diff --git a/.changeset/antfly-lifecycle-resilience.md b/.changeset/antfly-lifecycle-resilience.md
deleted file mode 100644
index 4aeb5e2..0000000
--- a/.changeset/antfly-lifecycle-resilience.md
+++ /dev/null
@@ -1,12 +0,0 @@
----
-'@prosdevlab/dev-agent': minor
----
-
-Antfly lifecycle resilience: graceful MCP startup, `dev doctor`, Podman support, captured startup logs
-
-- **MCP server no longer dies when Antfly is down.** The stdio handshake completes immediately; backend init (Antfly auto-start, index load, watcher catchup) runs behind a gate. Tool calls wait for it, and a failed init returns a legible error pointing at `dev doctor` — with automatic retry on the next call once the backend is reachable.
-- **New `dev doctor` command** diagnoses the stack from the CLI (install, server reachability, port conflicts, embedding model, repository index) — works when the MCP server itself can't start.
-- **Podman support.** All container fallbacks detect Docker or Podman instead of assuming Docker.
-- **Antfly startup output is captured** to `~/.antfly/antfly.log` (rotated at 5MB) instead of being discarded, so failed starts are diagnosable.
-- **`dev search`, `dev refs`, and `dev map` auto-start the backend** like `dev index` already did.
-- One shared lifecycle implementation in core replaces three divergent copies (CLI utils, MCP entry point, adapter registry) — the MCP copies were missing `--data-dir` and port-conflict detection.
diff --git a/packages/dev-agent/CHANGELOG.md b/packages/dev-agent/CHANGELOG.md
index 2135b21..ee8d6ce 100644
--- a/packages/dev-agent/CHANGELOG.md
+++ b/packages/dev-agent/CHANGELOG.md
@@ -1,5 +1,18 @@
 # @prosdevlab/dev-agent
 
+## 0.13.0
+
+### Minor Changes
+
+- a6d08da: Antfly lifecycle resilience: graceful MCP startup, `dev doctor`, Podman support, captured startup logs
+
+  - **MCP server no longer dies when Antfly is down.** The stdio handshake completes immediately; backend init (Antfly auto-start, index load, watcher catchup) runs behind a gate. Tool calls wait for it, and a failed init returns a legible error pointing at `dev doctor` — with automatic retry on the next call once the backend is reachable.
+  - **New `dev doctor` command** diagnoses the stack from the CLI (install, server reachability, port conflicts, embedding model, repository index) — works when the MCP server itself can't start.
+  - **Podman support.** All container fallbacks detect Docker or Podman instead of assuming Docker.
+  - **Antfly startup output is captured** to `~/.antfly/antfly.log` (rotated at 5MB) instead of being discarded, so failed starts are diagnosable.
+  - **`dev search`, `dev refs`, and `dev map` auto-start the backend** like `dev index` already did.
+  - One shared lifecycle implementation in core replaces three divergent copies (CLI utils, MCP entry point, adapter registry) — the MCP copies were missing `--data-dir` and port-conflict detection.
+
 ## 0.12.2
 
 ### Patch Changes
diff --git a/packages/dev-agent/package.json b/packages/dev-agent/package.json
index 2df1864..3408a58 100644
--- a/packages/dev-agent/package.json
+++ b/packages/dev-agent/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@prosdevlab/dev-agent",
-  "version": "0.12.2",
+  "version": "0.13.0",
   "private": false,
   "description": "Deep code intelligence + AI subagents via MCP. Local-first semantic code search, GitHub integration, and development planning for AI tools like Cursor and Claude Code.",
   "keywords": [