@panaversity/ksor 0.0.22 → 0.0.24

package/CHANGELOG.md

@@ -1,5 +1,76 @@
 # @panaversity/ksor
 
+## 0.0.24
+
+### Patch Changes
+
+- cbdc4c2: The MCP tool surface is now adopter-owned code. `ksor init` emits
+  `system/gateways/content.ts` — ordinary `registerTool` calls with ordinary zod —
+  where a record sets what its tools are called, what it says it covers, what they
+  accept, and which of them exist at all.
+
+  Agents are the operator, and an agent pays for this surface out of its context
+  window twice. Measured against a live 81-document record: the three tool
+  definitions cost ~2,990 tokens and stay resident for the whole session, and one
+  `search` at the default `k=10` costs ~3,541 tokens per call. Until now a record
+  could change none of it. Deleting the two tools your agents never call gives back
+  ~1,643 tokens a session — verified live at 5,337 bytes of definitions against the
+  default's 11,960.
+
+  Real code rather than a config API, because models are trained on the MCP SDK and
+  on zod and not on our field names — and because `registerTool` lets a record add
+  its own tools, which no config schema could have anticipated. One import
+  (`@panaversity/ksor/gateway`, which re-exports `z` and `McpServer`), so a
+  registration stays a file: no package, no build step, nothing new in your
+  lockfile. It is **deletable** — without it the door serves the identical surface.
+
+  The handlers, output schemas and framework description text stay in the package,
+  because a hand-written handler returning fabricated hits with plausible
+  `stable_id`s passes every shape check there is. Your prose composes ABOVE the
+  framework text, never instead of it — and since that is now a template literal in
+  a file you own, **the door inspects its own served surface at boot** and refuses
+  to start when a guarantee is gone: `ksor-gateway-floor-missing`,
+  `ksor-gateway-no-tools`, `ksor-gateway-unloadable`.
+
+  Adds a public subpath export, `@panaversity/ksor/gateway`.
+
+  New: `docs/tool-surface.md`.
+
+## 0.0.23
+
+### Patch Changes
+
+- 6d306a2: Ship the deployment artifacts: `ksor init` now emits a `Dockerfile` and
+  `.dockerignore`, and its `vercel.json` declares both surfaces — the static site
+  and the MCP door — behind one domain.
+
+  The served MCP rung is a core surface, not an optional extra, so a scaffolded
+  project should be able to reach a host without anyone hand-writing a container
+  recipe first. The emitted `Dockerfile` names no vendor: it installs the pinned
+  `@panaversity/ksor`, honours `$PORT`, and runs `ksor serve`, so the same image
+  runs on Cloud Run, Fly, Render, ECS, Kubernetes or a VPS. `vercel.json` points
+  AT that file rather than replacing it, which is what keeps the host a choice —
+  moving is a redeploy, not a rewrite. A test asserts that neutrality directly,
+  and CI now builds the emitted image, boots it against real Postgres and asks it
+  a question over MCP, with no hosting vendor involved.
+
+  Verified live before shipping, and the verification paid for itself: a
+  project-level `trailingSlash: true` — harmless while the config was static-only
+  — 308-redirected every door route including `POST /mcp`. It is removed (the
+  site's own Next config already sets it where it belongs); shipping it would have
+  broken the MCP endpoint of every adopter who deployed.
+
+  Two new documents: `docs/deploying.md` (both surfaces onto a host, the
+  configuration each needs, and what a cold start costs — measured) and
+  `docs/ingesting.md` (why serving does not publish, so a first deploy with no
+  ingest serves an empty record; where ingest belongs, which is never inside the
+  container; and how the abstention gate gets turned on).
+
+  Also drops the scaffold's build-script denials for `@google/genai` and
+  `protobufjs`. The embedding provider speaks the vendor's REST API directly now,
+  so neither package is installed at all and the entries described a dependency
+  that no longer exists.
+
 ## 0.0.22
 
 ### Patch Changes