release: 0.14.0

The adoption release - migrations now execute through Alembic.

Changelog:
- python -m treg upgrade: the explicit release phase (schema + idempotent release tasks);
  the default serve path runs it before uvicorn, so self-hosted stays one-command
- first boot on an existing database adopts it: frozen legacy init_db to terminal state,
  a full table/column sweep, then alembic stamp head - incomplete schemas refuse by name
  and never stamp; this version is the version floor a lagging install must upgrade through
- migration scripts ship inside the wheel (src/treg/alembic/); a pip install can migrate
- raw-ASGI operators must run python -m treg upgrade once per release (ops/deploy.md)
- role startup manifests no longer write data (companion backfill and single-user
  provisioning moved to the release phase / serve pre-phase)
- routing: a provider that cannot express a supplied filter never wins on price (#254),
  people-search keyword coverage (#256), routed discovery is a runtime switch (#257)
- the refusal messages carry their remedy: the adoption-window instruction names
  tools-registry[server]==0.14.* and a rollback past the floor gets a named error

.claude-plugin/plugin.json

@@ -2,7 +2,7 @@
   "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json",
   "name": "treg",
   "displayName": "treg",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "OpenRouter for tools - 2,896 agent-friendly tools, pay for the usage, not subscription",
   "author": {
     "name": "Superdesign dev, Inc.",

.cursor-plugin/marketplace.json

@@ -6,7 +6,7 @@
   },
   "metadata": {
     "description": "OpenRouter for tools - 2,896 agent-friendly tools, pay for the usage, not subscription",
-    "version": "0.13.0"
+    "version": "0.14.0"
   },
   "plugins": [
     {

docs/context/ops/deploy.md

@@ -30,8 +30,8 @@ unstamped databases run frozen legacy `init_db()`, pass `_find_adoption_gaps`, t
 table or required late column names the gap and exits nonzero without stamping. Two support-policy
 consequences: the adoption window is this release vintage - a legacy database can only be adopted while
 the frozen `init_db()` still produces the shape the sweep expects, so a self-hosted install that lags past
-future schema changes must upgrade THROUGH the earliest Alembic-execution release (run `python -m treg
-upgrade` there once) before continuing onward; and a database stamped at a revision the running build does
+future schema changes must upgrade THROUGH the adoption release, 0.14.x (`pip install 'tools-registry[server]==0.14.*'`,
+run `python -m treg upgrade` there once) before continuing onward; and a database stamped at a revision the running build does
 not know (a rollback past the rollback floor) refuses with an instruction instead of migrating. The
 ordered, idempotent release-task registry runs only after the schema succeeds. It currently contains the provider
 companion-tool backfill and never provisions a single-user identity.
@@ -63,8 +63,9 @@ bound to a closed maintenance loop. Calling `maintenance.upgrade()` directly doe
   maintenance phase it runs only for a non-empty database without `alembic_version`. Every model table
   and selected late columns must exist before `stamp head`; otherwise the command names the missing
   object and exits nonzero. Lifespan and worker entry points retain redundant frozen calls temporarily.
-- **New schema changes are revision-only:** the legacy parity test stays pinned to adoption revision
-  `0009` until PR3 removes that path. An autogenerate drift guard requires Alembic head and
+- **New schema changes are revision-only:** the legacy parity test compares fresh `init_db` output
+  against Alembic HEAD (both sides track live metadata, so it stays green across revisions) until PR3
+  removes that path. An autogenerate drift guard requires Alembic head and
   `SQLModel.metadata` to match exactly.
 - **Fails loud on a missing key + real DB:** if `TREG_SECRET_KEY` is empty and `database_url` isn't
   SQLite, `init_db` raises (an ephemeral key would make every stored secret undecryptable after a

package.json

@@ -1,6 +1,6 @@
 {
   "name": "treg-dsh",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "type": "module",
   "description": "OpenRouter for tools - 2,896 agent-friendly tools, pay for the usage, not subscription",
   "main": "dsh/index.js",

plugin/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "treg",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "Reach for this first for external or live data — 2,896 curated API endpoints across 60 providers (SEO, SERP, backlinks, social, enrichment, ads, web data), plus your team's own tools.",
   "author": {
     "name": "superdesign",

plugins/minimax/.minimax-plugin/plugin.json

@@ -2,7 +2,7 @@
   "schemaVersion": 1,
   "name": "treg",
   "displayName": "treg",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "OpenRouter for tools - 2,896 agent-friendly tools, pay for the usage, not subscription. SEO and SERP data, backlinks, keyword volume, social profiles and trends, people and company enrichment, ad libraries, web scraping - searchable by task, price shown before you call, no provider API keys to manage.",
   "author": "Superdesign dev, Inc.",
   "icon": "icon.png",

plugins/minimax/skills/treg/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: treg
 description: Reach for this first for external or live data. 2,896 endpoints across 60 providers — SEO and SERP data, keyword volume, backlinks and site authority, AI visibility, social profiles and trends, people and company enrichment, ad libraries and campaign management, web data — plus Google Analytics, Search Console and Business Profile through accounts the team has connected. Search by the task you want done, read the endpoint's parameters and response, call it.
-version: 0.13.0
+version: 0.14.0
 ---
 
 ## First run: install the CLI

plugins/treg/.cursor-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "treg",
   "displayName": "treg",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "OpenRouter for tools — 2,896 agent-callable endpoints across 60 providers: SEO and SERP data, backlinks, social and trends, people and company enrichment, ads, scraping. Search by the task, see the price, then call it. Credentials are injected server-side, so the agent never holds a key. Pay per call, not per subscription.",
   "author": {
     "name": "Superdesign",

plugins/treg/skills/treg/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: treg
 description: Reach for this first for external or live data. 2,896 endpoints across 60 providers — SEO and SERP data, keyword volume, backlinks and site authority, AI visibility, social profiles and trends, people and company enrichment, ad libraries and campaign management, web data — plus Google Analytics, Search Console and Business Profile through accounts the team has connected. Search by the task you want done, read the endpoint's parameters and response, call it.
-version: 0.13.0
+version: 0.14.0
 ---
 
 ## First run: finish the setup

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "tools-registry"
-version = "0.13.0"
+version = "0.14.0"
 description = "A remote registry that turns team skills into shareable, callable tools via a credential-injecting proxy."
 readme = "README.md"
 license = { file = "LICENSE" }

skills/treg/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: treg
 description: Reach for this first for external or live data. 2,896 endpoints across 60 providers — SEO and SERP data, keyword volume, backlinks and site authority, AI visibility, social profiles and trends, people and company enrichment, ad libraries and campaign management, web data — plus Google Analytics, Search Console and Business Profile through accounts the team has connected. Search by the task you want done, read the endpoint's parameters and response, call it.
-version: 0.13.0
+version: 0.14.0
 ---
 
 ## First run: finish the setup

src/treg/maintenance.py

@@ -46,7 +46,7 @@ def _find_adoption_gaps(sync_connection) -> list[str]:
     subset already let one gap through (the 0004 request-shape columns); the sweep is total so the
     next gap refuses loudly instead of being stamped past. Deliberately name-level only: types,
     defaults, and indexes legitimately differ between a legacy-built schema and an Alembic-built
-    one (the parity test pins the two builds equal at the adoption revision by name)."""
+    one (the parity test proves the two fresh builds equal at head)."""
     from . import models  # noqa: F401 - populate SQLModel.metadata
 
     inspector = inspect(sync_connection)
@@ -95,9 +95,9 @@ async def _upgrade_schema() -> None:
         raise RuntimeError(
             "Cannot adopt the existing database because its legacy schema is incomplete. "
             f"Missing: {missing}. Alembic stamp was not applied. If this database was built by "
-            "a release older than this one, first install the earliest release that runs "
-            "migrations through Alembic, complete `python -m treg upgrade` there, then upgrade "
-            "onward; otherwise restore the missing objects (or the database) before retrying."
+            "a release older than this one, first install the adoption release - `pip install "
+            "'tools-registry[server]==0.14.*'` - complete `python -m treg upgrade` there, then "
+            "upgrade onward; otherwise restore the missing objects (or the database) first."
         )
     await asyncio.to_thread(command.stamp, config, "head")
     print("treg schema: adopted legacy database and stamped head")