This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@gemstack/ai-autopilot@0.2.0

Minor Changes

• 70168c5: Add bootstrap's deploy phase and the DeployTarget adapter seam. The final phase decides the rendering mode (SSR/SSG/SPA) and the deploy target (Dockploy vs Cloudflare), narrates the plan, and hands it to a DeployTarget — the same pattern as the runner seam. agentDeploy is the default step (an ai-sdk agent decides { render, target, reason }, normalized against the allowed sets); planOnlyTarget is the v1 default that decides and narrates without shipping, and FakeDeployTarget backs tests. v1 decides + narrates only; real Dockploy / Cloudflare adapters implement DeployTarget and are infra-gated follow-ups, so bootstrap never does a blind deploy. The deploy step is optional and its outcome rides on BootstrapResult.deploy. Closes Bootstrap deploy step: decide SSR/SSG/SPA + target, expose adapter seam #123.
• 11dce73: Add bootstrap mode's orchestrator core: the spine that sequences autopilot's primitives into scope → architect → build → full-fledged loop, taking a user from nothing to a running, production-grade app. Bootstrap owns the control flow (the loop, the gate, the interrupt) over four injectable steps, narrating each phase over the generic surface stream and recording the architect's choices to the decisions ledger — no permission asked. The full-fledged loop repeats the production-grade checklist with fresh context, improving against its { blockers } verdict until it is empty or a maxPasses budget stops it; prototype scope skips it. Default step builders wire the steps onto the real primitives — agentArchitect (an ai-sdk agent + the decisions briefing), supervisorBuild (the Supervisor over personas + runner), and loopChecklist / loopImprove (the Loop) — so the same orchestrator runs against real agents in production or stubs + FakeRunner in a test. Verified end-to-end offline. Closes Bootstrap orchestrator core: scope -> architect -> build -> full-fledged loop #122.
• aa1925f: Add scale mode: an always-current CODE-OVERVIEW.md the agent reads first in a large repo so it stays oriented without re-scanning the tree. The hard part is keeping it fresh — a stale overview is worse than none — so refreshes are gated by a deterministic material-change detector (detectMaterialChange): a build/config change, a test-framework migration, a directory restructure, or a large change across many areas, not every routine edit. CodeOverviewMaintainer owns the policy (refresh on material change, skip otherwise, persist over an OverviewFs); agentOverview regenerates the map with an ai-sdk agent (seeding the previous overview so it revises rather than rewrites); and overviewLoopPrompt drops the maintainer into the loop ("The loop" — event-triggered prompt chains (major change → review/quality/security; UI flow → QA/UX) #113) so it self-maintains on major-change. The CODE-OVERVIEW.md markdown round-trips via parseOverview / serializeOverview. Regeneration is injected, so the whole policy is tested offline against a stub. Closes Scale mode — maintain CODE-OVERVIEW.md for large-codebase navigation #114.
• c9b4d0d: Add the decisions ledger: durable project memory so a run stops re-pitching ideas that were already turned down. DecisionLedger records rejected ideas and settled choices and answers consult(idea) / wasRejected(idea) (lexical token-overlap matching, deterministic) before the agent proposes; it round-trips a human-editable DECISIONS.md via loadLedger / saveLedger over a storage-agnostic LedgerFs seam (a subset of the runner's RunnerFs, with a nodeLedgerFs() host adapter). decisionTools(ledger) exposes consult_decisions + record_decision to an agent and decisionBriefing(ledger) renders the rejected set as a system-prompt fragment. Verified end-to-end on real disk. First child (Decisions bookkeeping — stop the AI re-pitching rejected ideas #112) of the AI-framework epic (Epic: The AI framework — turnkey end-to-end AI orchestration for building software (web-app first) #110); the state layer "the loop" ("The loop" — event-triggered prompt chains (major change → review/quality/security; UI flow → QA/UX) #113) will consult.
• 57cbba4: Add the web-app preset seam: framework-specific knowledge selected by detecting the app's framework, on top of the agnostic core. A Preset bundles a framework's personas with the signals that identify it; detectFramework scores a project's dependencies + files (deps weigh more than files) and PresetRegistry.select picks the preset (falling back to the flagship when nothing matches). Ships two built-ins — vikePreset (flagship) and nextPreset — plus a new nextPageBuilder persona (App Router + React Server Components). presetPersonas(preset) returns the framework page builder followed by the shared, framework-neutral personas (sharedPersonas: the universal-orm modeler + intent-UI designer), so only the page builder changes between frameworks while the rest of the stack stays put and prompts stay neutral. One shared core; a new framework is a new Preset, not a runtime fork. Closes Web-app preset: Next.js + Vike targeting (persona/prompt/loop layer) #115.
• 04965c5: Generalize the surface stream to be event-type generic. EventStream<E>, AutopilotHandle<E, R>, and launchAutopilot<E, R> now take the event (and result) type as parameters, defaulting to the supervisor's SupervisorEvent / SupervisorRun so every existing supervisor surface is unchanged. This lets bootstrap (and any future surface) stream its own narration events and return its own result over the same replayable, detached transport. Closes Generalize the surface stream to be event-type generic #120.
• 07cc623: Add LocalRunner, the first real adapter behind the runner seam. Where FakeRunner simulates a workspace in memory, LocalRunner boots each workspace as a real temp directory on the host: real files via node:fs (path-traversal guarded to the workspace root), real commands via child_process (shell, per-command cwd/env/timeoutMs), a localhost preview, and a dispose() that removes the workspace. It is the reference the sandboxed adapters (WebContainer, Docker, Flue) mirror. It runs commands unsandboxed on the host, so it is documented for trusted/CI use only, not untrusted agent-authored code. Part of the ai-autopilot epic (Epic: end-to-end AI orchestration for building web apps (the ai-autopilot thesis) #97), issue ai-autopilot: LocalRunner — first real runner adapter (host fs + child processes) #106.
• 7fde5fc: Add the personas layer: stack-aware roles that make autopilot opinionated about the GemStack stack (Vike + universal-orm) instead of generic. definePersona() builds a role from a system prompt + skills (composed over @gemstack/ai-skills) + tools; personaAgent()/personaWorkers() materialize personas into Supervisor workers; personaRoster() describes them to a planner so plans route to the right role. Ships three built-ins: vikePageBuilder, universalOrmModeler, and uiIntentDesigner (the "declare intent, decouple implementation" UI guardrail). First child (ai-autopilot: personas + prompts library (stack-aware knowledge) #98) of the ai-autopilot end-to-end epic (Epic: end-to-end AI orchestration for building web apps (the ai-autopilot thesis) #97).
• 55b3697: Add the production-grade checklist prompt and a { blockers } verdict convention. The new built-in prompt judges an app against a production-grade checklist (auth, data layer, error handling, instrumentation, emailing, validation, tests, build/config) and ends with a machine-readable { "blockers": [...] } verdict — an empty list means production-grade. parseVerdict() / isPassing() read it back, and the loop now gates on that outcome: a PromptOutcome carries verdict and a passing flag (executed and no blockers), and continueOnError: false stops the chain on !passing. Backward compatible — with no verdict reported, passing === ok; pass verdict: null for an execution-only gate. This is what bootstrap's full-fledged loop repeats against until blockers is empty. Closes Production-grade checklist prompt + { blockers } verdict convention #121.
• 8f780f3: Add the built-in prompts library: the stack-aware prompt bodies the loop dispatches, shipped as data. Eight markdown bundles under prompts/ (review TLDR + thorough, code-quality, security, refactor, UX, QA, knowledge-base) that already know Vike + universal-orm, loaded with builtinLibrary() / loadPromptsFrom(dir) into a PromptLibrary and parsed via @gemstack/ai-skills' frontmatter (a contributor edits prose, not code). loopPromptsFor(library, makeAgent) materializes them into loop prompts so defaultLoopRules() ids resolve to real bodies (a fresh agent per pass); promptInstructions composes a body with the decisions briefing (Decisions bookkeeping — stop the AI re-pitching rejected ideas #112) and renderTask turns a LoopEvent into the worker's task. Closes the turnkey wire across Built-in prompts library (review, quality, security, refactor, UX, QA, knowledge base) #111 / Decisions bookkeeping — stop the AI re-pitching rejected ideas #112 / "The loop" — event-triggered prompt chains (major change → review/quality/security; UI flow → QA/UX) #113. Verified end-to-end against the built package. Child Built-in prompts library (review, quality, security, refactor, UX, QA, knowledge base) #111 of the AI-framework epic (Epic: The AI framework — turnkey end-to-end AI orchestration for building software (web-app first) #110).
• b0c0647: Add the runner seam: a pluggable Runner execution environment (workspace filesystem + shell + optional preview URL), shaped after Flue's sandbox contract so WebContainer, Docker, and Flue sandboxes drop in behind one interface. Ships the interface plus FakeRunner (the runner analog of ai-sdk's AiFake) for infra-free testing, and runnerTools(session) to expose a booted session to an agent as sandbox tools (read_file, write_file, list_files, exec, preview). Real adapters (FlueRunner, WebContainer, Docker) land separately. Interface-first slice of the runner child (ai-autopilot: pluggable execution runner (WebContainer + Docker) #99) of the ai-autopilot epic (Epic: end-to-end AI orchestration for building web apps (the ai-autopilot thesis) #97).
• d261873: Add surfaces: run the same autopilot in the terminal, an in-page UI, or a background process, all over the Supervisor's onEvent stream. terminalSink() prints events inline (formatEvent() renders one event as a line); EventStream is a replayable multi-consumer transport with offset/tail replay (borrowing Flue's Durable-Streams tail=N); launchAutopilot(start) runs a Supervisor detached and returns an AutopilotHandle (status(), events(offset), live async stream(), result()) that backs both the background and in-page (SSE) surfaces. Verified end-to-end against a real Supervisor. Closes the surfaces child (ai-autopilot: surfaces (terminal / in-page / background) #100) of the ai-autopilot epic (Epic: end-to-end AI orchestration for building web apps (the ai-autopilot thesis) #97).
• 481c2f0: Add "the loop": the event-to-prompt-chain policy. The agent declares a semantic LoopEvent (a change kind) and the matching prompt chain fires: a major change runs review + code-quality + security, a new UI flow runs QA + UX. Loop.handle(event) resolves the chain from LoopRules (defineRule / defaultLoopRules() as the built-in policy) and runs each LoopPrompt (definePrompt) for its passes with fresh context every pass; matches(event) is the pure preview, watch(stream) handles a stream fire-and-report. Design calls on the two open questions: the trigger is agent-declared (not heuristic), and both modes ship (handle awaits; continueOnError: false is a blocking gate). Consults the decisions ledger via ctx.ledger, and references the prompts library (Built-in prompts library (review, quality, security, refactor, UX, QA, knowledge base) #111) by id. Verified end-to-end. Child "The loop" — event-triggered prompt chains (major change → review/quality/security; UI flow → QA/UX) #113 of the AI-framework epic (Epic: The AI framework — turnkey end-to-end AI orchestration for building software (web-app first) #110).

@gemstack/connector-github@0.1.0

Minor Changes

• f360932: New package: the GitHub connector for GemStack AI orchestration. Read and act on issues, pull requests, and repository files over the GitHub REST API — get-repo, list-issues, get-issue, list-pull-requests, get-pull-request, get-file, search-issues, comment-on-issue, create-issue. Built with @gemstack/connectors; consumes a bearer token (PAT or OAuth) via the mount credentials seam. First real connector on the contract (epic Epic: AI orchestration — 3rd-party tool connectors (GitHub, Google Drive, …) #86).

Patch Changes

• Updated dependencies [b0430f9]
  • @gemstack/connectors@0.1.0

@gemstack/connector-google-drive@0.1.0

Minor Changes

• 6e37d60: New package: the Google Drive connector for GemStack AI orchestration. Browse, read, and share Drive files over the Drive REST API (v3) — get-about, list-files, search-files, get-file, get-file-content (Docs/Sheets/Slides exported to text, other files downloaded), list-permissions, create-folder, share-file, trash-file. Built with @gemstack/connectors; consumes a Google OAuth 2.0 access token via the mount credentials seam. Second connector on the contract (epic Epic: AI orchestration — 3rd-party tool connectors (GitHub, Google Drive, …) #86).

Patch Changes

• Updated dependencies [b0430f9]
  • @gemstack/connectors@0.1.0

@gemstack/connectors@0.1.0

Minor Changes

• b0430f9: New package: the connector contract for GemStack AI orchestration. defineConnector declares a tool connector to an external service (its auth requirement + tools); mountConnectors composes any number into a single @gemstack/mcp server, namespacing tools by connector id and resolving credentials at call time. Built on @gemstack/mcp, framework-agnostic. First step of the connectors epic (GitHub, Google Drive to follow).

@gemstack/example-autopilot-quickstart@0.0.1

Patch Changes

• Updated dependencies [70168c5]
• Updated dependencies [11dce73]
• Updated dependencies [aa1925f]
• Updated dependencies [c9b4d0d]
• Updated dependencies [57cbba4]
• Updated dependencies [04965c5]
• Updated dependencies [07cc623]
• Updated dependencies [7fde5fc]
• Updated dependencies [55b3697]
• Updated dependencies [8f780f3]
• Updated dependencies [b0c0647]
• Updated dependencies [d261873]
• Updated dependencies [481c2f0]
  • @gemstack/ai-autopilot@0.2.0

@gemstack/example-bootstrap-quickstart@0.0.1

Patch Changes

• Updated dependencies [70168c5]
• Updated dependencies [11dce73]
• Updated dependencies [aa1925f]
• Updated dependencies [c9b4d0d]
• Updated dependencies [57cbba4]
• Updated dependencies [04965c5]
• Updated dependencies [07cc623]
• Updated dependencies [7fde5fc]
• Updated dependencies [55b3697]
• Updated dependencies [8f780f3]
• Updated dependencies [b0c0647]
• Updated dependencies [d261873]
• Updated dependencies [481c2f0]
  • @gemstack/ai-autopilot@0.2.0

@gemstack/example-connectors-quickstart@0.0.1

Patch Changes

• Updated dependencies [b0430f9]
  • @gemstack/connectors@0.1.0