feat(growth): add synthetic Dawn research app for LangSmith

.gitignore

@@ -80,3 +80,7 @@ keys/
 
 # Aborted fixture-record runs strand raw recordings under a committed path
 **/e2e/fixtures/.staging/
+
+# Growth research generated deployment artifacts
+apps/growth-research/.dawn/
+apps/growth-research/.deployment/

apps/growth-research/.env.example

@@ -0,0 +1,12 @@
+# Runtime names only. Never copy environment files into the deployment artifact.
+OPENAI_API_KEY=
+DAWN_DATABASE_URL=
+GROWTH_RESEARCH_TEST_DATABASE_URL=
+# Explicit operator-only synthetic invocation gate; blank disables model calls.
+GROWTH_RESEARCH_FIXTURE_MODE=
+# Trusted synthetic memory slot: atlas or beacon. Not an authenticated tenant ID.
+GROWTH_RESEARCH_FIXTURE_SLOT=
+# Optional cancellation probe pause; integer 0..5000, default 0.
+GROWTH_RESEARCH_FIXTURE_DELAY_MS=
+GROWTH_RESEARCH_URL=
+LANGSMITH_API_KEY=

apps/growth-research/README.md

@@ -0,0 +1,147 @@
+# Growth research compatibility application
+
+Private synthetic Dawn application, separate from lifecycle and the Python cockpit.
+Published Dawn packages are pinned to `0.8.24`; the app and deployment require Node 24.
+The only public graph ID is `growth_research`, pointing to the unchanged generated
+Dawn `/enrichment/research#agent` entry. The safe alias avoids slash/hash routing
+failures in the Agent Server's internal per-graph HTTP endpoints. Its registered researcher is
+private to coordinator delegation; staging verifies the known generated specialist
+entry but removes its standalone public graph key.
+
+This app exercises authored plans, skills, scoped delegation, candidate memory and
+platform thread continuation against a fixed synthetic corpus. It is disabled by
+default and has no connection to Growth ingestion or campaign delivery. Configure
+a dedicated memory database; do not point it at Growth's canonical database.
+
+Dawn 0.8.24 sets the child checkpointer to `false`; `task` accepts only `subagent`
+and `input`. Each delegation starts a fresh child conversation. Carry relevant
+context explicitly through the checkpointed parent when delegating follow-up work.
+
+From the workspace root on Node 24:
+
+```sh
+npm ci --ignore-scripts
+npx nx test growth-research
+npx nx run growth-research:check
+npx nx lint growth-research
+npx nx build growth-research
+```
+
+The build uses the CLI resolved from this application and checks its version before
+execution. Dawn emits a LangSmith entry under `.dawn/build`. Packaging preserves
+that entry and the relative `src/` and `dawn.config.ts` layout, stages approved files
+under `.deployment`, and normalizes `langgraph.json` to Node 24, `dependencies: ["."]`
+and `env: {}`. Configure secret values in the deployment environment. Generated
+`.dawn/routes/*/tools.json` schemas are preserved because actual tool execution needs
+them; arbitrary build files and all environment files remain excluded. The artifact
+pins Agent Server `api_version: "0.13.4"` and contains a standalone NodeNext
+`tsconfig.json` for the official server's static schema extractor. It does not inherit
+the monorepo's compiler configuration or path aliases.
+
+`deployment-package-lock.json` is the standalone runtime dependency lock. To update
+it after changing direct dependencies, use `deploymentManifest()` from
+`scripts/package-langsmith.mts` to write a temporary standalone `package.json`, run
+`npm install --package-lock-only --ignore-scripts --workspaces=false` there, and copy
+its lock to `deployment-package-lock.json`. The build rejects stale direct dependency
+locks. Do not copy the monorepo lock or workspace dependencies into the artifact.
+The workspace lock keeps this app's dependency tree nested so its testing helpers
+and runtime resolve Dawn 0.8.24 while lifecycle retains Dawn 0.8.21.
+
+Set `GROWTH_RESEARCH_FIXTURE_MODE=synthetic-only` explicitly to permit model calls.
+The default blocks them. The fixed corpus contains `atlas` and `beacon`; tools accept
+only those identifiers and cannot fetch URLs, read arbitrary files or execute shell
+commands. The specialist is explicitly registered with delegation denied by default
+and only that specialist allowed. It can read fixtures but is denied the shared
+coordinator summary tool. Planning and skill instructions are authored beside the
+coordinator route.
+
+For a local active-child cancellation probe, the operator may set
+`GROWTH_RESEARCH_FIXTURE_DELAY_MS` to an integer from 0 to 5000. It defaults to zero;
+the model cannot choose a delay. The fixture tool cooperatively observes cancellation
+while paused and rechecks both cancellation and fixture mode before returning data.
+
+The public Dawn `seedModelImporter` bootstrap installs a process-wide bounded
+OpenAI model for this isolated app. Every request is gated, including cached models.
+It uses `gpt-4.1-mini`, a 1,024-token output cap, zero provider retries and a 20-second
+request timeout. Credential-free schema extraction can construct the model with a
+construction-only placeholder; invocation and actual HTTP fetch reject absent real
+credentials, so the placeholder is never sent. Route recursion is limited to 12 steps and Dawn retries to one
+attempt. These are compatibility-probe bounds, not a shared spending reservation or
+production provider selection. Provider-free tests inspect actual request bodies,
+verify one request on a retryable failure, and observe a stalled request timing out.
+
+Candidate memory uses an explicit lazy pgvector store via `DAWN_DATABASE_URL`, with
+8-dimensional deterministic synthetic embeddings. Generated `remember` writes are
+candidates and normal `recall` excludes them. Missing database configuration fails
+when durable memory is accessed; there is no SQLite fallback. The eager prompt index
+is explicitly disabled with `indexMaxEntries: 0`; a zero-result search returns an
+empty list without opening a connection. This allows credential-free graph import
+and packaging while positive-limit recall and all writes still require the database.
+This index setting is necessary because Dawn 0.8.24 does not consult `memory.enabled`
+when a route-local memory declaration exists.
+
+Run the separate, uncached integration target only against a disposable database:
+
+```sh
+GROWTH_RESEARCH_TEST_DATABASE_URL='postgres://…' npx nx run growth-research:test-memory-integration
+```
+
+The probe requires that variable and never falls back to a production URL. It uses
+fresh child processes for generated candidate writes, active recall, slot isolation
+and deletion, and deletes only the fixture record it created. Memory namespaces use
+the stable `growth-research` workspace and route plus a server-owned `GROWTH_RESEARCH_FIXTURE_SLOT` (`atlas` or
+`beacon`, default `atlas`). These are trusted synthetic deployment slots, not
+authenticated account identities. The explicit workspace remains stable across source,
+staging and relocated deployment directories, which the subprocess test verifies.
+Dawn's scope callback has no authenticated user;
+production tenancy still requires separate application-owned authorization. Synthetic
+hash embeddings do not establish semantic retrieval quality for live data.
+
+Build, staging, standalone installation, and native Node graph import require no
+model or database credentials. Native import is only a packaging check; run server
+and cloud smoke checks separately to exercise the deployment boundary. The server's static
+schema extractor still emits a nonfatal `Unsupported type: never` diagnostic; the
+tested runtime operations succeeded despite it. Fast tests use the public Dawn
+harness and a local mock model; memory persistence is verified separately against
+PostgreSQL.
+
+Run the fast and database suites sequentially: Dawn's local testing harness uses a
+shared checkpoint file, so overlapping those commands can produce a SQLite lock
+error. This does not change the deployed graph's LangSmith checkpoint ownership or
+its separate pgvector memory store.
+
+Agent Server `0.13.4-node24` can acknowledge interruption before its JavaScript child
+stops, allowing a later result checkpoint. The generated Dawn graph cancels when a
+live `config.signal` is supplied; the official JS sidecar does not forward that
+signal. No vendor patch is included. Cancellation and protection against writes
+after cancellation remain failed live-use gates. The smoke client's cleanup command
+refuses interrupted threads; an operator must independently establish worker
+quiescence before deleting those records. A terminal run status alone is insufficient.
+Deploy the verified artifact with the official CLI `0.4.21` source archive layout
+and the LangSmith control-plane source-upload API. Updates should target the existing deployment ID:
+request its upload URL, upload only the verified `.deployment` archive, and submit
+the returned object path with `revision_source: "internal_source"`,
+`langgraph_config_path: "langgraph.json"`, and `install_command: "npm ci --ignore-scripts"`.
+The signed upload requires `Content-Type: application/gzip` and
+`X-Goog-Content-Length-Range: 0,209715200`. Configure secrets through the deployment
+API; never include an environment file in the archive. Re-enabling synthetic model
+tests requires both a provider key and the explicit fixture-mode value. Do not wire
+real Growth signals into this deployment until its remaining live-use gates pass.
+
+The uncached platform smoke target takes positional fixture, thread and correlation
+identifiers. Set `GROWTH_RESEARCH_URL`, `LANGSMITH_API_KEY` when authentication is
+required, and the explicit fixture-mode gate in the operator environment:
+
+```sh
+npx nx run growth-research:smoke-langsmith -- direct THREAD_UUID SMOKE_ID
+```
+
+Other phases are `delegated`, `memory`, `continuation`, and `cleanup`. Continuation
+uses the same thread and smoke ID after a direct run; cleanup verifies ownership
+and rejects active or interrupted runs, then deletes the fixture thread and verifies
+absence. Interrupted fixtures require the separate operator procedure described above.
+
+This application is restricted to synthetic compatibility work. It does not collect
+real people or companies, publish account facts, or dispatch campaigns. Live use still
+requires trusted scopes, source controls, budget enforcement, a durable Growth work
+ledger, publication validation and cross-store deletion safeguards.

apps/growth-research/dawn.config.ts

@@ -0,0 +1,18 @@
+import type { DawnConfig } from '@dawn-ai/core';
+import './src/runtime/model-boundary.js';
+import { candidateMemoryStore, syntheticEmbedder, trustedFixtureScope } from './src/runtime/memory-store.js';
+
+export default {
+  appDir: 'src/app',
+  build: { targets: ['langsmith'] },
+  toolOutput: { noOffloadTools: ['readFixture', 'coordinatorSummary', 'readSkill', 'writeTodos', 'recall', 'remember'] },
+  summarization: { enabled: false },
+  memory: {
+    store: candidateMemoryStore,
+    indexMaxEntries: 0,
+    writes: 'candidate',
+    vector: { embedder: syntheticEmbedder },
+    resolveScope: trustedFixtureScope,
+    episodes: { enabled: false },
+  },
+} satisfies DawnConfig;