diff --git a/docs/adr/0017-agent-session-is-provider-bound.md b/docs/adr/0017-agent-session-is-provider-bound.md
index 37af679..331cf3a 100644
--- a/docs/adr/0017-agent-session-is-provider-bound.md
+++ b/docs/adr/0017-agent-session-is-provider-bound.md
@@ -1,50 +1,53 @@
-# ADR 0017: AgentSession Is Provider-Bound
+# ADR 0017: Agent Runtime — Claude Code SDK
 
 ## Status
 
-Accepted.
+Accepted — supersedes the original ADR-0017 (provider-agnostic session).
 
 ## Context
 
-ADR-0002 says a long-lived `AgentSession` can be reused by many runs, but never
-addressed the provider/model dimension. The agent layer is provider-agnostic
-(ADR-0001..0003 never required Claude specifically; the `@Claude` trigger name
-is a product brand, not a provider commitment). In practice the Hub routes
-model calls through an OpenAI-compatible client (e.g. OpenRouter), so a run may
-target different models — and the motivation for multi-model is **role-based
-routing**: drafting, reviewing, and triage suit different models.
+The original ADR-0017 mandated a provider-agnostic agent layer: any
+OpenAI-compatible model via OpenRouter, custom agent loop built on Vercel AI
+SDK's `streamText`. The motivation was to avoid vendor lock-in.
 
-That raises the question ADR-0002 left open: when a teacher switches model
-mid-project, does the `AgentSession` carry live context across the switch, or
-is it bound to a single provider/model?
+In practice, the hand-rolled agent loop lacked capabilities that a production
+agent needs: context compaction (long conversations), tool approval flows,
+partial-message streaming, and a battle-tested multi-turn loop. Building these
+ourselves meant re-implementing what the Claude Code SDK already provides.
+
+The Feishu integration also proved costly to hand-roll — card schema quirks,
+patch limitations, file support — each API quirk cost a round-trip. The
+Claude-to-IM library (2.7K stars) has already solved these, but its
+LLMProvider interface expects Claude Code SDK's SSE event format. Using it with
+a generic provider would require a conversion layer.
 
 ## Decision
 
-An `AgentSession` is **provider/model-bound**. A model or provider switch
-starts a new `AgentSession`. The `AgentSession` does not carry live context
-across a switch.
+Adopt `@anthropic-ai/claude-agent-sdk` as the agent runtime. The SDK's `query()`
+function owns the agent loop (tool dispatch, compaction, streaming). Built-in
+tools (Read, Write, Bash, Glob, Grep) cover the entire tool surface — no custom
+tools needed. `cph check` / `cph build` run via Bash.
 
-Cross-session continuity is not the session's job — it is carried by ADR-0003's
-project memory and anchors, which the Hub reads on demand when seeding a new
-run. This keeps B consistent with ADR-0003 ("the Hub avoids becoming a full
-chat history store"): session continuity and history continuity are the same
-mechanism, both via memory/anchors, never via a live cross-provider session.
-
-Per-run model selection (the run carries its model) is the switching point.
+This means:
+- The agent layer is **Anthropic-specific**, not provider-agnostic. Claude Code
+  SDK speaks the Anthropic Messages API and cannot be pointed at OpenRouter
+  (which is OpenAI-compatible).
+- An Anthropic API key (or OAuth via Claude Pro/Max) is required.
+- Model selection is within the Anthropic model family (Sonnet, Haiku, Opus),
+  not across vendors.
 
 ## Consequences
 
-- Switching model mid-project = new session; the new run seeds from project
-  memory/anchors (ADR-0003), not the prior session's live context.
-- The agent layer is provider-agnostic: model calls go through an
-  OpenAI-compatible client, and the tool-calling loop is delegated to the
-  Vercel AI SDK (`generateText` + `tool`). The provider seam is the SDK's
-  `LanguageModel` interface, not a hand-rolled loop; swapping provider is
-  swapping the `createOpenAICompatible` factory, not rewriting the loop. No
-  hard dependency on a single-vendor agent SDK.
-- Role-based model routing (different run kinds default to different models) is
-  a product/admin configuration concern, not a spec invariant.
-- "AgentSession reused by many runs" (ADR-0002) holds *within* one
-  provider/model; it does not span a switch.
-- The `@Claude` trigger name remains the product mention brand regardless of the
-  backing model.
+- We lose the ability to route to GLM or other non-Anthropic models via
+  OpenRouter. This is the accepted cost of getting a mature agent loop.
+- The Vercel AI SDK dependency is dropped.
+- Custom tools (read_file, write_file, cph_check, cph_build) are replaced by
+  the SDK's built-in Read/Write/Bash/Glob/Grep.
+- The SSE event format matches Claude-to-IM's expected protocol natively —
+  adopting Claude-to-IM for Feishu integration is a direct fit, no conversion.
+- Per-run model selection still works (role→model mapping), but models are
+  Anthropic aliases, not OpenRouter routing IDs.
+- If provider-agnosticism becomes a hard requirement again, the migration path
+  is: replace `query()` with a Vercel AI SDK `streamText` call + an SSE format
+  converter (~100 lines). The rest of the architecture (trigger, lock,
+  permission, Feishu client) is provider-agnostic already.