@@ -27,5 +27,5 @@ None.
 
 - `hooks/user_prompt_validator.py`：触发判定与语言选择逻辑改动；外部行为变化仅在误触发场景（更少误触发为正向变化）。
 - `scripts/run_check.py`：新增官方 SDK 依赖（`mcp` 包需写入 `requirements.txt` 或类似清单），新增 `out/.codeguard-last.log` 落盘逻辑，删除占位 `mcp_main` 分支。
-- 新建根 `requirements.txt` 声明 `mcp>=1.0`（本仓当前没有任何 requirements 文件；`mcp` 1.26.0 已在开发环境安装，CI 需显式安装）。
+- 新建根 `requirements.txt` 声明 `mcp>=1.0,<2`（2.x 更改了装饰器 API；本仓此前无 requirements 文件，CI 显式安装）。
 - 不修改 vendored 技能（codeguard-* 走 vendor 流程）；不修改 marketplace 清单。
@@ -42,7 +42,7 @@
 
 ## 7. Release
 
-- [ ] 7.1 `node scripts/bump-plugin.mjs codeguard minor` (0.6.6 → 0.7.0; MCP server is a new feature).
-- [ ] 7.2 Sync marketplace manifest + hub catalog.
-- [ ] 7.3 Push, confirm CI green, tag v0.7.0, GitHub Release.
-- [ ] 7.4 `openspec archive 2026-09-22-fix-gate-trigger-and-mcp`.
+- [x] 7.1 Bump minor 0.6.8 → 0.7.0 (MCP server is a new feature; worktree depth broke the hub resolver so manifests were completed manually: 4 manifests + marketplace ref/icon/logo pins + README rows).
+- [x] 7.2 Sync marketplace manifest + hub catalog (repo pins v0.7.0; hub chain landed as `release: codeguard 0.7.0` + README row alignment).
+- [x] 7.3 Push, confirm CI green (PR #21 feature, PR #24 release), tag v0.7.0 at dbc133a, GitHub Release published (README row follow-up at main 0e84186).
+- [x] 7.4 `openspec archive 2026-09-22-fix-gate-trigger-and-mcp`（delta specs 已并入 openspec/specs：gate-trigger-policy +3、mcp-tool-server +4）。
@@ -0,0 +1,44 @@
+# gate-trigger-policy Specification
+
+## Purpose
+TBD - created by archiving change 2026-09-22-fix-gate-trigger-and-mcp. Update Purpose after archive.
+## Requirements
+### Requirement: The UserPromptSubmit hook SHALL run the lint gate only when intent is commit-like
+The `UserPromptSubmit` hook MUST evaluate the user's prompt for commit intent using word-boundary matching of trigger words (`commit|push|deploy|发布|部署`) and a separate action-verb check. The hook SHALL exit silently when either (a) no trigger word is present, (b) the only trigger occurrence is followed by `?`, or (c) a trigger word appears in an interrogative sentence identified by `QUESTION_MARKERS` (`?`, `？`, `么`, `吗`, `如何`, `怎么`, `有没有`, `是不是`, `什么是`, `哪些`).
+
+#### Scenario: User asks a question containing a trigger word
+
+- **WHEN** the user prompt is `使用过程中的情况` or `Should I commit this?` or `先读项目结构`
+- **THEN** the hook exits 0 with empty stdout; no linter runs
+
+#### Scenario: User issues a commit-like imperative
+
+- **WHEN** the user prompt is `请帮我 commit` or `commit this now` or `请帮我 push`
+- **THEN** the hook enters the gate path and invokes `run_gate()`
+
+#### Scenario: Trigger word followed by question mark is exempt
+
+- **WHEN** the user prompt is `commit?` or `要不要 push?` or `should I deploy?`
+- **THEN** the hook exits silently regardless of other QUESTION_MARKERS content
+
+### Requirement: The gate SHALL run only languages named in the user message when they can be identified
+When the trigger fires, the hook MUST extract candidate language ids from the user prompt using case-insensitive token matching against `scripts/languages.json` ids. When the intersection is non-empty, the gate SHALL run only that subset. When the intersection is empty, the gate SHALL fall back to the existing `detect_languages()` behavior.
+
+#### Scenario: User mentions one language explicitly
+
+- **WHEN** the user prompt is `请帮我 commit 这次 ruff 改动`
+- **THEN** the gate runs only Python; other detected languages are skipped for this invocation
+
+#### Scenario: User mentions no language
+
+- **WHEN** the user prompt is `请帮我 commit` with no language token
+- **THEN** the gate runs every language detected by `detect_languages()` (current behavior preserved)
+
+### Requirement: Question markers SHALL cover both Chinese and English
+`QUESTION_MARKERS` MUST include the ASCII `?` in addition to the existing Chinese markers, so that an English prompt ending with `?` is recognized as a question even without other markers.
+
+#### Scenario: ASCII question mark without any other marker
+
+- **WHEN** the user prompt is `Should I push this?`
+- **THEN** the hook exits silently because `?` is a recognized question marker
+
@@ -0,0 +1,47 @@
+# mcp-tool-server Specification
+
+## Purpose
+TBD - created by archiving change 2026-09-22-fix-gate-trigger-and-mcp. Update Purpose after archive.
+## Requirements
+### Requirement: The MCP server SHALL expose three tools over stdio JSON-RPC
+`scripts/run_check.py --mcp` MUST start an stdio JSON-RPC server using the official `mcp` Python SDK and register exactly three tools: `check_code_style`, `auto_fix`, and `list_languages`. The server MUST NOT take any other CLI flag in MCP mode.
+
+#### Scenario: Listing tools at startup
+
+- **WHEN** a client sends `tools/list`
+- **THEN** the response enumerates `check_code_style`, `auto_fix`, `list_languages` with their input schemas
+
+### Requirement: `check_code_style` SHALL return a structured envelope with full stderr on disk
+`check_code_style(path, languages?)` MUST run the configured linters, write the full stderr of any failing invocation to `<log_dir>/.codeguard-last.log` (default `<project>/out/.codeguard-last.log`), and return a JSON envelope with one entry per language: `{language, passed: bool, exit_code: int, stderr_path: str, log_path: str}`.
+
+#### Scenario: One language fails
+
+- **WHEN** the path contains a single Python project whose ruff lint fails
+- **THEN** the envelope contains one entry with `passed=false`, `exit_code` from ruff, `stderr_path` pointing at the on-disk log, and `log_path` echoing the same path
+
+#### Scenario: All languages pass
+
+- **WHEN** every linter exits 0
+- **THEN** every entry has `passed=true`, `exit_code=0`, and the log file is either empty or absent (no stderr written)
+
+### Requirement: `auto_fix` SHALL run formatters and re-verify
+`auto_fix(path, languages?)` MUST invoke the per-language formatter chain (`format` from `scripts/languages.json`) and then re-run `check_code_style` on the same scope before returning. The response envelope MUST contain both `fixed: bool` (whether any file was modified) and the `check_code_style` envelope for the post-fix re-check.
+
+#### Scenario: Formatter succeeds and lint passes
+
+- **WHEN** ruff `--fix` removes unused imports and the re-check passes
+- **THEN** `fixed=true` and the embedded `check_code_style` envelope is all-pass
+
+#### Scenario: Formatter succeeds but lint still fails
+
+- **WHEN** ruff `--fix` ran but a remaining error is non-fixable
+- **THEN** `fixed=true` and the embedded `check_code_style` envelope contains at least one failing entry
+
+### Requirement: `list_languages` SHALL return id + name only
+`list_languages()` MUST return a JSON array of `{id, name}` objects from `scripts/languages.json`, omitting the rest of each language record (lint command, format command, install hint, etc.) so internal commands are not exposed as part of the MCP surface.
+
+#### Scenario: Listing all supported languages
+
+- **WHEN** the client calls `list_languages`
+- **THEN** the response enumerates every entry in `scripts/languages.json` with exactly two fields (`id`, `name`) per record
+