v1.18.1 — i18n defaultStage/defaultMode

Settings i18n leftovers

• bddPilot.defaultStage / bddPilot.defaultMode descriptions + enumDescriptions via package.nls EN/ES
• Mode copy clarifies xUnit parallelism ≠ Debug session / dashboard runKind

Install

Download bdd-pilot.vsix below. Marketplace not published in this ship (GitHub release only).

v1.18.0 — Tree↔TE parity

Tree ↔ Test Explorer parity

• Core contract tests for displayMode labels, leaf tags, outcome keys, and shared TRX mapping
• treeDisplaySettings parsers + shared domain/tag structural bases
• Test Explorer reads settings via treeSettings (no coupling to tree provider types)

Install

Download bdd-pilot.vsix below, or wait for Marketplace sync after publish.

v1.17.0 — Matching v2

Matching v2 (FQN / Theory-first)

• TRX / live / history → Gherkin leaves prefer Reqnroll FeatureClass.ScenarioMethod and Theory name: "value" params
• Same scenario title across features and prefix siblings no longer cross-attribute when the TRX carries a feature-class FQN
• Shared helpers in tree, Test Explorer, mapping report path, and failure artifacts

Install

Download bdd-pilot.vsix below, or wait for Marketplace sync after publish.

v1.16.0 — Dashboard runKind

Dashboard runKind

Added

• runKind on history (run | debug | profile) — independent of xUnit parallelism mode
• Recent runs badges — Debug / Profile on Env column (normal runs stay quiet)
• Run API — optional additive runKind on PilotRunHistoryEntryDto (no apiVersion bump)

Changed

• Execution profiles — bddPilot.runProfile sets runKind: profile (Re-run Failed stays run)

432 unit tests · install bdd-pilot.vsix from this release or update from Marketplace.

v1.15.1 — Settings i18n ES

Settings i18n (Capa C parcial)

Changed

• Spanish enumDescriptions for 7 settings: tree display/group/tags/duration, outline row filter, post-run toast, rehydrate on activate
• packageNls.test.ts — validates configuration nls key parity EN/ES

428 unit tests · install bdd-pilot.vsix from this release or update from Marketplace.

v1.15.0 — Env Tier 1

Env Tier 1

Added

• .env.<stage>.local loader — merge order: config/.env.<stage> → .env.<stage>.local → .env.local (last wins)
• Hub env indicator — execution hub tooltip shows env file basenames or optional-missing hint (no values logged)

Changed

• resolveStageEnvFileStatus — core helper for existence-only checks
• Docs — env.example and README three-step load order + gitignore hints

426 unit tests · install bdd-pilot.vsix from this release.

v1.14.0 — Agent Surface

Agent Surface

Added

• Discover enrich — pilot discover [--enrich] and MCP pilot_discover_bdd param enrich (opt-in dotnet test --list-tests; outline examples + executableTestCount)
• Rehydrate artifact — TRX rehydrate with failures writes bdd-pilot-last-failure.json when no newer live artifact exists (TRX-only, no log)
• README Agent recipes — Copilot/Cursor workflow: map → filter → failure context → reopen workspace

Changed

• buildCliDiscoverReport — async with 90s timeout, partial success via listTestsWarnings
• MCP T2 — optional testTarget for enrich list-tests scope

418 unit tests · install bdd-pilot.vsix from this release.

v1.13.0

MCP Fase 2

Extension auto-registers BDD Pilot as an MCP server in VS Code 1.101+ / Cursor (agent mode) — no manual mcp.json when installed from VSIX.

Tools (unchanged from Fase 1): pilot_analyze_log · pilot_discover_bdd · pilot_build_filter · pilot_failure_context

What's new

• MCP provider — registerMcpServerDefinitionProvider discovers server on extension install
• VSIX bundle — dist/pilot-mcp.cjs + headless core packaged in the extension
• Last-failure artifact — after a failed run, writes TestResults/bdd-pilot-last-failure.json so pilot_failure_context works with only projectDir

Install

Download bdd-pilot.vsix below → Extensions → Install from VSIX…

Open a BDD workspace, enable BDD Pilot in Copilot Chat tools (agent mode), run tests, then ask the agent to use pilot_failure_context after a failure.

Repo dev: npm run pilot:mcp still works from a git clone (config/mcp.json.example).

Security: read-only tools; outputs sanitized and path-jailed — review before sharing with external AI.

v1.12.0 — MCP Fase 1

MCP Fase 1

Headless MCP stdio server for Cursor / Claude Desktop (repo clone required):

npm run pilot:mcp

Tools: pilot_analyze_log · pilot_discover_bdd · pilot_build_filter · pilot_failure_context

• Thin wrapper over pilot-cli (read-only)
• Path jail via BDD_PILOT_WORKSPACE_ROOT
• Sanitized outputs + 5 MB artifact cap
• Setup: config/mcp.json.example

Extension VSIX: unchanged UX — install below for the VS Code panel; MCP runs from a git clone.

Install extension: download bdd-pilot.vsix → Extensions → Install from VSIX…

v1.11.0 — pilot-cli v2

pilot-cli v2

Headless JSON subcommands for agents, CI, and pre-MCP tooling:

npm run pilot -- discover <project-dir>
npm run pilot -- build-filter <project-dir> --tag smoke
npm run pilot -- failure-context --project-dir . --trx path.trx --log run.log

• discover — BDD map (domains, features, scenarios, tags)
• build-filter — Reqnroll/xUnit dotnet test --filter (same defaults as the extension)
• failure-context — markdown “Copy for AI” from TRX/log on disk (sanitized)
• Auto-compile out-test/ when missing

Install extension: download bdd-pilot.vsix below → Extensions → Install from VSIX…