps: TESTNEO_ROUTE_MAP_JSON or testneo_set_project_route_map; optional auto_align_saucedemo_route_map + SauceDemo preset ties route phrases to SauceDemo paths.

Preview generated tests in both NLP and Playwright SDK spec.ts draft format for human-in-loop approval. Applies route hardening (env + optional route_hardening) to Navigate-to lines when a phrase map is configured.

Execute a generated test case by numeric test_case_id OR by project_id + name_query (resolved via GET …/test-cases/?search= then exact/unique name match). Human-in-loop gated. POST /api/web/v1/test-cases/{id}/execute — passes use_agent when TESTNEO_MCP_DEFAULT_EXECUTION_MODE=local and TESTNEO_MCP_PREFER_LOCAL_AGENT=true (same routing as batch). Optional wait_for_agent_seconds + env TESTNEO_MCP_WAIT_FOR_AGENT_MS / TESTNEO_MCP_REQUIRE_LOCAL_AGENT_FOR_BATCH control preflight polling. Use testneo_find_test_cases to browse matches before running.

End-to-end: execute a generated test by test_case_id OR (project_id + name_query), poll until terminal, return analytics summary + step-level execution, optional failure triage bundle, and project pass/fail trend. POST /api/web/v1/test-cases/{id}/execute sends use_agent when local+prefer-local (same as testneo_execute_generated_test_case). name_match_mode: auto|exact|substring. Optional wait_for_agent_seconds for preflight.

Update NLP commands for a specific test case ID and return verification snapshot. Optionally rewrites Navigate-to lines using route hardening (same env / route_hardening as generate).

Export a test case as Playwright SDK TypeScript spec text.

Run a Playwright SDK spec preview by extracting ai.run commands and executing via Playwright SDK execute endpoint.

End-to-end workflow: ingest Figma -> create unified context -> generate tests -> preview NLP + Playwright drafts.

No Figma token: upload exported UI image (PNG/JPEG/GIF/WebP) like the product 'Upload Figma Image' flow → wait for vision ETL → create unified context → generate tests → preview. Guarded: TESTNEO_MCP_ALLOW_WRITE + confirm=true. Backend: POST /api/web/v1/etl/upload-figma-image (multipart field: file).

Rerun failed tests for a project using test-case execute endpoint (guarded write action).

Trigger NLP execution via Playwright SDK execute endpoint (write tool, requires confirm + allow-write).