chore: version packages

.doc-bridge/capabilities.json

@@ -7,7 +7,7 @@
     "name": "AgentsKit",
     "root": "."
   },
-  "contentHash": "e30c41afd86eef2cbd264342d8925c4531db6df699a1bccf023998fa4e496bf9",
+  "contentHash": "dc9fe5e9beded7091d00c6a2e99f0629bd2f706d91a5fad443e20ef5a4e42e24",
   "artifacts": {
     "index": ".doc-bridge/index.json",
     "llmsTxt": ".doc-bridge/llms.txt"

apps/docs-next/content/docs/for-agents/mcp.mdx

@@ -56,7 +56,7 @@ tools bridge. A thin bridge over `@agentskit/tools/mcp` (`createMcpServer` +
 ## Install
 
 ```bash
-npm install @agentskit/mcp@0.5.0
+npm install @agentskit/mcp@0.5.1
 ```
 
 ## Primary exports
@@ -100,7 +100,7 @@ other features from newer protocol revisions.
 
 ## Bin
 
-`agentskit-mcp` — `npx -y @agentskit/mcp@0.5.0 --tools fetch,search`. Expose whole agents
+`agentskit-mcp` — `npx -y @agentskit/mcp@0.5.1 --tools fetch,search`. Expose whole agents
 with `--agents <ids> --provider <p>` (any of ~20 providers + catalog OpenAI-compatible). Flags: `--tools`,
 `--fs-root`, `--sqlite`, `--allow-shell`, `--model`, `--max-steps`, `--api-key`,
 `--base-url`. Unknown, duplicate, or valueless flags fail closed. stdout is the

apps/docs-next/content/docs/reference/packages/adapters.mdx

@@ -136,7 +136,7 @@ Testing: `mockAdapter` · `recordingAdapter` · `replayAdapter` · `inMemorySink
 
 ## Stability
 
-- **Version:** `0.19.0`
+- **Version:** `0.19.1`
 - **Tier:** beta
 - **Contract:** evolving
 - **Roadmap:** see [packages roadmap](./roadmap) for what this package needs to reach v1.0.

apps/docs-next/content/docs/reference/packages/cli.mdx

@@ -134,7 +134,7 @@ The CLI makes AgentsKit feel like a product, not just a pile of packages. It sho
 
 ## Stability
 
-- **Version:** `0.13.47`
+- **Version:** `0.13.48`
 - **Tier:** beta
 - **Contract:** evolving
 - **Roadmap:** see [packages roadmap](./roadmap) for what this package needs to reach v1.0.

apps/docs-next/content/docs/reference/packages/memory.mdx

@@ -80,7 +80,7 @@ In most real products, memory is the difference between a clever demo and a usef
 
 ## Stability
 
-- **Version:** `0.12.0`
+- **Version:** `0.12.1`
 - **Tier:** beta
 - **Contract:** evolving
 - **Roadmap:** see [packages roadmap](./roadmap) for what this package needs to reach v1.0.

apps/docs-next/content/docs/reference/packages/net.mdx

@@ -82,7 +82,7 @@ await retry(
 
 ## Stability
 
-- **Version:** `0.2.3`
+- **Version:** `0.2.4`
 - **Tier:** beta
 - **Contract:** evolving
 

apps/docs-next/content/docs/reference/packages/observability-langfuse.mdx

@@ -23,7 +23,7 @@ docbridge:
       detection: static
 ---
 
-- **Version:** `0.2.27`
+- **Version:** `0.2.28`
 
 `@agentskit/observability/langfuse` is a drop-in observer that ships AgentsKit trace spans to [Langfuse](https://langfuse.com). LLM generations become `generation` nodes; tool calls and planning steps become `span` nodes. Token usage and error state are captured on each node end. Cost is not normalized by this adapter, and HITL is not represented until the canonical event contract includes it.
 

apps/docs-next/content/docs/reference/packages/observability.mdx

@@ -82,7 +82,7 @@ For many teams, this is the point where an agent stops feeling magical and start
 
 ## Stability
 
-- **Version:** `0.13.0`
+- **Version:** `0.13.1`
 - **Tier:** beta
 - **Contract:** evolving
 - **Roadmap:** see [packages roadmap](./roadmap) for what this package needs to reach v1.0.

apps/docs-next/content/docs/reference/packages/rag.mdx

@@ -75,7 +75,7 @@ HTTPS origin in `allowedOrigins` and rejects redirects.
 
 ## Stability
 
-- **Version:** `0.7.0`
+- **Version:** `0.7.1`
 - **Tier:** beta
 - **Contract:** evolving
 - **Roadmap:** see [packages roadmap](./roadmap) for what this package needs to reach v1.0.

apps/docs-next/content/docs/reference/packages/runtime.mdx

@@ -84,7 +84,7 @@ That one `run()` call is the center of the agent stack. Everything else in the e
 
 ## Stability
 
-- **Version:** `0.11.2`
+- **Version:** `0.11.3`
 - **Tier:** beta
 - **Contract:** evolving
 - **Roadmap:** see [packages roadmap](./roadmap) for what this package needs to reach v1.0.

apps/docs-next/content/docs/reference/packages/tools.mdx

@@ -103,7 +103,7 @@ default byte cap (`maxResponseBytes`) and cancels the stream on overflow.
 
 ## Stability
 
-- **Version:** `0.13.15`
+- **Version:** `0.13.16`
 - **Tier:** beta
 - **Contract:** evolving
 - **Roadmap:** see [packages roadmap](./roadmap) for what this package needs to reach v1.0.

apps/docs-next/content/docs/reference/recipes/coding-agent-mcp.mdx

@@ -35,7 +35,7 @@ host application's UI, login, or trust prompt.
 Add the server with the CLI:
 
 ```bash
-codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search
+codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search
 codex mcp list
 ```
 
@@ -44,7 +44,7 @@ Or commit a project-scoped `.codex/config.toml` in a trusted project:
 ```toml
 [mcp_servers.agentskit]
 command = "npx"
-args = ["-y", "@agentskit/mcp@0.5.0", "--tools", "fetch,search"]
+args = ["-y", "@agentskit/mcp@0.5.1", "--tools", "fetch,search"]
 default_tools_approval_mode = "prompt"
 ```
 
@@ -56,7 +56,7 @@ Use `/mcp` to inspect the connected server.
 Create a project-scoped server:
 
 ```bash
-claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search
+claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search
 claude mcp list
 ```
 
@@ -68,7 +68,7 @@ The equivalent `.mcp.json` is:
     "agentskit": {
       "type": "stdio",
       "command": "npx",
-      "args": ["-y", "@agentskit/mcp@0.5.0", "--tools", "fetch,search"]
+      "args": ["-y", "@agentskit/mcp@0.5.1", "--tools", "fetch,search"]
     }
   }
 }
@@ -89,7 +89,7 @@ Create `.cursor/mcp.json`:
     "agentskit": {
       "type": "stdio",
       "command": "npx",
-      "args": ["-y", "@agentskit/mcp@0.5.0", "--tools", "fetch,search"]
+      "args": ["-y", "@agentskit/mcp@0.5.1", "--tools", "fetch,search"]
     }
   }
 }
@@ -109,7 +109,7 @@ edits `%APPDATA%\Claude\claude_desktop_config.json`.
   "mcpServers": {
     "agentskit": {
       "command": "npx",
-      "args": ["-y", "@agentskit/mcp@0.5.0", "--tools", "fetch,search"]
+      "args": ["-y", "@agentskit/mcp@0.5.1", "--tools", "fetch,search"]
     }
   }
 }
@@ -128,7 +128,7 @@ Open MCP Servers, choose Configure MCP Servers, and add:
   "mcpServers": {
     "agentskit": {
       "command": "npx",
-      "args": ["-y", "@agentskit/mcp@0.5.0", "--tools", "fetch,search"],
+      "args": ["-y", "@agentskit/mcp@0.5.1", "--tools", "fetch,search"],
       "env": {},
       "disabled": false,
       "autoApprove": []
@@ -151,15 +151,15 @@ Create `.continue/mcpServers/agentskit.yaml`:
 
 ```yaml
 name: AgentsKit MCP
-version: 0.5.0
+version: 0.5.1
 schema: v1
 mcpServers:
   - name: agentskit
     type: stdio
     command: npx
     args:
       - "-y"
-      - "@agentskit/mcp@0.5.0"
+      - "@agentskit/mcp@0.5.1"
       - "--tools"
       - "fetch,search"
 ```
@@ -174,7 +174,7 @@ server arguments rather than changing host-specific code:
 
 ```bash
 OPENROUTER_API_KEY=your-key \
-  npx -y @agentskit/mcp@0.5.0 \
+  npx -y @agentskit/mcp@0.5.1 \
   --tools fetch,search \
   --agents code-review \
   --provider openrouter \

apps/docs-next/content/docs/reference/recipes/mcp-bridge.mdx

@@ -55,7 +55,7 @@ For a ready-made stdio server, install `@agentskit/mcp` instead of writing a
 transport wrapper:
 
 ```bash
-npx -y @agentskit/mcp@0.5.0 --tools fetch,search
+npx -y @agentskit/mcp@0.5.1 --tools fetch,search
 ```
 
 Use the smallest tool set that fits the task. `filesystem`, `sqlite`, and
@@ -71,7 +71,7 @@ Use this JSON in Claude Desktop or `.cursor/mcp.json` in a project:
   "mcpServers": {
     "agentskit": {
       "command": "npx",
-      "args": ["-y", "@agentskit/mcp@0.5.0", "--tools", "fetch,search"]
+      "args": ["-y", "@agentskit/mcp@0.5.1", "--tools", "fetch,search"]
     }
   }
 }
@@ -84,7 +84,7 @@ Codex uses TOML configuration, not the JSON wrapper above:
 ```toml
 [mcp_servers.agentskit]
 command = "npx"
-args = ["-y", "@agentskit/mcp@0.5.0", "--tools", "fetch,search"]
+args = ["-y", "@agentskit/mcp@0.5.1", "--tools", "fetch,search"]
 ```
 
 ### OpenClaw
@@ -95,7 +95,7 @@ OpenClaw can save and probe the stdio server directly:
 openclaw mcp add agentskit \
   --command npx \
   --arg -y \
-  --arg @agentskit/mcp@0.5.0 \
+  --arg @agentskit/mcp@0.5.1 \
   --arg --tools \
   --arg fetch,search
 openclaw mcp doctor agentskit --probe

docs/ecosystem/content-pipeline/atoms/coding-agent-mcp/atom.json

@@ -43,8 +43,8 @@
       "expectStdoutIncludes": "verified local MCP stdio protocol; cli tools: fetch_url, web_search"
     },
     "commands": [
-      "codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search",
-      "claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search",
+      "codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search",
+      "claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search",
       "pnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts",
       "pnpm --filter @agentskit/mcp smoke:published"
     ],
@@ -173,22 +173,22 @@
     }
   ],
   "variants": {
-    "docsPage": "---\ntitle: Connect AgentsKit to your coding agent\ndescription: Developers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n---\n\n# Connect AgentsKit to your coding agent\n\nDevelopers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n\n## Audience\n\nJavaScript and TypeScript developers using Codex, Claude Code or Desktop, Cursor, Cline, Continue, or another MCP-compatible host\n\n## Commands\n\n```bash\ncodex mcp add agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search\nclaude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search\npnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts\npnpm --filter @agentskit/mcp smoke:published\n```\n\n## Verified claims\n\n- **agentskit/packages**: 22 packages (from ecosystem-claims.json)\n- **agentskit/integrations**: 50 integrations (from ecosystem-claims.json)\n\n## Citations\n\n- [coding-agent MCP host fixture](packages/mcp/fixtures/coding-agent-hosts.ts)\n- [coding-agent MCP recipe](/docs/reference/recipes/coding-agent-mcp)\n- [Codex MCP configuration](https://developers.openai.com/codex/mcp/)\n- [Claude Code MCP configuration](https://docs.anthropic.com/en/docs/claude-code/mcp)\n- [Cursor MCP configuration](https://cursor.com/docs/mcp)\n- [Cline MCP configuration](https://docs.cline.bot/mcp/mcp-overview)\n- [Continue MCP configuration](https://docs.continue.dev/customize/deep-dives/mcp)\n\n## Ecosystem next steps\n\n- [AgentsKit](https://www.agentskit.io/docs) — Build agents without gluing many libraries together.\n- [AgentsKit Registry](https://registry.agentskit.io/docs) — Copy ready-made agents and own the source.\n- [AgentsKit Chat](https://chat.agentskit.io/docs) — Define one agent experience and deliver it across interfaces.\n- [Doc Bridge](https://doc-bridge.agentskit.io/) — Turn repository documentation into executable agent handoffs.\n- [AgentsKit Code Review](https://code-review.agentskit.io/docs) — Run deep, low-noise review with the model already in use.\n- [AgentsKit Harness](https://harness.agentskit.io/docs) — The keep-pushing loop for your SDLC.\n- [Agents Playbook](https://playbook.agentskit.io/docs) — Make agents ship code a human would actually merge.\n\n> Draft generated by the content pipeline. Human approval required before publishing.\n",
+    "docsPage": "---\ntitle: Connect AgentsKit to your coding agent\ndescription: Developers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n---\n\n# Connect AgentsKit to your coding agent\n\nDevelopers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n\n## Audience\n\nJavaScript and TypeScript developers using Codex, Claude Code or Desktop, Cursor, Cline, Continue, or another MCP-compatible host\n\n## Commands\n\n```bash\ncodex mcp add agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search\nclaude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search\npnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts\npnpm --filter @agentskit/mcp smoke:published\n```\n\n## Verified claims\n\n- **agentskit/packages**: 22 packages (from ecosystem-claims.json)\n- **agentskit/integrations**: 50 integrations (from ecosystem-claims.json)\n\n## Citations\n\n- [coding-agent MCP host fixture](packages/mcp/fixtures/coding-agent-hosts.ts)\n- [coding-agent MCP recipe](/docs/reference/recipes/coding-agent-mcp)\n- [Codex MCP configuration](https://developers.openai.com/codex/mcp/)\n- [Claude Code MCP configuration](https://docs.anthropic.com/en/docs/claude-code/mcp)\n- [Cursor MCP configuration](https://cursor.com/docs/mcp)\n- [Cline MCP configuration](https://docs.cline.bot/mcp/mcp-overview)\n- [Continue MCP configuration](https://docs.continue.dev/customize/deep-dives/mcp)\n\n## Ecosystem next steps\n\n- [AgentsKit](https://www.agentskit.io/docs) — Build agents without gluing many libraries together.\n- [AgentsKit Registry](https://registry.agentskit.io/docs) — Copy ready-made agents and own the source.\n- [AgentsKit Chat](https://chat.agentskit.io/docs) — Define one agent experience and deliver it across interfaces.\n- [Doc Bridge](https://doc-bridge.agentskit.io/) — Turn repository documentation into executable agent handoffs.\n- [AgentsKit Code Review](https://code-review.agentskit.io/docs) — Run deep, low-noise review with the model already in use.\n- [AgentsKit Harness](https://harness.agentskit.io/docs) — The keep-pushing loop for your SDLC.\n- [Agents Playbook](https://playbook.agentskit.io/docs) — Make agents ship code a human would actually merge.\n\n> Draft generated by the content pipeline. Human approval required before publishing.\n",
     "shortPost": "Connect AgentsKit to your coding agent.\n\n4 commands. No provider key.\n\n```bash\npnpm --filter @agentskit/mcp smoke:published\n```\n\nBuilt on AgentsKit (22 packages, 50 integrations).\n\nDocs + fixture linked in the atom. Human-reviewed drafts only.",
-    "thread": "1/ Connect AgentsKit to your coding agent\n\nWhy it exists: Developers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n\n2/ The path:\n1. `codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search`\n2. `claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search`\n3. `pnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts`\n4. `pnpm --filter @agentskit/mcp smoke:published`\n\n3/ Claims (generated, not hand-typed):\n- 22 packages\n- 50 integrations\n\n4/ Next products:\n- AgentsKit: Build agents without gluing many libraries together.\n- AgentsKit Registry: Copy ready-made agents and own the source.\n- AgentsKit Chat: Define one agent experience and deliver it across interfaces.\n\n5/ This thread is a draft atom. Publishing requires explicit human approval.",
-    "communityPost": "## Connect AgentsKit to your coding agent\n\nDevelopers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n\n### Try it\n\n```bash\ncodex mcp add agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search\nclaude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search\npnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts\npnpm --filter @agentskit/mcp smoke:published\n```\n\n### Claims\n22 packages, 50 integrations\n\nSource fixture: `packages/mcp/fixtures/run-coding-agent-hosts.mjs`\n\nStatus: draft — awaiting human approval before community publish.",
+    "thread": "1/ Connect AgentsKit to your coding agent\n\nWhy it exists: Developers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n\n2/ The path:\n1. `codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search`\n2. `claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search`\n3. `pnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts`\n4. `pnpm --filter @agentskit/mcp smoke:published`\n\n3/ Claims (generated, not hand-typed):\n- 22 packages\n- 50 integrations\n\n4/ Next products:\n- AgentsKit: Build agents without gluing many libraries together.\n- AgentsKit Registry: Copy ready-made agents and own the source.\n- AgentsKit Chat: Define one agent experience and deliver it across interfaces.\n\n5/ This thread is a draft atom. Publishing requires explicit human approval.",
+    "communityPost": "## Connect AgentsKit to your coding agent\n\nDevelopers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\n\n### Try it\n\n```bash\ncodex mcp add agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search\nclaude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search\npnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts\npnpm --filter @agentskit/mcp smoke:published\n```\n\n### Claims\n22 packages, 50 integrations\n\nSource fixture: `packages/mcp/fixtures/run-coding-agent-hosts.mjs`\n\nStatus: draft — awaiting human approval before community publish.",
     "example": {
       "kind": "fixture-link",
       "path": "packages/mcp/fixtures/run-coding-agent-hosts.mjs",
       "commands": [
-        "codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search",
-        "claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search",
+        "codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search",
+        "claude mcp add --scope project --transport stdio agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search",
         "pnpm --filter @agentskit/mcp exec vitest run tests/coding-agent-hosts.test.ts",
         "pnpm --filter @agentskit/mcp smoke:published"
       ]
     }
   },
-  "storyboard": "# Storyboard — Connect AgentsKit to your coding agent\n\nFormat: 5-frame carousel / 15s silent GIF\n\n1. **Hook** — \"Developers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\"\n2. **Install** — show `codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.0 --tools fetch,search`\n3. **Copy fixture** — show the agent.ts path\n4. **Run** — terminal output includes the expected ready line\n5. **Next step** — point to Registry / Playbook / Chat without inventing counts\n\nAccessibility: every frame needs alt text; prefer reduced-motion static frames.\nNo auto-publish — storyboard is a draft for human design review.\n",
+  "storyboard": "# Storyboard — Connect AgentsKit to your coding agent\n\nFormat: 5-frame carousel / 15s silent GIF\n\n1. **Hook** — \"Developers should be able to use the same AgentsKit tools and agents inside their existing coding agent without building or maintaining a separate integration for every host.\"\n2. **Install** — show `codex mcp add agentskit -- npx -y @agentskit/mcp@0.5.1 --tools fetch,search`\n3. **Copy fixture** — show the agent.ts path\n4. **Run** — terminal output includes the expected ready line\n5. **Next step** — point to Registry / Playbook / Chat without inventing counts\n\nAccessibility: every frame needs alt text; prefer reduced-motion static frames.\nNo auto-publish — storyboard is a draft for human design review.\n",
   "publish": {
     "status": "blocked",
     "reason": "Human approval required before publish packaging",