skill create-adr

Reviewed-on: #22

.codex/skills/create-adr/SKILL.md

@@ -0,0 +1,89 @@
+---
+name: create-adr
+description: Use when a change introduces or updates a durable cross-cutting technical decision that future work in this repository must preserve, such as dependency, storage, integration, workflow, or quality-gate policy changes
+---
+
+# Create ADR
+
+## Overview
+
+Create a new ADR in `docs/adr/**` when the current task locks in a durable technical decision.
+Do not use this for bug fixes, styling changes, routine refactors, or ordinary test additions.
+
+## Use This Skill For
+
+- introducing or removing a major dependency
+- choosing or changing a storage strategy
+- defining release, branch, or workflow policy
+- changing how agents, jobs, or external tools are integrated
+- adopting or tightening repository quality gates
+- changing a behavior-defining contract future work must preserve
+
+## Do Not Use This Skill For
+
+- bug fixes
+- styling-only changes
+- routine refactors
+- ordinary test additions
+- temporary plans or execution notes
+
+## Workflow
+
+1. Check whether an active ADR already covers the same decision.
+   - Read `docs/adr/README.md` and scan existing files under `docs/adr/`.
+   - If the current change is only implementing an already-recorded decision, do not create a new ADR.
+2. If the new decision changes or replaces an older ADR, create a new ADR that supersedes it.
+   - Do not rewrite the meaning of an active ADR in place.
+3. Find the next numeric ID under `docs/adr/`.
+4. Create `docs/adr/NNNN-short-kebab-title.md`.
+5. Use this template:
+
+```md
+---
+type: ADR
+id: "NNNN"
+title: "Short decision title"
+status: active
+date: YYYY-MM-DD
+---
+
+## Context
+
+What changed? Why does this decision need to exist now?
+
+## Decision
+
+**State the decision clearly in one or two sentences.**
+
+## Options considered
+
+- **Option A** (chosen): description with pros and cons
+- **Option B**: description with pros and cons
+- **Option C**: description with pros and cons
+
+## Consequences
+
+What becomes easier?
+What becomes harder?
+What risks were accepted?
+What would trigger re-evaluation?
+
+## Advice
+
+Optional. Capture important external guidance or review input.
+```
+
+6. Add the new ADR to the index in `docs/adr/README.md`.
+7. Commit the ADR in the same change as the implementation it documents.
+8. Mention the ADR in the issue comment and PR body under `Docs/ADR`.
+
+## Rule Of Thumb
+
+If the decision changes how future work should be done across the repository, it probably deserves an ADR.
+
+## Repository Rules
+
+- Keep one decision per file.
+- Name files `NNNN-short-kebab-title.md`.
+- Keep the `docs/adr/README.md` index aligned with the files on disk.
+- Prefer concise, durable wording over implementation detail.

.codex/skills/create-adr/agents/openai.yaml

@@ -0,0 +1,3 @@
+interface:
+  display_name: "Create ADR"
+  short_description: "Record a new architecture decision for durable repo changes"

.codex/skills/implement-task/SKILL.md

@@ -0,0 +1,103 @@
+---
+name: implement-task
+description: Phase 2 skill. Use when issue scope is locked by Product Owner (Phase 1 done). Plan, implement, create PR, and own it until merged as Developer.
+---
+
+# Implement Task
+
+**Phase:** 2 — Requirement Implementation
+**Role:** Developer
+
+## When to invoke
+
+- User says: "implement issue #X", "/implement-task #X"
+- Phase 1 is complete (issue has a "Final Requirements" comment from PO).
+
+## Prerequisite
+
+- Issue must have a **Final Requirements** comment locked by PO. If not found, stop and tell the user to run `/validate-issue` first.
+
+## Workflow
+
+1. **Load issue**
+   - Read issue: `mcp__gitea__.issue_read` (`method: "get"`).
+   - Read comments, find the **Final Requirements** comment: `mcp__gitea__.issue_read` (`method: "get_comments"`).
+   - If Final Requirements not found → stop, tell user to validate issue first.
+   - Read current Gitea user: `mcp__gitea__.get_me`.
+   - If current user is not already in the issue assignee list, update the issue with `mcp__gitea__.issue_write` (`method: "update"`) and set `assignees` to the existing assignees plus the current username.
+
+2. **Read codebase**
+   - Use `mcp__gitea__.get_file_contents`, `mcp__gitea__.get_repository_tree` to understand current state.
+   - Identify files and functions that need to change.
+
+3. **Write plan and post on issue**
+   - List implementation steps with verification criteria.
+   - List files to change.
+   - State whether `README.md` or `.env.example` must change.
+   - State whether an ADR is required or not required.
+   - If ADR is required or recommended, invoke `create-adr` before finalizing the implementation summary and PR body.
+   - Flag risks or tradeoffs.
+   - Post plan as comment: `mcp__gitea__.issue_write` (`method: "add_comment"`).
+   - **Wait for "Approved" before coding.** Stop here.
+
+4. **Implement** (only after plan is approved)
+   - Create feature branch from `main`:
+     ```
+     git checkout main && git pull origin main
+     git checkout -b feat/<short-description>
+     ```
+   - If currently on another branch, use worktree:
+     ```
+     git worktree add .worktrees/<branch-name> main
+     ```
+   - Implement following the plan (surgical changes only).
+   - Update `README.md` and `.env.example` when config, endpoint, or runtime behavior changes.
+   - If the change introduces a durable behavior, contract, architecture, or workflow decision, use `create-adr` and create/update `docs/adr/**`.
+   - Run validation:
+     ```
+     gofmt -l .
+     go test ./...
+     go vet ./...
+     staticcheck ./...
+     ```
+   - Post an implementation summary comment on the issue with sections:
+     - `Summary`
+     - `Changes`
+     - `Test Plan`
+     - `Refactor`
+     - `Docs/ADR`
+     - `Risks/Follow-ups`
+
+5. **Create PR**
+   - Use `mcp__gitea__.pull_request_write` (`method: "create"`).
+   - PR body must include:
+     - `Summary`
+     - `Changes`
+     - `Test Plan` with exact commands and output
+     - `Refactor`
+     - `Docs/ADR`
+     - `Risks/Follow-ups`
+     - `Fixes #<issue-id>`
+   - Add reviewer.
+
+6. **Own the PR until merged**
+   - Monitor for review feedback: `mcp__gitea__.pull_request_read` (`method: "get_reviews"`, `get_review_comments`).
+   - When feedback arrives:
+     - Checkout the PR branch (or use worktree if on another branch).
+     - Fix code locally, commit, push.
+     - Reply on PR: `mcp__gitea__.pull_request_review_write` (`method: "create"`, `state: "COMMENT"`) explaining what was fixed.
+     - Never edit remote files directly via API.
+   - After reviewer approves (APPROVED), merge the PR if user confirms.
+
+## Rules
+- Use Gitea MCP tools for all issue/PR operations.
+- Preserve existing assignees when adding yourself unless the user explicitly asks to replace them.
+- Never edit remote files directly — always work on local branch.
+- Satisfy the repo Definition of Done before claiming completion.
+- Missing required tests, docs, `.env.example` updates, or ADRs is a quality failure unless explicitly justified in both the issue comment and PR body.
+- Never claim completion without fresh verification output.
+- Keep changes strictly scoped to the approved plan.
+- If the change needs durable documentation, use `create-adr` and create/update `docs/adr/**`. Do not commit `docs/superpowers/**` planning artifacts.
+
+## Next skill
+After PR is created, the user invokes `/review-pr <pr-number>` on the Reviewer agent.

.codex/skills/review-pr/SKILL.md

@@ -0,0 +1,126 @@
+---
+name: review-pr
+description: Phase 2 skill. Use when a PR is created or updated. Review PR against locked issue requirements and post verdict as Reviewer. Do NOT edit code on the PR.
+---
+
+# Review PR
+
+**Phase:** 2 — Requirement Implementation
+**Role:** Reviewer
+
+## When to invoke
+
+- User says: "review PR #X", "check PR #X", "/review-pr #X"
+- A PR is created or updated and needs review.
+
+## Prerequisite
+
+- PR exists on Gitea.
+- The referenced issue has a **Final Requirements** comment from PO (Phase 1 done).
+
+## Workflow
+
+1. **Load PR and issue**
+   - Read PR: `mcp__gitea__.pull_request_read` (`method: "get"`).
+   - If PR is already merged or closed, say so and stop.
+   - Read current Gitea user: `mcp__gitea__.get_me`.
+   - Read existing reviews: `mcp__gitea__.pull_request_read` (`method: "get_reviews"`).
+   - If the current user is not already requested as a reviewer and has not already reviewed this PR, add them with `mcp__gitea__.pull_request_write` (`method: "add_reviewers"`).
+   - Identify the linked issue from PR body (`Fixes #<id>` or `Closes #<id>`).
+   - Read issue comments, find the **Final Requirements** comment: `mcp__gitea__.issue_read` (`method: "get_comments"`).
+   - Read PR discussion comments: `mcp__gitea__.issue_read` (`method: "get_comments"`), using the PR number as the index.
+
+2. **Read review history** (for re-reviews)
+   - Get review comments: `mcp__gitea__.pull_request_read` (`method: "get_review_comments"`).
+   - Treat reviews, inline review comments, and PR discussion comments as separate context sources.
+   - If author addressed feedback in a follow-up commit, focus on new commits but still sanity-check the full diff.
+
+3. **Get the diff**
+   - Use `mcp__gitea__.pull_request_read` (`method: "get_diff"`).
+   - Use `mcp__gitea__.get_file_contents` for surrounding context when needed.
+
+4. **Validate locally** (when needed)
+   - Create worktree:
+     ```
+     git worktree add .worktrees/pr-<number> origin/<head-branch>
+     ```
+   - Run checks:
+     ```
+     gofmt -l .
+     go test ./...
+     go vet ./...
+     staticcheck ./...
+     ```
+   - Remove worktree when done:
+     ```
+     git worktree remove .worktrees/pr-<number>
+     ```
+
+5. **Evaluate**
+   - Does PR address **all** Final Requirements from the issue?
+   - Does PR satisfy the repo Definition of Done?
+     - relevant tests for behavior/config changes
+     - required `README.md` / `.env.example` updates
+     - ADR present when the change locks in a durable decision
+     - required implementation evidence exists on the issue and in the PR body
+   - Does PR contain unrelated changes or refactoring? → FAIL immediately.
+   - Code correctness, style, test coverage.
+
+6. **Write verdict**
+   - Start with **PASS** or **FAIL**.
+   - Include file:line references for findings.
+   - Explain concrete impact, not style preferences.
+
+7. **Post review**
+   - Post proactively if verdict is clear.
+   - Use `mcp__gitea__.pull_request_review_write` (`method: "create"`):
+     - `PASS` → `state: "APPROVED"`
+     - `FAIL` → `state: "REQUEST_CHANGES"`
+     - Non-blocking note → `state: "COMMENT"`
+   - For plain discussion (not changing review state): `mcp__gitea__.issue_write` (`method: "add_comment"`).
+
+## Review Format
+
+### PASS
+```
+PASS
+
+No blocking issues found.
+
+Validation:
+- gofmt -l .: OK
+- go vet ./...: OK
+- go test ./...: OK
+- staticcheck ./...: OK
+
+Non-blocking:
+- <optional note>
+```
+
+### FAIL
+```
+FAIL
+
+1. [High] <issue with path:line and impact>
+2. [Medium] <issue with path:line and impact>
+
+Validation:
+- go test ./...: FAIL (<details>)
+```
+
+## Rules
+- Use Gitea MCP tools for all PR/review operations.
+- Do not add yourself as reviewer twice; check requested reviewers and existing reviews first.
+- Do NOT edit code on the PR.
+- FAIL if PR contains unrelated changes or scope creep.
+- FAIL if the PR misses required DoD evidence, docs, or ADR work for the accepted change.
+- Post review proactively if verdict is clear.
+- Do not post a verdict until PR discussion comments have been checked.
+- Mention validation commands you actually ran.
+- No bracketed author/tool tags.
+- Accept `docs/adr/**` when it records a direct, durable decision tied to the issue scope.
+- Treat `docs/superpowers/**` and other agent execution artifacts as scope creep. When rejecting docs, explain whether the content should be deleted entirely or rewritten as an ADR.
+
+## What happens next
+- **PASS (APPROVED):** Developer merges the PR.
+- **FAIL (REQUEST_CHANGES):** Developer fixes feedback, then user re-invokes `/review-pr` on this PR.

.codex/skills/validate-issue/SKILL.md

@@ -0,0 +1,77 @@
+---
+name: validate-issue
+description: Phase 1 skill. Use when a Gitea issue is created or updated and needs requirement validation. Analyze, clarify, and lock scope as Product Owner. Do NOT implement.
+---
+
+# Validate Issue
+
+**Phase:** 1 — Requirement Refinement
+**Role:** Product Owner
+
+## When to invoke
+
+- User says: "check issue #X", "validate issue #X", "/validate-issue #X"
+- A new issue is created or an existing issue is updated with new information.
+
+## Prerequisite
+
+- Issue exists on Gitea.
+
+## Workflow
+
+1. **Load issue**
+   - Read issue: `mcp__gitea__.issue_read` (`method: "get"`).
+   - Read comments: `mcp__gitea__.issue_read` (`method: "get_comments"`).
+
+2. **Analyze**
+   - Is the problem statement clear?
+   - Is the scope well-defined or too broad?
+   - Is it technically feasible?
+   - What acceptance criteria will prove the change is done?
+   - Will the change require docs updates or a durable ADR?
+   - If needed, read relevant code: `mcp__gitea__.get_file_contents`.
+
+3. **Decide**
+   - **Unclear or incomplete:**
+     - Post clarifying questions: `mcp__gitea__.issue_write` (`method: "add_comment"`).
+     - Stop. Wait for user to respond, then re-run this skill.
+   - **Clear and feasible:**
+     - Post Final Requirements (see format below) with specific acceptance criteria.
+     - State whether docs updates are expected.
+     - State whether an ADR is required, recommended, or not needed.
+     - If ADR is `required` or `recommended`, tell the Developer to use `create-adr` during implementation.
+     - This locks the scope and signals Phase 2 can begin.
+
+## Final Requirements format
+
+```
+## Final Requirements
+
+**Summary:** <one-line summary>
+
+**Scope:**
+- <requirement 1>
+- <requirement 2>
+
+**Out of scope:**
+- <explicitly excluded items>
+
+**Acceptance criteria:**
+- <verifiable check 1>
+- <verifiable check 2>
+
+**Documentation impact:**
+- <README.md / .env.example updates required or not required>
+
+**ADR:**
+- <required / recommended / not needed, with one-line reason>
+```
+
+## Rules
+- Use Gitea MCP tools for all operations.
+- Do NOT write code, create branches, or create PRs.
+- Make acceptance criteria specific enough that Dev and Reviewer can verify Definition of Done without guessing.
+- Do NOT proceed to Phase 2 — that is a separate skill.
+
+## Next skill
+After Final Requirements are posted, the user invokes `/implement-task <issue-id>` on the Developer agent.

.env.example

@@ -18,6 +18,7 @@ OPENCLAW_API_KEY=
 BRIDGE_CALLBACK_TOKEN=a-strong-random-token
 SYSTEM_EMAIL=bridge@example.com
 WHITELIST_EMAILS=user1@example.com,user2@example.com
+BLOCKLIST_EMAILS=
 
 # === Optional ===
 IMAP_PROXY_URL=

.gitea/workflows/ci.yml

@@ -6,11 +6,10 @@ on:
 jobs:
   fmt:
     runs-on: linux
+    container:
+      image: git.thuanle.me/public/node-go
     steps:
       - uses: actions/checkout@v4
-      - uses: actions/setup-go@v5
-        with:
-          go-version-file: go.mod
       - run: |
           out=$(gofmt -l .)
           if [ -n "$out" ]; then
@@ -21,9 +20,25 @@ jobs:
 
   test:
     runs-on: linux
+    container:
+      image: git.thuanle.me/public/node-go
     steps:
       - uses: actions/checkout@v4
-      - uses: actions/setup-go@v5
-        with:
-          go-version-file: go.mod
       - run: go test ./...
+
+  vet:
+    runs-on: linux
+    container:
+      image: git.thuanle.me/public/node-go
+    steps:
+      - uses: actions/checkout@v4
+      - run: go vet ./...
+
+  staticcheck:
+    runs-on: linux
+    container:
+      image: git.thuanle.me/public/node-go
+    steps:
+      - uses: actions/checkout@v4
+      - run: go install honnef.co/go/tools/cmd/staticcheck@2025.1.1
+      - run: $(go env GOPATH)/bin/staticcheck ./...

AGENT.md

@@ -0,0 +1,166 @@
+# Agent Instructions
+
+## 1. Coding Principles
+
+### 1.1 Simplicity First
+
+- No out-of-scope features.
+- No abstractions for single-use code.
+- If 50 lines work instead of 200, write 50.
+
+### 1.2 Surgical Changes
+
+- Only change lines directly related to the request.
+- Never refactor adjacent code unless asked.
+- Remove orphaned imports/variables/functions your changes created.
+
+### 1.3 Goal-Driven Execution
+
+- Define success criteria before coding.
+- Verify each step before moving to the next.
+
+---
+
+## 2. Workflow Overview
+
+All work starts from a Gitea issue. Two phases, each with a dedicated role:
+
+1. **Phase 1 — Requirement Refinement** (Product Owner): Clarify and lock scope.
+2. **Phase 2 — Requirement Implementation** (Developer & Reviewer): Plan, code, review, merge.
+
+### 2.1 Definition of Done
+
+A change is done only when all of the following are true:
+
+- The issue has locked **Final Requirements** with verifiable acceptance criteria.
+- The implementation stays within approved scope; unrelated refactors are not allowed.
+- Behavior, contract, or config changes have relevant tests added or updated, or the change record explains why tests were not added.
+- Validation passes with fresh output: `gofmt -l .`, `go test ./...`, `go vet ./...`, `staticcheck ./...`.
+- `README.md` and `.env.example` and related documents are updated when config, endpoint, or runtime behavior changes.
+- A new ADR is added in `docs/adr/**` when the change introduces a durable architecture, contract, behavior, or workflow decision.
+- The implementation evidence is posted on the issue and reflected in the PR body:
+  - what changed
+  - what tests were run
+  - whether any refactor was included
+  - whether docs or ADRs were added or not needed
+  - risks or follow-ups
+- After merge, the linked issue is verified closed.
+
+---
+
+## 3. Phase 1: Requirement Refinement
+
+**Role:** Product Owner (PO)
+**Goal:** Turn a raw issue into a clear, scoped, feasible requirement.
+**Constraint:** PO must NOT write code or create PRs.
+
+### Steps
+
+1. Read and analyze the issue.
+2. Is the requirement clear and well-scoped?
+   - **No:** Comment clarifying questions on the issue. Stop and wait for user response.
+   - **Yes:** Comment a summary and lock the scope (Final Requirements).
+3. Locked issue moves to Phase 2.
+
+---
+
+## 4. Phase 2: Requirement Implementation
+
+**Roles:** Developer (Dev) and Reviewer (Rev)
+**Goal:** Implement the locked requirement from Phase 1.
+
+### 4.1 Developer: Plan & Implement
+
+1. Read the locked issue and any clarification comments.
+2. Read relevant code to understand current state.
+3. Write an implementation plan — list steps, files to change, risks, verification criteria, and any docs/ADR impact.
+4. **Comment the plan on the issue. Wait for "Approved" before coding.**
+5. Once approved:
+   - Create a feature branch from `main`.
+   - Implement following the plan (surgical changes only).
+   - Update `README.md` and `.env.example` if config, endpoint, or runtime behavior changes.
+   - If the change locks in a durable decision, write/update an ADR in `docs/adr/**`.
+   - Run validation: `gofmt -l .`, `go test ./...`, `go vet ./...`, `staticcheck ./...`.
+   - Post an implementation summary comment on the issue with sections: `Summary`, `Changes`, `Test Plan`, `Refactor`, `Docs/ADR`, `Risks/Follow-ups`.
+   - Create PR with the same evidence in the body plus `Fixes #<issue>` so Gitea auto-closes the issue on merge.
+   - After merge, verify the issue is closed. If not (e.g. missing keyword), close it manually with a summary comment.
+6. **Own the PR until merged:**
+   - Monitor for review feedback.
+   - Fix feedback on the correct branch (checkout or worktree, never edit remote via API).
+   - Always reply on PR after pushing a fix.
+
+### 4.2 Reviewer: Review PR
+
+**Constraint:** Reviewer must NOT edit code on the PR.
+
+1. Re-read the original issue — verify PR addresses the locked requirement.
+2. Read PR diff, review history, and PR discussion comments.
+3. Validate in worktree if needed (`.worktrees/pr-<number>`, run `gofmt -l .`, `go test ./...`, `go vet ./...`, `staticcheck ./...`).
+4. Check for scope creep — FAIL if PR contains unrelated changes or refactoring.
+5. Verify the repo Definition of Done:
+   - relevant tests exist for behavior/config changes
+   - docs and `.env.example` were updated when required
+   - ADR exists when a durable decision was introduced
+   - the issue comment and PR body contain the required implementation evidence
+6. Write verdict — start with **PASS** or **FAIL**.
+7. Post review proactively if verdict is clear:
+   - `PASS` → `APPROVED`
+   - `FAIL` → `REQUEST_CHANGES`
+   - Non-blocking note → `COMMENT`
+
+---
+
+## 5. Gitea Tools
+
+- Always use Gitea MCP tools in this project.
+- Scope: read PR/issue, list comments, post replies, create/edit PRs, reviews.
+- Use `tea` CLI only as fallback when MCP is unavailable.
+- Local CLI tools (`git`, `go`, `rg`) are fine for local validation.
+
+---
+
+## 6. Review Format
+
+```
+PASS
+
+No blocking issues found.
+
+Validation:
+- gofmt -l .: OK
+- go test ./...: OK
+- go vet ./...: OK
+- staticcheck ./...: OK
+
+Non-blocking:
+- <optional note>
+```
+
+```
+FAIL
+
+1. [High] <issue with path:line and impact>
+2. [Medium] <issue with path:line and impact>
+
+Validation:
+- <list only the commands you actually ran and their result>
+```
+
+---
+
+## 7. Documentation Rules
+
+- Durable architecture or behavior decisions belong in `docs/adr/**`.
+- `docs/superpowers/**` is for agent execution artifacts and must not be merged as part of product implementation PRs.
+- If a PR needs a lasting technical rationale, author an ADR in `docs/adr/` instead.
+- Reviewers should treat temporary plan/spec docs as scope creep, but accept ADRs that directly document the accepted behavior of the change.
+
+---
+
+## 8. Comment Style
+
+- Use readable multiline markdown.
+- Use short sections (`Summary`, `Changes`, `Test Plan`).
+- Implementation completion comments should also include `Refactor`, `Docs/ADR`, and `Risks/Follow-ups`.
+- Concise bullet points and explicit line breaks.
+- No bracketed author/tool tags.

CLAUDE.md

@@ -1,70 +1,5 @@
 # CLAUDE.md
 
-Behavioral guidelines to reduce common LLM coding mistakes. Merge with project-specific instructions as needed.
+See [AGENT.md](AGENT.md) for full project instructions and workflow.
 
-**Tradeoff:** These guidelines bias toward caution over speed. For trivial tasks, use judgment.
-
-## 1. Think Before Coding
-
-**Don't assume. Don't hide confusion. Surface tradeoffs.**
-
-Before implementing:
-
-- State your assumptions explicitly. If uncertain, ask.
-- If multiple interpretations exist, present them - don't pick silently.
-- If a simpler approach exists, say so. Push back when warranted.
-- If something is unclear, stop. Name what's confusing. Ask.
-
-## 2. Simplicity First
-
-**Minimum code that solves the problem. Nothing speculative.**
-
-- No features beyond what was asked.
-- No abstractions for single-use code.
-- No "flexibility" or "configurability" that wasn't requested.
-- No error handling for impossible scenarios.
-- If you write 200 lines and it could be 50, rewrite it.
-
-Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.
-
-## 3. Surgical Changes
-
-**Touch only what you must. Clean up only your own mess.**
-
-When editing existing code:
-
-- Don't "improve" adjacent code, comments, or formatting.
-- Don't refactor things that aren't broken.
-- Match existing style, even if you'd do it differently.
-- If you notice unrelated dead code, mention it - don't delete it.
-
-When your changes create orphans:
-
-- Remove imports/variables/functions that YOUR changes made unused.
-- Don't remove pre-existing dead code unless asked.
-
-The test: Every changed line should trace directly to the user's request.
-
-## 4. Goal-Driven Execution
-
-**Define success criteria. Loop until verified.**
-
-Transform tasks into verifiable goals:
-
-- "Add validation" → "Write tests for invalid inputs, then make them pass"
-- "Fix the bug" → "Write a test that reproduces it, then make it pass"
-- "Refactor X" → "Ensure tests pass before and after"
-
-For multi-step tasks, state a brief plan:
-
-```
-1. [Step] → verify: [check]
-2. [Step] → verify: [check]
-3. [Step] → verify: [check]
-```
-
-Strong success criteria let you loop independently. Weak criteria ("make it work") require constant clarification.
-
----
-
-**These guidelines are working if:** fewer unnecessary changes in diffs, fewer rewrites due to overcomplication, and clarifying questions come before implementation rather than after mistakes.
+Agent skills are stored in `.codex/skills`

README.md

@@ -3,7 +3,7 @@
 Tự động xử lý email bằng OpenClaw AI theo luồng:
 
 ```
-Ingress (IMAP) → Dispatch (OpenClaw) → Callback → Egress (SMTP reply)
+Ingress (IMAP) → Safety Checks → External Rules → Apply Context → Dispatch (OpenClaw) → Callback → Egress (SMTP reply)
 ```
 
 ## Tính năng
@@ -12,6 +12,7 @@ Ingress (IMAP) → Dispatch (OpenClaw) → Callback → Egress (SMTP reply)
 - **Threading** — giữ nguyên email thread qua `In-Reply-To` / `References`
 - **Idempotency** — chống duplicate theo `message_id`
 - **Anti-loop & Whitelist** — bỏ qua email từ chính hệ thống, chỉ xử lý sender được phép
+- **External Rules** — pipeline configurable: whitelist, blocklist, và rule metadata cho dispatch
 - **Retry** — tối đa 3 lần với backoff 1s → 5s → 15s cho cả OpenClaw và SMTP
 - **Stateful** — SQLite lưu trạng thái pipeline, không mất dữ liệu khi restart
 

cmd/bridge/main.go

@@ -3,10 +3,12 @@ package main
 import (
 	"context"
 	"log/slog"
+	"net/http"
 	"os"
 	"os/signal"
 	"sync"
 	"syscall"
+	"time"
 
 	"thuanle.me/claw-email-bridge/internal/ai_client"
 	"thuanle.me/claw-email-bridge/internal/api"
@@ -62,16 +64,30 @@ func main() {
 	// Start HTTP server with SMTP egress on callback.
 	smtpSender := mail.NewSMTPSender(cfg, db)
 	router := api.NewRouter(db, cfg.BridgeCallbackToken, smtpSender.SendReplyFunc())
+	srv := &http.Server{Addr: cfg.ListenAddr, Handler: router}
+
 	wg.Add(1)
 	go func() {
 		defer wg.Done()
 		slog.Info("starting HTTP server", "addr", cfg.ListenAddr)
-		if err := router.Run(cfg.ListenAddr); err != nil {
+		if err := srv.ListenAndServe(); err != nil && err != http.ErrServerClosed {
 			slog.Error("server exited with error", "error", err)
 			cancel()
 		}
 	}()
 
+	// Graceful shutdown: wait for context cancel then shutdown HTTP server.
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		<-ctx.Done()
+		shutdownCtx, shutdownCancel := context.WithTimeout(context.Background(), 5*time.Second)
+		defer shutdownCancel()
+		if err := srv.Shutdown(shutdownCtx); err != nil {
+			slog.Error("server shutdown error", "error", err)
+		}
+	}()
+
 	wg.Wait()
 	slog.Info("bridge stopped")
 }

docs/adr/0001-imap-thread-resolution.md

@@ -0,0 +1,35 @@
+---
+type: ADR
+id: "0001"
+title: "IMAP thread resolution prefers In-Reply-To and falls back to References"
+status: active
+date: 2026-04-28
+---
+
+## Context
+
+The IMAP ingress flow stores a `thread_id` before dispatching work to OpenClaw. Previously the bridge only used `In-Reply-To` to find a parent task. Some valid email clients preserve thread ancestry through `References` even when `In-Reply-To` is absent. In those cases, the bridge created a new `thread_id`, which broke conversation continuity and downstream session reuse.
+
+This decision needs to be durable because future refactors of mail parsing, dispatch, or storage must preserve the same thread-resolution behavior.
+
+## Decision
+
+The bridge resolves parent thread linkage by trying `In-Reply-To` first. If no parent task is found, it falls back to `References`, scanning message IDs from last to first and using the first match found in the database. If neither header yields a known parent, the message's own `message_id` becomes the new `thread_id`.
+
+## Options considered
+
+- **Use `In-Reply-To` first, then `References` last-to-first** (chosen): preserves existing behavior, matches common email ancestry ordering, and favors the most recent known parent.
+- **Use only `In-Reply-To`**: simpler, but breaks valid threads when clients omit `In-Reply-To`.
+- **Search `References` from first to last**: may attach to an older ancestor even when a more recent known parent exists.
+- **Normalize all thread linkage into a more complex conversation graph**: more flexible, but out of scope for the current bridge behavior.
+
+## Consequences
+
+Thread continuity is improved for replies that rely on `References`.
+Future contributors have an explicit behavioral contract for threading semantics.
+Mail parsing changes must preserve the header precedence and reverse scan behavior unless a new ADR supersedes this one.
+The current approach still depends on existing message IDs being present in the database; if upstream behavior changes significantly, this decision may need reevaluation.
+
+## Advice
+
+If a future change wants different precedence rules or deeper conversation reconstruction, create a new ADR rather than silently changing this behavior.

docs/adr/0002-external-rules-pipeline.md

@@ -0,0 +1,30 @@
+---
+type: ADR
+id: "0002"
+title: "Ingress uses a configurable external rules pipeline before dispatch"
+status: active
+date: 2026-04-28
+---
+
+## Context
+
+The IMAP ingress flow previously hardcoded whitelist and idempotency checks directly in processMessage. There was no extension point for operator-defined policy such as sender blocking or context-based routing. The flow mixed internal safety rules (anti-loop) with configurable policy (whitelist).
+
+## Decision
+
+Separate internal safety checks (anti-loop, idempotency) from external configurable rules. External rules run in a pipeline after safety checks. Each rule implements a `Rule` interface that evaluates `EmailContext` and returns `RuleResult` (accept/reject + metadata). Rule metadata is persisted as `DispatchContext` on the Task and passed to OpenClaw dispatch.
+
+Pipeline order: anti-loop (internal) → idempotency (internal) → external rules → threading → save → dispatch.
+
+## Options considered
+
+- **Pipeline with Rule interface** (chosen): extensible, testable, each rule is isolated.
+- **Keep everything in processMessage**: simpler but not configurable or testable.
+- **Plugin-based rules via config files**: more flexible but over-engineered for current needs.
+
+## Consequences
+
+New rules can be added by implementing the Rule interface and registering in config.
+Whitelist and blocklist are now env-configurable and independently testable.
+Future rules (domain routing, context selection) fit the same pipeline.
+Task model has a new DispatchContext column (auto-migrated).

docs/adr/README.md

@@ -0,0 +1,76 @@
+# Architecture Decision Records
+
+This directory contains Architecture Decision Records (ADRs) for durable technical decisions in this repository.
+
+## Purpose
+
+ADRs capture cross-cutting or behavior-defining decisions that future contributors need to preserve.
+
+Use an ADR when a change introduces or locks in a decision such as:
+
+- runtime behavior that future work must preserve
+- repository-wide technical policy
+- architecture or integration contracts
+- storage or threading semantics
+- quality gate or workflow policy
+
+Do not use ADRs for:
+
+- temporary implementation plans
+- agent execution checklists
+- one-off debugging notes
+- PR-local task breakdowns
+
+## Format
+
+Each ADR is a markdown file with YAML frontmatter.
+
+```md
+---
+type: ADR
+id: "NNNN"
+title: "Short decision title"
+status: active
+date: YYYY-MM-DD
+---
+
+## Context
+
+Why does this decision need to exist now?
+
+## Decision
+
+**State the decision clearly in one or two sentences.**
+
+## Options considered
+
+- **Option A** (chosen): description with pros and cons
+- **Option B**: description with pros and cons
+- **Option C**: description with pros and cons
+
+## Consequences
+
+What becomes easier?
+What becomes harder?
+What risks were accepted?
+What would trigger re-evaluation?
+
+## Advice
+
+Optional. Capture important external input or review notes.
+```
+
+## Rules
+
+- One decision per file.
+- Name files `NNNN-short-kebab-title.md`.
+- Once active, do not rewrite an ADR's meaning later. Supersede it with a new ADR instead.
+- Keep the index below aligned with the current files in this directory.
+- Do not keep placeholder or mock entries in the index.
+
+## Index
+
+| ID | Title | Status |
+|----|-------|--------|
+| 0001 | IMAP thread resolution prefers In-Reply-To and falls back to References | active |
+| 0002 | Ingress uses a configurable external rules pipeline before dispatch | active |

internal/ai_client/client.go

@@ -73,7 +73,7 @@ func (d *Dispatcher) Dispatch(task *database.Task) {
 		SessionID:   task.ThreadID,
 		History:     history,
 		CallbackURL: callbackURL,
-		Metadata:    map[string]string{"task_uuid": task.TaskUUID},
+		Metadata:    buildDispatchMetadata(task),
 	}
 
 	// Update status to AI_PROCESSING before first attempt.
@@ -195,3 +195,25 @@ func (d *Dispatcher) buildHistory(threadID string) ([]historyEntry, error) {
 
 	return history, nil
 }
+
+// ExportBuildDispatchMetadata exports buildDispatchMetadata for testing.
+func ExportBuildDispatchMetadata(task *database.Task) map[string]string {
+	return buildDispatchMetadata(task)
+}
+
+// buildDispatchMetadata constructs the metadata map for OpenClaw dispatch,
+// merging the task UUID with any DispatchContext from the rules pipeline.
+func buildDispatchMetadata(task *database.Task) map[string]string {
+	metadata := map[string]string{"task_uuid": task.TaskUUID}
+	if task.DispatchContext != "" {
+		var ctx map[string]string
+		if err := json.Unmarshal([]byte(task.DispatchContext), &ctx); err == nil {
+			for k, v := range ctx {
+				if k != "task_uuid" {
+					metadata[k] = v
+				}
+			}
+		}
+	}
+	return metadata
+}

internal/ai_client/client_test.go

@@ -12,6 +12,53 @@ import (
 	"thuanle.me/claw-email-bridge/internal/database"
 )
 
+func TestBuildDispatchMetadata_ForwardsContext(t *testing.T) {
+	task := &database.Task{
+		TaskUUID:        "uuid-ctx-test",
+		DispatchContext: `{"rule":"whitelist","source":"env"}`,
+	}
+	metadata := ai_client.ExportBuildDispatchMetadata(task)
+
+	if metadata["task_uuid"] != "uuid-ctx-test" {
+		t.Errorf("expected task_uuid=uuid-ctx-test, got %q", metadata["task_uuid"])
+	}
+	if metadata["rule"] != "whitelist" {
+		t.Errorf("expected rule=whitelist, got %q", metadata["rule"])
+	}
+	if metadata["source"] != "env" {
+		t.Errorf("expected source=env, got %q", metadata["source"])
+	}
+}
+
+func TestBuildDispatchMetadata_ProtectsTaskUUID(t *testing.T) {
+	task := &database.Task{
+		TaskUUID:        "original-uuid",
+		DispatchContext: `{"task_uuid":"malicious-override","extra":"data"}`,
+	}
+	metadata := ai_client.ExportBuildDispatchMetadata(task)
+
+	if metadata["task_uuid"] != "original-uuid" {
+		t.Errorf("task_uuid should not be overwritten, got %q", metadata["task_uuid"])
+	}
+	if metadata["extra"] != "data" {
+		t.Errorf("expected extra=data, got %q", metadata["extra"])
+	}
+}
+
+func TestBuildDispatchMetadata_NoContext(t *testing.T) {
+	task := &database.Task{
+		TaskUUID: "uuid-no-ctx",
+	}
+	metadata := ai_client.ExportBuildDispatchMetadata(task)
+
+	if len(metadata) != 1 {
+		t.Errorf("expected 1 key, got %d", len(metadata))
+	}
+	if metadata["task_uuid"] != "uuid-no-ctx" {
+		t.Errorf("expected task_uuid=uuid-no-ctx, got %q", metadata["task_uuid"])
+	}
+}
+
 // Test matrix #2: OpenClaw fail 2 lần, lần 3 thành công → COMPLETED, attempt_openclaw = 3.
 func TestDispatch_RetryThenSuccess(t *testing.T) {
 	tdb := database.NewTestDB(t)

internal/config/config.go

@@ -31,6 +31,7 @@ type Config struct {
 	BridgeCallbackToken string
 	SystemEmail         string
 	WhitelistEmails     []string
+	BlocklistEmails     []string
 
 	// Optional
 	IMAPProxyURL string
@@ -74,6 +75,14 @@ func Load() (*Config, error) {
 		}
 	}
 
+	if raw := os.Getenv("BLOCKLIST_EMAILS"); raw != "" {
+		for _, email := range strings.Split(raw, ",") {
+			if trimmed := strings.TrimSpace(email); trimmed != "" {
+				cfg.BlocklistEmails = append(cfg.BlocklistEmails, trimmed)
+			}
+		}
+	}
+
 	if cfg.ListenAddr == "" {
 		cfg.ListenAddr = ":8080"
 	}

internal/database/models.go

@@ -24,6 +24,7 @@ type Task struct {
 	Sender          string
 	Subject         string
 	BodyPlain       string
+	DispatchContext string // JSON metadata from external rules pipeline.
 	AIResponse      string
 	Status          string `gorm:"index;not null"`
 	AttemptOpenClaw int    `gorm:"default:0"`

internal/mail/imap.go

@@ -1,12 +1,16 @@
 package mail
 
 import (
+	"bufio"
 	"context"
 	"crypto/tls"
+	"encoding/base64"
+	"encoding/json"
 	"fmt"
 	"io"
 	"log/slog"
 	"net"
+	"net/http"
 	"net/url"
 	"strings"
 	"time"
@@ -21,6 +25,7 @@ import (
 	"thuanle.me/claw-email-bridge/internal/config"
 	"thuanle.me/claw-email-bridge/internal/database"
 	"thuanle.me/claw-email-bridge/internal/logging"
+	"thuanle.me/claw-email-bridge/internal/rules"
 )
 
 // IMAPWatcher monitors an IMAP mailbox via IDLE and processes new emails.
@@ -32,6 +37,8 @@ type IMAPWatcher struct {
 	dialIMAP         func(addr string, options *imapclient.Options) (*imapclient.Client, error)
 	dialIMAPViaProxy func(addr, proxyURL string, options *imapclient.Options) (*imapclient.Client, error)
 
+	rules *rules.Pipeline
+
 	// onReceived is called after a task is saved as RECEIVED.
 	// This will be wired to the OpenClaw dispatch in a future step.
 	OnReceived func(task *database.Task)
@@ -55,13 +62,26 @@ func (d timeoutContextDialer) DialContext(ctx context.Context, network, address
 
 var socks5DialerFactory = proxy.SOCKS5
 
+var proxyTLSConfigForTest = func(host string) *tls.Config {
+	return &tls.Config{ServerName: host}
+}
+
 // NewIMAPWatcher creates a new IMAP watcher.
 func NewIMAPWatcher(cfg *config.Config, db *gorm.DB) *IMAPWatcher {
+	var externalRules []rules.Rule
+	if len(cfg.WhitelistEmails) > 0 {
+		externalRules = append(externalRules, rules.NewWhitelistRule(cfg.WhitelistEmails))
+	}
+	if len(cfg.BlocklistEmails) > 0 {
+		externalRules = append(externalRules, rules.NewBlocklistRule(cfg.BlocklistEmails))
+	}
+
 	return &IMAPWatcher{
 		cfg:              cfg,
 		db:               db,
 		dialIMAP:         imapclient.DialTLS,
-		dialIMAPViaProxy: dialTLSViaSOCKS5,
+		dialIMAPViaProxy: dialTLSViaProxy,
+		rules:            rules.NewPipeline(externalRules),
 	}
 }
 
@@ -180,34 +200,20 @@ func (w *IMAPWatcher) connectAndWatch(ctx context.Context) error {
 	}
 }
 
-func dialTLSViaSOCKS5(addr, proxyURL string, options *imapclient.Options) (*imapclient.Client, error) {
+func dialTLSViaProxy(addr, proxyURL string, options *imapclient.Options) (*imapclient.Client, error) {
 	u, err := url.Parse(proxyURL)
 	if err != nil {
 		return nil, fmt.Errorf("parse proxy url: %w", err)
 	}
-	if u.Scheme != "socks5" {
-		return nil, fmt.Errorf("unsupported proxy scheme: %s", u.Scheme)
-	}
-
-	var auth *proxy.Auth
-	if u.User != nil {
-		pw, _ := u.User.Password()
-		auth = &proxy.Auth{User: u.User.Username(), Password: pw}
-	}
-
-	dialer, err := socks5DialerFactory("tcp", u.Host, auth, timeoutContextDialer{timeout: imapDialTimeout})
-	if err != nil {
-		return nil, fmt.Errorf("create socks5 dialer: %w", err)
-	}
-
-	dialCtx, dialCancel := context.WithTimeout(context.Background(), imapDialTimeout)
-	defer dialCancel()
 
 	var conn net.Conn
-	if cd, ok := dialer.(proxy.ContextDialer); ok {
-		conn, err = cd.DialContext(dialCtx, "tcp", addr)
-	} else {
-		conn, err = dialer.Dial("tcp", addr)
+	switch u.Scheme {
+	case "socks5":
+		conn, err = dialViaSOCKS5(u, addr)
+	case "http", "https":
+		conn, err = dialViaCONNECT(u, addr)
+	default:
+		return nil, fmt.Errorf("unsupported proxy scheme: %s", u.Scheme)
 	}
 	if err != nil {
 		return nil, fmt.Errorf("proxy dial: %w", err)
@@ -230,6 +236,96 @@ func dialTLSViaSOCKS5(addr, proxyURL string, options *imapclient.Options) (*imap
 	return imapclient.New(tlsConn, options), nil
 }
 
+func dialViaSOCKS5(u *url.URL, targetAddr string) (net.Conn, error) {
+	var auth *proxy.Auth
+	if u.User != nil {
+		pw, _ := u.User.Password()
+		auth = &proxy.Auth{User: u.User.Username(), Password: pw}
+	}
+
+	dialer, err := socks5DialerFactory("tcp", u.Host, auth, timeoutContextDialer{timeout: imapDialTimeout})
+	if err != nil {
+		return nil, fmt.Errorf("create socks5 dialer: %w", err)
+	}
+
+	dialCtx, dialCancel := context.WithTimeout(context.Background(), imapDialTimeout)
+	defer dialCancel()
+
+	if cd, ok := dialer.(proxy.ContextDialer); ok {
+		return cd.DialContext(dialCtx, "tcp", targetAddr)
+	}
+	return dialer.Dial("tcp", targetAddr)
+}
+
+// proxyConn wraps a net.Conn, draining buffered data from a bufio.Reader
+// before delegating reads to the underlying connection.
+type proxyConn struct {
+	net.Conn
+	reader io.Reader
+}
+
+func (c *proxyConn) Read(b []byte) (int, error) {
+	return c.reader.Read(b)
+}
+
+// dialToProxy is the function used to establish a TCP connection to the proxy.
+var dialToProxy = func(ctx context.Context, network, addr string) (net.Conn, error) {
+	return (&net.Dialer{Timeout: imapDialTimeout}).DialContext(ctx, network, addr)
+}
+
+func dialViaCONNECT(u *url.URL, targetAddr string) (net.Conn, error) {
+	ctx, cancel := context.WithTimeout(context.Background(), imapDialTimeout)
+	defer cancel()
+
+	conn, err := dialToProxy(ctx, "tcp", u.Host)
+	if err != nil {
+		return nil, fmt.Errorf("dial proxy %s: %w", u.Host, err)
+	}
+
+	if u.Scheme == "https" {
+		tlsConn := tls.Client(conn, proxyTLSConfigForTest(u.Hostname()))
+		if err := tlsConn.HandshakeContext(ctx); err != nil {
+			_ = tlsConn.Close()
+			return nil, fmt.Errorf("tls handshake to proxy: %w", err)
+		}
+		conn = tlsConn
+	}
+
+	_ = conn.SetDeadline(time.Now().Add(imapDialTimeout))
+
+	connectReq := fmt.Sprintf("CONNECT %s HTTP/1.1\r\nHost: %s\r\n", targetAddr, targetAddr)
+	if u.User != nil {
+		username := u.User.Username()
+		password, _ := u.User.Password()
+		creds := base64.StdEncoding.EncodeToString([]byte(username + ":" + password))
+		connectReq += fmt.Sprintf("Proxy-Authorization: Basic %s\r\n", creds)
+	}
+	connectReq += "\r\n"
+
+	if _, err := fmt.Fprint(conn, connectReq); err != nil {
+		_ = conn.Close()
+		return nil, fmt.Errorf("send connect: %w", err)
+	}
+
+	br := bufio.NewReader(conn)
+	resp, err := http.ReadResponse(br, nil)
+	if err != nil {
+		_ = conn.Close()
+		return nil, fmt.Errorf("read connect response: %w", err)
+	}
+	if resp.StatusCode != http.StatusOK {
+		_ = conn.Close()
+		return nil, fmt.Errorf("proxy connect failed: %s", resp.Status)
+	}
+
+	_ = conn.SetDeadline(time.Time{})
+
+	if br.Buffered() > 0 {
+		return &proxyConn{Conn: conn, reader: io.MultiReader(br, conn)}, nil
+	}
+	return conn, nil
+}
+
 // fetchUnseen searches for UNSEEN messages and processes each one.
 func (w *IMAPWatcher) fetchUnseen(c *imapclient.Client) error {
 	criteria := &imap.SearchCriteria{
@@ -283,6 +379,39 @@ func (w *IMAPWatcher) fetchUnseen(c *imapclient.Client) error {
 	return nil
 }
 
+// resolveThreadID determines the thread_id for a message.
+// It tries In-Reply-To first, then falls back to References (last-to-first).
+// Returns the message's own ID if no parent is found.
+func resolveThreadID(messageID, inReplyTo, references string, lookup func(string) (*database.Task, error)) string {
+	if inReplyTo != "" {
+		if parent, err := lookup(inReplyTo); err == nil && parent != nil {
+			return parent.ThreadID
+		}
+	}
+
+	for _, refID := range parseReferences(references) {
+		if parent, err := lookup(refID); err == nil && parent != nil {
+			return parent.ThreadID
+		}
+	}
+
+	return messageID
+}
+
+// parseReferences splits a References header value into individual message IDs,
+// returned in reverse order (most recent first).
+func parseReferences(header string) []string {
+	if header == "" {
+		return nil
+	}
+	raw := strings.Fields(header)
+	// Reverse: last (most recent) first.
+	for i, j := 0, len(raw)-1; i < j; i, j = i+1, j-1 {
+		raw[i], raw[j] = raw[j], raw[i]
+	}
+	return raw
+}
+
 // processMessage handles a single fetched email message.
 func (w *IMAPWatcher) processMessage(c *imapclient.Client, buf *imapclient.FetchMessageBuffer, bodySection *imap.FetchItemBodySection) {
 	env := buf.Envelope
@@ -310,15 +439,7 @@ func (w *IMAPWatcher) processMessage(c *imapclient.Client, buf *imapclient.Fetch
 		return
 	}
 
-	// 2. Whitelist: skip if sender not in allowed list.
-	if !w.isWhitelisted(sender) {
-		log.Info("imap: sender not whitelisted, ignoring", "sender", sender)
-		w.saveIgnored(messageID, sender, subject, "not whitelisted")
-		w.markSeen(c, buf.UID)
-		return
-	}
-
-	// 3. Idempotency: skip if message_id already exists.
+	// 2. Idempotency: skip if message_id already exists.
 	existing, err := database.FindByMessageID(w.db, messageID)
 	if err != nil {
 		log.Error("imap: db lookup failed", "error", err)
@@ -330,9 +451,23 @@ func (w *IMAPWatcher) processMessage(c *imapclient.Client, buf *imapclient.Fetch
 		return
 	}
 
+	// 3. External rules: evaluate operator-configurable policy.
+	ruleResult := w.rules.Evaluate(rules.EmailContext{
+		Sender:    sender,
+		Subject:   subject,
+		MessageID: messageID,
+	})
+	if !ruleResult.Accepted {
+		log.Info("imap: rejected by external rules", "sender", sender, "reason", ruleResult.Reason)
+		w.saveIgnored(messageID, sender, subject, ruleResult.Reason)
+		w.markSeen(c, buf.UID)
+		return
+	}
+
 	// 4. Threading: determine thread_id.
 	threadID := messageID
 	inReplyTo := ""
+	references := ""
 
 	// Parse body to get In-Reply-To / References headers and plain text.
 	bodyPlain := ""
@@ -340,10 +475,12 @@ func (w *IMAPWatcher) processMessage(c *imapclient.Client, buf *imapclient.Fetch
 	if rawBody != nil {
 		mr, err := gomessage.CreateReader(io.NopCloser(strings.NewReader(string(rawBody))))
 		if err == nil {
-			// Extract In-Reply-To header.
 			if irt, err := mr.Header.Text("In-Reply-To"); err == nil && irt != "" {
 				inReplyTo = strings.TrimSpace(irt)
 			}
+			if refs, err := mr.Header.Text("References"); err == nil && refs != "" {
+				references = refs
+			}
 
 			// Read body parts for plain text.
 			for {
@@ -361,23 +498,21 @@ func (w *IMAPWatcher) processMessage(c *imapclient.Client, buf *imapclient.Fetch
 		}
 	}
 
-	if inReplyTo != "" {
-		parent, err := database.FindByMessageID(w.db, inReplyTo)
-		if err == nil && parent != nil {
-			threadID = parent.ThreadID
-		}
-	}
+	threadID = resolveThreadID(messageID, inReplyTo, references, func(mid string) (*database.Task, error) {
+		return database.FindByMessageID(w.db, mid)
+	})
 
 	// 5. Save task with status RECEIVED.
 	taskUUID := uuid.New().String()
 	task := &database.Task{
-		TaskUUID:  taskUUID,
-		ThreadID:  threadID,
-		MessageID: messageID,
-		Sender:    sender,
-		Subject:   subject,
-		BodyPlain: bodyPlain,
-		Status:    database.StatusReceived,
+		TaskUUID:        taskUUID,
+		ThreadID:        threadID,
+		MessageID:       messageID,
+		Sender:          sender,
+		Subject:         subject,
+		BodyPlain:       bodyPlain,
+		DispatchContext: marshalDispatchContext(ruleResult.Metadata),
+		Status:          database.StatusReceived,
 	}
 
 	taskLog := logging.TaskLogger(taskUUID, threadID, messageID)
@@ -420,19 +555,13 @@ func (w *IMAPWatcher) saveIgnored(messageID, sender, subject, reason string) {
 	}
 }
 
-// isWhitelisted checks if the sender is in the allowed list.
-// Empty whitelist means all senders are allowed.
-func (w *IMAPWatcher) isWhitelisted(sender string) bool {
-	if len(w.cfg.WhitelistEmails) == 0 {
-		return true
+// marshalDispatchContext serializes rule metadata to JSON.
+func marshalDispatchContext(metadata map[string]string) string {
+	if len(metadata) == 0 {
+		return ""
 	}
-	lower := strings.ToLower(sender)
-	for _, email := range w.cfg.WhitelistEmails {
-		if strings.ToLower(email) == lower {
-			return true
-		}
-	}
-	return false
+	data, _ := json.Marshal(metadata)
+	return string(data)
 }
 
 // markSeen flags a message as \Seen in IMAP.

internal/mail/imap_test.go

@@ -1,10 +1,20 @@
 package mail
 
 import (
+	"bufio"
 	"context"
+	"crypto/ecdsa"
+	"crypto/elliptic"
+	"crypto/rand"
+	"crypto/tls"
+	"crypto/x509"
+	"encoding/base64"
 	"errors"
+	"fmt"
 	"io"
+	"math/big"
 	"net"
+	"net/http"
 	"strings"
 	"testing"
 	"time"
@@ -73,13 +83,22 @@ func TestConnectAndWatch_UsesProxyDial_WhenProxySet(t *testing.T) {
 	}
 }
 
-func TestDialTLSViaSOCKS5_RejectsNonSocks5Scheme(t *testing.T) {
-	_, err := dialTLSViaSOCKS5("imap.example.com:993", "http://proxy:8080", nil)
+func TestDialTLSViaProxy_RejectsUnsupportedScheme(t *testing.T) {
+	_, err := dialTLSViaProxy("imap.example.com:993", "ftp://proxy:21", nil)
 	if err == nil || !strings.Contains(err.Error(), "unsupported proxy scheme") {
 		t.Fatalf("expected unsupported scheme error, got %v", err)
 	}
 }
 
+func TestDialTLSViaProxy_RejectsMalformedURL(t *testing.T) {
+	_, err := dialTLSViaProxy("imap.example.com:993", "://bad", nil)
+	if err == nil {
+		t.Fatal("expected error for malformed URL")
+	}
+}
+
+// --- SOCKS5 path ---
+
 type fakeContextDialer struct{}
 
 func (fakeContextDialer) Dial(network, address string) (net.Conn, error) {
@@ -93,7 +112,7 @@ func (fakeContextDialer) DialContext(ctx context.Context, network, address strin
 	return nil, errors.New("used dialcontext")
 }
 
-func TestDialTLSViaSOCKS5_UsesDialContextWithTimeout(t *testing.T) {
+func TestDialTLSViaProxy_SOCKS5_UsesDialContextWithTimeout(t *testing.T) {
 	orig := socks5DialerFactory
 	t.Cleanup(func() { socks5DialerFactory = orig })
 
@@ -104,7 +123,7 @@ func TestDialTLSViaSOCKS5_UsesDialContextWithTimeout(t *testing.T) {
 		return fakeContextDialer{}, nil
 	}
 
-	_, err := dialTLSViaSOCKS5("imap.example.com:993", "socks5://127.0.0.1:1080", nil)
+	_, err := dialTLSViaProxy("imap.example.com:993", "socks5://127.0.0.1:1080", nil)
 	if err == nil || !strings.Contains(err.Error(), "proxy dial: used dialcontext") {
 		t.Fatalf("expected DialContext path, got %v", err)
 	}
@@ -143,7 +162,7 @@ func (d contextConnDialer) DialContext(ctx context.Context, network, address str
 	return d.conn, nil
 }
 
-func TestDialTLSViaSOCKS5_SetsDeadlineForTLSHandshake(t *testing.T) {
+func TestDialTLSViaProxy_SOCKS5_SetsDeadlineForTLSHandshake(t *testing.T) {
 	orig := socks5DialerFactory
 	t.Cleanup(func() { socks5DialerFactory = orig })
 
@@ -152,7 +171,7 @@ func TestDialTLSViaSOCKS5_SetsDeadlineForTLSHandshake(t *testing.T) {
 		return contextConnDialer{conn: conn}, nil
 	}
 
-	_, err := dialTLSViaSOCKS5("imap.example.com:993", "socks5://127.0.0.1:1080", nil)
+	_, err := dialTLSViaProxy("imap.example.com:993", "socks5://127.0.0.1:1080", nil)
 	if err == nil || !strings.Contains(err.Error(), "tls handshake") {
 		t.Fatalf("expected tls handshake error, got %v", err)
 	}
@@ -160,3 +179,191 @@ func TestDialTLSViaSOCKS5_SetsDeadlineForTLSHandshake(t *testing.T) {
 		t.Fatal("expected TLS handshake deadline to be set")
 	}
 }
+
+// --- HTTP CONNECT path ---
+
+func startFakeProxy(t *testing.T, handler func(host string) bool) (addr string, cleanup func()) {
+	t.Helper()
+	ln, err := net.Listen("tcp", "127.0.0.1:0")
+	if err != nil {
+		t.Fatalf("listen: %v", err)
+	}
+	done := make(chan struct{})
+	go func() {
+		defer close(done)
+		for {
+			conn, err := ln.Accept()
+			if err != nil {
+				return
+			}
+			go func(c net.Conn) {
+				defer c.Close()
+				br := bufio.NewReader(c)
+				req, err := http.ReadRequest(br)
+				if err != nil {
+					return
+				}
+				if handler(req.URL.Host) {
+					fmt.Fprintf(c, "HTTP/1.1 200 OK\r\n\r\n")
+				} else {
+					fmt.Fprintf(c, "HTTP/1.1 403 Forbidden\r\n\r\n")
+				}
+			}(conn)
+		}
+	}()
+	return ln.Addr().String(), func() {
+		ln.Close()
+		<-done
+	}
+}
+
+func TestDialTLSViaProxy_HTTPConnect_ReachesTLSHandshake(t *testing.T) {
+	addr, cleanup := startFakeProxy(t, func(host string) bool { return true })
+	defer cleanup()
+
+	_, err := dialTLSViaProxy("imap.example.com:993", fmt.Sprintf("http://%s", addr), nil)
+	if err == nil || !strings.Contains(err.Error(), "tls handshake") {
+		t.Fatalf("expected tls handshake error (CONNECT succeeded), got %v", err)
+	}
+}
+
+func TestDialTLSViaProxy_HTTPConnect_RejectsOnProxyFailure(t *testing.T) {
+	addr, cleanup := startFakeProxy(t, func(host string) bool { return false })
+	defer cleanup()
+
+	_, err := dialTLSViaProxy("imap.example.com:993", fmt.Sprintf("http://%s", addr), nil)
+	if err == nil || !strings.Contains(err.Error(), "proxy connect failed") {
+		t.Fatalf("expected proxy connect failed error, got %v", err)
+	}
+}
+
+func TestDialTLSViaProxy_HTTPConnect_SendsProxyAuth(t *testing.T) {
+	var gotAuth string
+	ln, err := net.Listen("tcp", "127.0.0.1:0")
+	if err != nil {
+		t.Fatalf("listen: %v", err)
+	}
+	go func() {
+		defer ln.Close()
+		conn, err := ln.Accept()
+		if err != nil {
+			return
+		}
+		defer conn.Close()
+		br := bufio.NewReader(conn)
+		req, err := http.ReadRequest(br)
+		if err != nil {
+			return
+		}
+		gotAuth = req.Header.Get("Proxy-Authorization")
+		fmt.Fprintf(conn, "HTTP/1.1 200 OK\r\n\r\n")
+	}()
+
+	_, _ = dialTLSViaProxy("imap.example.com:993",
+		fmt.Sprintf("http://testuser:testpass@%s", ln.Addr().String()), nil)
+
+	expected := "Basic " + base64.StdEncoding.EncodeToString([]byte("testuser:testpass"))
+	if gotAuth != expected {
+		t.Fatalf("expected auth %q, got %q", expected, gotAuth)
+	}
+}
+
+func TestDialTLSViaProxy_HTTPConnect_DialError(t *testing.T) {
+	origDial := dialToProxy
+	t.Cleanup(func() { dialToProxy = origDial })
+
+	dialToProxy = func(ctx context.Context, network, addr string) (net.Conn, error) {
+		return nil, errors.New("dial refused")
+	}
+
+	_, err := dialTLSViaProxy("imap.example.com:993", "http://127.0.0.1:1", nil)
+	if err == nil || !strings.Contains(err.Error(), "dial proxy") {
+		t.Fatalf("expected dial proxy error, got %v", err)
+	}
+}
+
+// --- HTTPS CONNECT path ---
+
+func generateTestCert(t *testing.T) tls.Certificate {
+	t.Helper()
+	key, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
+	if err != nil {
+		t.Fatalf("generate key: %v", err)
+	}
+	tmpl := &x509.Certificate{
+		SerialNumber: big.NewInt(1),
+		NotBefore:    time.Now(),
+		NotAfter:     time.Now().Add(time.Hour),
+		KeyUsage:     x509.KeyUsageDigitalSignature,
+		DNSNames:     []string{"127.0.0.1"},
+		IPAddresses:  []net.IP{net.ParseIP("127.0.0.1")},
+	}
+	certDER, err := x509.CreateCertificate(rand.Reader, tmpl, tmpl, &key.PublicKey, key)
+	if err != nil {
+		t.Fatalf("create cert: %v", err)
+	}
+	cert, err := x509.ParseCertificate(certDER)
+	if err != nil {
+		t.Fatalf("parse cert: %v", err)
+	}
+	return tls.Certificate{
+		Certificate: [][]byte{certDER},
+		PrivateKey:  key,
+		Leaf:        cert,
+	}
+}
+
+func TestDialTLSViaProxy_HTTPSConnect_ReachesTLSHandshake(t *testing.T) {
+	cert := generateTestCert(t)
+
+	ln, err := net.Listen("tcp", "127.0.0.1:0")
+	if err != nil {
+		t.Fatalf("listen: %v", err)
+	}
+	tlsLn := tls.NewListener(ln, &tls.Config{
+		Certificates: []tls.Certificate{cert},
+	})
+	done := make(chan struct{})
+	go func() {
+		defer close(done)
+		for {
+			conn, err := tlsLn.Accept()
+			if err != nil {
+				return
+			}
+			go func(c net.Conn) {
+				defer c.Close()
+				br := bufio.NewReader(c)
+				req, err := http.ReadRequest(br)
+				if err != nil {
+					return
+				}
+				if req.URL.Host != "" {
+					fmt.Fprintf(c, "HTTP/1.1 200 OK\r\n\r\n")
+				}
+			}(conn)
+		}
+	}()
+	defer func() {
+		tlsLn.Close()
+		<-done
+	}()
+
+	// Override dialToProxy to return raw TCP, but inject InsecureSkipVerify
+	// so the TLS handshake to proxy succeeds with the test cert.
+	origDial := dialToProxy
+	t.Cleanup(func() { dialToProxy = origDial })
+
+	origProxyTLS := proxyTLSConfigForTest
+	t.Cleanup(func() { proxyTLSConfigForTest = origProxyTLS })
+
+	proxyTLSConfigForTest = func(host string) *tls.Config {
+		return &tls.Config{ServerName: host, InsecureSkipVerify: true}
+	}
+
+	_, err = dialTLSViaProxy("imap.example.com:993",
+		fmt.Sprintf("https://%s", ln.Addr().String()), nil)
+	if err == nil || !strings.Contains(err.Error(), "tls handshake") {
+		t.Fatalf("expected tls handshake error (HTTPS CONNECT succeeded), got %v", err)
+	}
+}

internal/mail/resolve_thread_test.go

@@ -0,0 +1,86 @@
+package mail
+
+import (
+	"errors"
+	"testing"
+
+	"thuanle.me/claw-email-bridge/internal/database"
+)
+
+func TestResolveThreadID_NoHeaders_ReturnsMessageID(t *testing.T) {
+	got := resolveThreadID("msg-1", "", "", nilLookup)
+	if got != "msg-1" {
+		t.Fatalf("expected msg-1, got %s", got)
+	}
+}
+
+func TestResolveThreadID_InReplyToMatches_ReturnsParentThreadID(t *testing.T) {
+	lookup := func(messageID string) (*database.Task, error) {
+		if messageID == "<parent>" {
+			return &database.Task{ThreadID: "thread-abc"}, nil
+		}
+		return nil, nil
+	}
+	got := resolveThreadID("msg-2", "<parent>", "", lookup)
+	if got != "thread-abc" {
+		t.Fatalf("expected thread-abc, got %s", got)
+	}
+}
+
+func TestResolveThreadID_ReferencesOnly_LastIDMatches(t *testing.T) {
+	lookup := func(messageID string) (*database.Task, error) {
+		if messageID == "<C>" {
+			return &database.Task{ThreadID: "thread-c"}, nil
+		}
+		return nil, nil
+	}
+	got := resolveThreadID("msg-3", "", "<A> <B> <C>", lookup)
+	if got != "thread-c" {
+		t.Fatalf("expected thread-c, got %s", got)
+	}
+}
+
+func TestResolveThreadID_ReferencesOnly_MiddleIDMatches(t *testing.T) {
+	lookup := func(messageID string) (*database.Task, error) {
+		if messageID == "<B>" {
+			return &database.Task{ThreadID: "thread-b"}, nil
+		}
+		return nil, nil
+	}
+	got := resolveThreadID("msg-4", "", "<A> <B> <C>", lookup)
+	if got != "thread-b" {
+		t.Fatalf("expected thread-b, got %s", got)
+	}
+}
+
+func TestResolveThreadID_ReferencesOnly_NoMatch_ReturnsMessageID(t *testing.T) {
+	got := resolveThreadID("msg-5", "", "<X> <Y>", nilLookup)
+	if got != "msg-5" {
+		t.Fatalf("expected msg-5, got %s", got)
+	}
+}
+
+func TestResolveThreadID_InReplyToMiss_ReferencesMatches(t *testing.T) {
+	lookup := func(messageID string) (*database.Task, error) {
+		if messageID == "<ref>" {
+			return &database.Task{ThreadID: "thread-ref"}, nil
+		}
+		return nil, nil
+	}
+	got := resolveThreadID("msg-6", "<unknown>", "<ref>", lookup)
+	if got != "thread-ref" {
+		t.Fatalf("expected thread-ref, got %s", got)
+	}
+}
+
+func TestResolveThreadID_LookupError_ReturnsMessageID(t *testing.T) {
+	lookup := func(messageID string) (*database.Task, error) {
+		return nil, errors.New("db error")
+	}
+	got := resolveThreadID("msg-7", "<parent>", "", lookup)
+	if got != "msg-7" {
+		t.Fatalf("expected msg-7, got %s", got)
+	}
+}
+
+var nilLookup = func(string) (*database.Task, error) { return nil, nil }

internal/rules/blocklist.go

@@ -0,0 +1,26 @@
+package rules
+
+import "strings"
+
+// BlocklistRule rejects emails from configured senders.
+type BlocklistRule struct {
+	emails map[string]bool
+}
+
+// NewBlocklistRule creates a blocklist rule from a list of email addresses.
+func NewBlocklistRule(emails []string) *BlocklistRule {
+	m := make(map[string]bool, len(emails))
+	for _, e := range emails {
+		m[strings.ToLower(e)] = true
+	}
+	return &BlocklistRule{emails: m}
+}
+
+func (r *BlocklistRule) Name() string { return "blocklist" }
+
+func (r *BlocklistRule) Evaluate(ctx EmailContext) RuleResult {
+	if r.emails[strings.ToLower(ctx.Sender)] {
+		return Reject("blocked sender")
+	}
+	return Accept()
+}

internal/rules/blocklist_test.go

@@ -0,0 +1,35 @@
+package rules
+
+import "testing"
+
+func TestBlocklistRule_SenderBlocked_Rejects(t *testing.T) {
+	r := NewBlocklistRule([]string{"spam@test.com"})
+	result := r.Evaluate(EmailContext{Sender: "spam@test.com"})
+	if result.Accepted {
+		t.Fatal("expected reject for blocked sender")
+	}
+}
+
+func TestBlocklistRule_SenderNotBlocked_Accepts(t *testing.T) {
+	r := NewBlocklistRule([]string{"spam@test.com"})
+	result := r.Evaluate(EmailContext{Sender: "ok@test.com"})
+	if !result.Accepted {
+		t.Fatal("expected accept for non-blocked sender")
+	}
+}
+
+func TestBlocklistRule_EmptyList_Accepts(t *testing.T) {
+	r := NewBlocklistRule(nil)
+	result := r.Evaluate(EmailContext{Sender: "anyone@test.com"})
+	if !result.Accepted {
+		t.Fatal("empty blocklist should accept all")
+	}
+}
+
+func TestBlocklistRule_CaseInsensitive(t *testing.T) {
+	r := NewBlocklistRule([]string{"Spam@Test.com"})
+	result := r.Evaluate(EmailContext{Sender: "spam@test.com"})
+	if result.Accepted {
+		t.Fatal("blocklist should be case-insensitive")
+	}
+}

internal/rules/pipeline.go

@@ -0,0 +1,30 @@
+package rules
+
+// Pipeline evaluates a sequence of rules, stopping on first rejection.
+type Pipeline struct {
+	rules []Rule
+}
+
+// NewPipeline creates a rules pipeline from the given rules.
+func NewPipeline(rules []Rule) *Pipeline {
+	return &Pipeline{rules: rules}
+}
+
+// Evaluate runs all rules in order. Returns the first rejection or a merged accept.
+func (p *Pipeline) Evaluate(ctx EmailContext) RuleResult {
+	merged := Accept()
+	for _, r := range p.rules {
+		result := r.Evaluate(ctx)
+		if !result.Accepted {
+			return result
+		}
+		for k, v := range result.Metadata {
+			if merged.Metadata == nil {
+				merged.Metadata = map[string]string{k: v}
+			} else {
+				merged.Metadata[k] = v
+			}
+		}
+	}
+	return merged
+}

internal/rules/pipeline_test.go

@@ -0,0 +1,106 @@
+package rules
+
+import (
+	"testing"
+)
+
+func TestPipeline_NoRules_Accepts(t *testing.T) {
+	p := NewPipeline(nil)
+	result := p.Evaluate(EmailContext{Sender: "anyone@test.com"})
+	if !result.Accepted {
+		t.Fatal("expected accept with no rules")
+	}
+}
+
+func TestPipeline_AllRulesAccept_Accepts(t *testing.T) {
+	allowAll := &stubRule{name: "allow-all", result: Accept()}
+	p := NewPipeline([]Rule{allowAll})
+	result := p.Evaluate(EmailContext{Sender: "a@b.com"})
+	if !result.Accepted {
+		t.Fatal("expected accept")
+	}
+}
+
+func TestPipeline_OneRuleRejects_Rejects(t *testing.T) {
+	rejector := &stubRule{name: "rejector", result: Reject("nope")}
+	p := NewPipeline([]Rule{rejector})
+	result := p.Evaluate(EmailContext{Sender: "a@b.com"})
+	if result.Accepted {
+		t.Fatal("expected reject")
+	}
+	if result.Reason != "nope" {
+		t.Fatalf("expected reason 'nope', got %s", result.Reason)
+	}
+}
+
+func TestPipeline_MergesMetadata(t *testing.T) {
+	r1 := &stubRule{name: "r1", result: Accept().WithMetadata("a", "1")}
+	r2 := &stubRule{name: "r2", result: Accept().WithMetadata("b", "2")}
+	p := NewPipeline([]Rule{r1, r2})
+	result := p.Evaluate(EmailContext{})
+	if result.Metadata["a"] != "1" {
+		t.Fatal("missing metadata a")
+	}
+	if result.Metadata["b"] != "2" {
+		t.Fatal("missing metadata b")
+	}
+}
+
+func TestPipeline_StopsOnFirstReject(t *testing.T) {
+	rejector := &stubRule{name: "rejector", result: Reject("stop")}
+	neverCalled := &trackingRule{}
+	p := NewPipeline([]Rule{rejector, neverCalled})
+	p.Evaluate(EmailContext{})
+	if neverCalled.called {
+		t.Fatal("expected second rule not to be called after rejection")
+	}
+}
+
+func TestWhitelistRule_EmptyList_Accepts(t *testing.T) {
+	r := NewWhitelistRule(nil)
+	result := r.Evaluate(EmailContext{Sender: "anyone@test.com"})
+	if !result.Accepted {
+		t.Fatal("empty whitelist should accept all")
+	}
+}
+
+func TestWhitelistRule_SenderInList_Accepts(t *testing.T) {
+	r := NewWhitelistRule([]string{"allowed@test.com"})
+	result := r.Evaluate(EmailContext{Sender: "allowed@test.com"})
+	if !result.Accepted {
+		t.Fatal("expected accept for whitelisted sender")
+	}
+}
+
+func TestWhitelistRule_SenderNotInList_Rejects(t *testing.T) {
+	r := NewWhitelistRule([]string{"allowed@test.com"})
+	result := r.Evaluate(EmailContext{Sender: "unknown@test.com"})
+	if result.Accepted {
+		t.Fatal("expected reject for non-whitelisted sender")
+	}
+}
+
+func TestWhitelistRule_CaseInsensitive(t *testing.T) {
+	r := NewWhitelistRule([]string{"Allowed@Test.com"})
+	result := r.Evaluate(EmailContext{Sender: "allowed@test.com"})
+	if !result.Accepted {
+		t.Fatal("whitelist should be case-insensitive")
+	}
+}
+
+// stubs
+
+type stubRule struct {
+	name   string
+	result RuleResult
+}
+
+func (s *stubRule) Name() string                       { return s.name }
+func (s *stubRule) Evaluate(_ EmailContext) RuleResult { return s.result }
+
+type trackingRule struct {
+	called bool
+}
+
+func (t *trackingRule) Name() string                       { return "tracker" }
+func (t *trackingRule) Evaluate(_ EmailContext) RuleResult { t.called = true; return Accept() }

internal/rules/types.go

@@ -0,0 +1,41 @@
+package rules
+
+// EmailContext contains the data available to rules for evaluation.
+type EmailContext struct {
+	Sender    string
+	Subject   string
+	MessageID string
+}
+
+// RuleResult is the output of evaluating a single rule.
+type RuleResult struct {
+	Accepted bool
+	Reason   string
+	Metadata map[string]string
+}
+
+// Accept returns a passing RuleResult.
+func Accept() RuleResult {
+	return RuleResult{Accepted: true}
+}
+
+// Reject returns a failing RuleResult with a reason.
+func Reject(reason string) RuleResult {
+	return RuleResult{Accepted: false, Reason: reason}
+}
+
+// WithMetadata attaches a key-value pair to the result.
+func (r RuleResult) WithMetadata(key, value string) RuleResult {
+	if r.Metadata == nil {
+		r.Metadata = map[string]string{key: value}
+	} else {
+		r.Metadata[key] = value
+	}
+	return r
+}
+
+// Rule evaluates an email against operator-configurable policy.
+type Rule interface {
+	Name() string
+	Evaluate(ctx EmailContext) RuleResult
+}

internal/rules/types_test.go

@@ -0,0 +1,42 @@
+package rules
+
+import (
+	"testing"
+)
+
+func TestRuleResult_Accept(t *testing.T) {
+	r := Accept()
+	if !r.Accepted {
+		t.Fatal("expected Accepted=true")
+	}
+	if r.Reason != "" {
+		t.Fatal("expected no reason for accept")
+	}
+}
+
+func TestRuleResult_Reject(t *testing.T) {
+	r := Reject("blocked sender")
+	if r.Accepted {
+		t.Fatal("expected Accepted=false")
+	}
+	if r.Reason != "blocked sender" {
+		t.Fatalf("expected reason 'blocked sender', got %s", r.Reason)
+	}
+}
+
+func TestRuleResult_WithMetadata(t *testing.T) {
+	r := Accept().WithMetadata("routing", "high-priority")
+	if r.Metadata["routing"] != "high-priority" {
+		t.Fatalf("expected routing=high-priority, got %s", r.Metadata["routing"])
+	}
+}
+
+func TestEmailContext_Fields(t *testing.T) {
+	ctx := EmailContext{
+		Sender:  "user@example.com",
+		Subject: "Test",
+	}
+	if ctx.Sender != "user@example.com" {
+		t.Fatal("sender mismatch")
+	}
+}

internal/rules/whitelist.go

@@ -0,0 +1,30 @@
+package rules
+
+import "strings"
+
+// WhitelistRule accepts emails from configured senders.
+// An empty list accepts all senders.
+type WhitelistRule struct {
+	emails map[string]bool
+}
+
+// NewWhitelistRule creates a whitelist rule from a list of email addresses.
+func NewWhitelistRule(emails []string) *WhitelistRule {
+	m := make(map[string]bool, len(emails))
+	for _, e := range emails {
+		m[strings.ToLower(e)] = true
+	}
+	return &WhitelistRule{emails: m}
+}
+
+func (r *WhitelistRule) Name() string { return "whitelist" }
+
+func (r *WhitelistRule) Evaluate(ctx EmailContext) RuleResult {
+	if len(r.emails) == 0 {
+		return Accept()
+	}
+	if r.emails[strings.ToLower(ctx.Sender)] {
+		return Accept()
+	}
+	return Reject("not whitelisted")
+}

requirements.md

@@ -47,6 +47,7 @@ Bảng `tasks`:
 - `status` TEXT
 - `attempt_openclaw` INTEGER DEFAULT 0
 - `attempt_smtp` INTEGER DEFAULT 0
+- `dispatch_context` TEXT (JSON metadata từ external rules pipeline, truyền qua OpenClaw dispatch)
 - `last_error` TEXT NULL
 - `next_retry_at` DATETIME NULL (chỉ cần nếu retry xử lý theo scheduler/worker)
 - `created_at` DATETIME
@@ -68,12 +69,12 @@ Danh sách trạng thái hợp lệ:
 - Nếu có `IMAP_PROXY_URL` thì kết nối IMAP qua proxy; nếu không thì direct.
 - Khi có email mới:
   1. Anti-loop: nếu `sender` trùng email hệ thống -> `IGNORED`.
-  2. Whitelist: nếu không nằm trong danh sách cho phép -> `IGNORED`.
-  3. Idempotency: nếu `message_id` đã tồn tại -> bỏ qua.
+  2. Idempotency: nếu `message_id` đã tồn tại -> bỏ qua.
+  3. External rules: chạy pipeline các rule configurable (whitelist, blocklist, v.v.). Nếu reject -> `IGNORED`. Rule metadata được lưu vào `dispatch_context`.
   4. Threading:
      - Nếu có `In-Reply-To`/`References`: tìm email cha trong DB, kế thừa `thread_id`.
      - Nếu không có: `thread_id = message_id`.
-  5. Lưu DB với `status = RECEIVED`.
+  5. Lưu DB với `status = RECEIVED`, bao gồm `dispatch_context` từ rules.
 
 4.2 Dispatch (OpenClaw)
 - Lấy nội dung email hiện tại và history của cùng `thread_id` (chỉ các bản ghi `COMPLETED`).
@@ -84,6 +85,7 @@ Danh sách trạng thái hợp lệ:
   - `history` (optional)
   - `callback_url`
   - `metadata.task_uuid`
+  - `metadata.*` từ `dispatch_context` (rule-derived context)
 - Cập nhật `status = AI_PROCESSING`.
 
 Retry OpenClaw:
@@ -146,6 +148,7 @@ Required:
 - `OPENCLAW_URL`, `OPENCLAW_API_KEY` (nếu dùng)
 - `BRIDGE_CALLBACK_TOKEN`
 - `WHITELIST_EMAILS` (comma-separated)
+- `BLOCKLIST_EMAILS` (comma-separated, optional)
 - `SYSTEM_EMAIL`
 - `IMAP_PROXY_URL` (optional)
 
@@ -166,7 +169,7 @@ Ví dụ:
 3. OpenClaw/SMTP retry đúng thứ tự 1s -> 5s -> 15s.
 4. Hết retry thì `FAILED` và có `last_error`.
 5. Callback sai token trả 401 và không đổi state task.
-6. Anti-loop + whitelist + idempotency hoạt động đúng.
+6. Anti-loop + idempotency + external rules hoạt động đúng thứ tự.
 
 10) Test matrix tối thiểu
 
@@ -175,4 +178,7 @@ Ví dụ:
 3. SMTP fail hết retry -> `FAILED`, có `last_error`.
 4. Callback sai token -> 401, state không đổi.
 5. Duplicate `message_id` -> không tạo task mới.
-6. Sender là chính hệ thống hoặc ngoài whitelist -> `IGNORED`.
+6. Sender là chính hệ thống -> `IGNORED`.
+7. Sender ngoài whitelist -> `IGNORED` bởi external rules.
+8. Sender trong blocklist -> `IGNORED` bởi external rules.
+9. Rule metadata được truyền qua dispatch context đến OpenClaw.