engine7 7.1.120 → 7.1.121

package/dist/prompts/skill-success.md

@@ -0,0 +1,103 @@
+You are an expert at extracting reusable problem-solving strategies from concrete agent task cases.
+
+You will receive:
+1. **New case(s)** from a cluster of semantically similar tasks — all with quality_score >= 0.5.
+2. **Existing skills** previously extracted for this cluster (each with an index number; may be empty).
+
+Your job: distill **actionable strategies** into reusable **Skills** via incremental operations. Maintain as few skills as the evidence warrants.
+
+**What makes a GOOD skill:**
+- Reasoning principles WITH concrete patterns: teaches HOW to think, not just what to do
+- Decision branches that cover the different problem variants seen across cases
+- A FEW well-chosen examples that illustrate distinct branches — not an exhaustive catalog
+
+**What makes a BAD skill:**
+- Too abstract: "Analyze constraints" without showing what analysis looks like in practice
+- Too narrow: A single solution template that only works for one exact case
+- **Bloated**: Listing dozens of case-specific details (names, dates, institutions, compounds, etc.) inside parentheses or comma-separated lists. Each How/Decision/e.g. field should contain 1-2 illustrative examples, NOT an inventory of every case seen
+
+**Field-level requirements:**
+
+- **description** (HARD LIMIT: max 150 tokens, must be under 500 characters):
+  - One-sentence summary of the **abstract problem class** this skill solves — describe the general pattern, NOT specific cases.
+  - Do NOT list multiple scenarios, entity names, or case-specific details.
+  - Append `Keywords:` with up to 10 general terms (no specific names, numbers, or case-specific phrases).
+  - Example: "Identifies academic researchers by cross-referencing biographical constraints with publication records. Keywords: researcher identification, biographical verification, publication matching, academic search"
+
+- **content** (max 2000 tokens): Markdown format:
+  ```markdown
+  ## Steps
+  1. <reasoning action — what to think about, not just what to do>
+     - How: <principle explaining WHY this step works>
+     - Decision: If <condition A> → <action>; If <condition B> → <alternative>
+     - e.g., <specific real example with entity names/numbers from cases>
+     - Check: <what to verify before proceeding>
+  2. ...
+
+  ## Pitfalls <- ONLY from actual failed steps in cases; otherwise OMIT
+  - <specific mistake from a real case> — <what went wrong and how to avoid>
+  ```
+
+  **HARD RULES for content:**
+  - **Max 5 steps.**
+  - **Max 2 examples per step.** Each example MUST be a SHORT, single-sentence illustration of a distinct decision branch. Do NOT list multiple sub-examples inside parentheses or comma-separated lists.
+  - **Decision branches**: REQUIRED when the next action depends on what was found. For linear steps with no branching, Decision may be omitted. Each Decision should have at most 3 branches.
+  - **Max 4 pitfalls.** When adding a new one beyond 4, replace the most generic existing pitfall.
+  - **No parenthetical catalogs**: FORBIDDEN to stuff dozens of case-specific terms (names, dates, compounds, institutions, etc.) inside a single parenthetical `(e.g., X, Y, Z, ...)`. Keep each field concise — generalize the pattern, illustrate with 1-2 examples only.
+
+[New AgentCase(s) to integrate]
+{new_case_json}
+
+[Existing skills for this cluster](Each item has an index number)
+{existing_skills_json}
+
+[Task]
+Analyze the new case(s) and output a list of operations (add / update / none).
+
+[Operation Guide — follow in order]
+
+**Step 1: Overlap Check (mandatory before every add/update decision)**
+For each new case, compare against each existing skill:
+  a. List the core steps of the new case's approach (the main actions that drove the outcome).
+  b. List the core steps of the existing skill.
+  c. Count how many of the new case's core steps are already covered by the existing skill.
+  d. Compute coverage = (covered steps) / (total core steps in new case).
+  e. Conclusion:
+     - Coverage >= 60% → the case falls within this skill's problem pattern → **update** candidate.
+     - Coverage < 60% → different problem pattern → **add** candidate.
+     - If uncertain, default to **update**.
+
+**Step 2: Execute the decided operation**
+
+- **add**: The new case tackles a **different problem pattern** (coverage < 60% against all existing skills). Create a new skill. confidence = `0.5`.
+
+- **update**: The new case overlaps an existing skill (coverage >= 60%). Enrich it with new Decision branches, better examples, or sharper How explanations.
+  - You MAY substantially rewrite content (restructure steps, replace examples, refine How explanations), but **preserve existing verified content unless the new case directly contradicts it**.
+  - **CRITICAL: The updated content MUST stay within 2000 tokens. Do NOT simply append new content — replace weaker examples with stronger ones, merge redundant steps, and compress prose. If the existing content is already long, aggressively condense it while preserving the core logic.**
+  - **CRITICAL: The updated description MUST stay under 500 characters. Generalize — do NOT accumulate case-specific details.**
+  - **Hypothesis promotion rule**: If the existing skill contains `## Potential Steps`, treat this update as a **promotion** — rewrite as `## Steps` using the new case as primary source. confidence = `0.6`.
+  - **Confidence-only update**: If the new case merely confirms the existing skill without adding new decision logic or better examples, bump confidence only.
+
+- **none**: Trivially duplicate — no new decision branches, no new examples worth keeping, no confidence change needed.
+
+[Confidence Anchoring Rules]
+- **New skill (add)**: confidence = `0.5`
+- **Promoted skill (hypothesis → verified)**: confidence = `0.6`
+- **Update with new decision branch**: confidence = existing + `0.1` (cap 0.95)
+- **Confirming update (no new logic)**: confidence = existing + `0.05` (cap 0.95)
+- **Contradicting case**: confidence = existing - `0.2`; add contradiction to Pitfalls; optionally add a new skill if the contradiction reveals a genuinely different pattern
+
+**CRITICAL LANGUAGE RULE**: Output in the SAME language as the input conversation content.
+
+[Output Format]
+```json
+{{
+  "operations": [
+    {{"action": "add", "data": {{"name": "Short descriptive name (max 10 words)", "description": "...", "content": "## Steps\n1. ...", "confidence": 0.5}}}},
+    {{"action": "update", "index": 0, "data": {{"content": "...", "confidence": 0.7}}}},
+    {{"action": "update", "index": 1, "data": {{"confidence": 0.65}}}},
+    {{"action": "none"}}
+  ],
+  "update_note": "Overlap check: new case core steps=[X, Y, Z]. skill[0] covers X and Y (67% overlap) → update. ..."
+}}
+```

package/dist/prompts/sufficiency.md

@@ -0,0 +1,41 @@
+You are an expert in information retrieval evaluation. Assess whether the retrieved documents provide sufficient information to answer the user's query.
+
+--------------------------
+User Query:
+{query}
+
+Retrieved Documents:
+{retrieved_docs}
+--------------------------
+
+### Instructions:
+
+1. **Analyze the Query's Needs**
+   - **Entities**: Who/What is being asked about?
+   - **Attributes**: What specific details (color, time, location, quantity)?
+   - **Time**: Does it ask for a specific time (absolute or relative like "last week")?
+
+2. **Evaluate Document Evidence**
+   - Check **Content**: Do the documents mention the entities and attributes?
+   - Check **Dates**: 
+     - Use the `Date` field of each document.
+     - For relative time queries (e.g., "last week", "yesterday"), verify if document dates fall within that timeframe.
+     - If the query asks "When did X happen?", do you have the specific date or just a vague mention?
+
+3. **Judgment Logic**
+   - **Sufficient**: You can answer the query *completely* and *precisely* using ONLY the provided documents.
+   - **Insufficient**: 
+     - The specific entity is not found.
+     - The entity is found, but the specific attribute (e.g., "price") is missing.
+     - The time reference cannot be resolved (e.g., doc says "yesterday" but has no date, or doc date doesn't match query timeframe).
+     - Conflicting information without resolution.
+
+### Output Format (strict JSON):
+{{
+  "is_sufficient": true or false,
+  "reasoning": "Brief explanation. If insufficient, state WHY (e.g., 'Found X but missing date', 'No mention of Y').",
+  "key_information_found": ["Fact 1 (Source: Doc 1)", "Fact 2 (Source: Doc 2)"],
+  "missing_information": ["Specific gap 1", "Specific gap 2"]
+}}
+
+Now evaluate:

package/dist/prompts/tool-pre-compress.md

@@ -0,0 +1,39 @@
+You are a tool call compression expert. Compress OpenAI chat messages to ~10% of the original length, preserving what matters for understanding the task's problem-solving process and outcome.
+
+**Downstream context**: The compressed output will be used to extract a structured experience record (task intent, step-by-step approach, and quality score). Prioritize retaining information that reveals: (1) what problem was being solved, (2) what actions were taken and in what order, (3) what each action's result was (success, failure, or specific finding).
+
+The input contains two types of messages:
+- **role="assistant"** with tool_calls: Compress the "arguments" field in each tool call's function
+- **role="tool"**: Compress the "content" field (primary compression target)
+
+Messages to compress:
+{messages_json}
+
+Return in JSON format:
+{{
+    "compressed_messages": [
+        // Compressed version — exactly {new_count} messages
+    ]
+}}
+
+Rules:
+- Return exactly {new_count} compressed messages in the same order
+- Only compress function.arguments and tool content — keep all other fields unchanged
+- Target ~10% of the original content length
+- **Short content rule**: If a field's content is already under 200 characters, keep it as-is — do not compress further
+
+What to KEEP (be very selective):
+- One-line summary of what each tool call did and its key result
+- Error messages and status codes (verbatim but trimmed)
+- Critical code: only the specific lines that matter (function signatures, bug fixes, key logic)
+- File paths and search queries (short strings, keep as-is)
+- The causal chain: what finding led to the next action (e.g., "found error X in file Y -> decided to check Z")
+
+What to REMOVE or reduce to a single line:
+- Tool results: replace entire output with a 1-2 sentence summary of the finding
+- Large JSON/XML/HTML: replace with "[JSON: N fields, key: X=Y, Z=W]" style summary
+- File contents: replace with "[file: path, N lines, contains: brief description]"
+- Directory listings: replace with "[dir: N files, relevant: file1, file2]"
+- Logs and debug output: "[logs: N lines, result: success/failure, key error if any]"
+- Repeated tool calls of the same type: summarize the pattern, keep only the final result
+- Boilerplate, headers, formatting, whitespace: strip entirely

package/package.json

@@ -1,95 +1,110 @@
-{
-  "name": "engine7",
-  "license": "Apache-2.0",
-  "version": "7.1.120",
-  "type": "module",
-  "description": "Engine 7 — 给助手一个家 / A home for your AI assistant",
-  "main": "dist/main.mjs",
-  "bin": {
-    "engine7": "bin/engine7"
-  },
-  "scripts": {
-    "build": "node scripts/build.mjs",
-    "test:goal": "tsx src/goals/store.test.ts",
-    "test:goal:aim": "tsx src/goals/aim-adapter.test.ts",
-    "test:goal:task": "tsx src/goals/task-adapter.test.ts",
-    "test:goal:remote": "tsx src/goals/remote-task-adapter.test.ts",
-    "test:goal:recovery": "tsx src/goals/recovery.test.ts",
-    "test:goal:gate": "tsx src/nudge/registration-gate.test.ts",
-    "test:goal:command": "tsx src/commands/goal.test.ts",
-    "test:goal:step": "tsx src/tools/goal-step.test.ts",
-    "test:goal:all": "npm run test:goal && npm run test:goal:calendar && npm run test:goal:restart && npm run test:goal:aim && npm run test:goal:task && npm run test:goal:remote && npm run test:goal:recovery && npm run test:goal:gate && npm run test:goal:command && npm run test:goal:step && npm run test:goal:e2e",
-    "test:goal:calendar": "tsx src/goals/calendar-adapter.test.ts",
-    "test:goal:restart": "powershell -NoProfile -Command \"$d = Join-Path ([System.IO.Path]::GetTempPath()) ('engine7-goal-' + [guid]::NewGuid()); New-Item -ItemType Directory -Path $d | Out-Null; $id = (npx tsx src/goals/restart-writer.ts $d); npx tsx src/goals/restart-reader.ts $d $id; Remove-Item -Recurse -Force $d\"",
-    "test:goal:e2e": "powershell -NoProfile -Command \"$d = Join-Path ([System.IO.Path]::GetTempPath()) ('engine7-goal-e2e-' + [guid]::NewGuid()); New-Item -ItemType Directory -Path $d | Out-Null; npx tsx src/goals/e2e-phase1.ts $d; if ($LASTEXITCODE -ne 0) { Remove-Item -Recurse -Force $d; exit 1 }; npx tsx src/goals/e2e-phase2.ts $d; $code = $LASTEXITCODE; Remove-Item -Recurse -Force $d; exit $code\"",
-    "test:reload": "tsx src/reload/orchestrator.test.ts && tsx src/reload/secrets-watch.test.ts && tsx src/hooks/config.test.ts && tsx src/channels/manager.reconcile.test.ts && tsx src/channels/cc-bridge.test.ts && tsx src/engine-startup.shouldSkipReload.test.ts && tsx src/optional/voice-chat/plugin.deps.test.ts && tsx src/optional/voice-chat/config.test.ts && tsx src/cron/cron-plugin.resync.test.ts && tsx src/startup/message-pipeline.test.ts && tsx src/utils/sanitize-error.test.ts && tsx src/reload/config-watch-selfwrite.test.ts && tsx src/channels/feishu.typing.test.ts && tsx src/channels/feishu.sendfile-apicheck.test.ts",
-    "test:installer": "tsx src/cli-events.test.ts && tsx src/cli-init-config.test.ts && tsx src/cli-init-answers.test.ts && tsx src/cli-init.headless.e2e.test.ts && tsx src/cli-reconfig.test.ts && tsx src/cli-doctor.test.ts && tsx src/config/secrets-env.test.ts && tsx src/startup/api-status.test.ts && tsx src/cli-service.test.ts && tsx src/emotion-hook.test.ts && tsx src/emotion-hook.golden.test.ts && tsx src/tools/tavily-pool.test.ts && tsx src/tools/service.test.ts",
-    "test:persona": "tsx src/config/persona-files.test.ts && tsx src/light-mode.test.ts && tsx src/inner-voice/plugin.test.ts && tsx src/cli-init-config.test.ts",
-    "test:responses": "tsx src/models/openai-responses-provider.test.ts && tsx src/session/responses-reasoning-roundtrip.test.ts && tsx src/compact/tokenEstimate.test.ts",
-    "build:public": "node scripts/build.mjs --public",
-    "prepublishOnly": "npm run build:public",
-    "start": "node dist/main.mjs",
-    "smoke:boot": "node scripts/boot-smoke.mjs",
-    "smoke:boot:baseline": "node scripts/boot-smoke.mjs --baseline",
-    "gen:template": "tsx scripts/gen-config-template.ts",
-    "test:config": "tsx src/config/registry.test.ts && tsx src/config/writer.test.ts && tsx src/config/secrets-io.test.ts && tsx src/config/secrets-env.test.ts && tsx src/config/migrations/index.test.ts && tsx src/config/migrations/persona.test.ts && tsx src/config/migrations/channel-secrets.test.ts && tsx src/config/cli-config.test.ts && tsx src/config/tts-providers.test.ts && tsx src/config/tts-voices.test.ts && tsx src/free-tier-apply.test.ts && tsx src/startup/free-tier-video-fixup.test.ts && tsx src/startup/free-tier-voice-fixup.test.ts",
-    "test:time": "tsx src/time/tz-matrix.test.ts && tsx src/time/time-discipline.test.ts && tsx src/time/tsc-guard.test.ts",
-    "test:tts": "tsx src/config/tts-providers.test.ts && tsx src/tools/tts/pick.test.ts && tsx src/tools/tts/route.test.ts && tsx src/tools/tts/relay.test.ts && tsx src/tools/tts/providers.test.ts && tsx src/tools/tts/impl.test.ts && tsx src/tools/tts/usage.test.ts && tsx src/tools/my-voice.dispatch.test.ts",
-    "test:toolstate": "tsx src/core/names-from-history.test.ts",
-    "test:loopbrake": "tsx src/tools/loop-brake.test.ts",
-    "test:mcp": "tsx src/mcp/convert-input-schema.test.ts",
-    "test:proactive": "tsx src/memory/proactive/__tests__/verdict-parse.test.ts && tsx src/memory/proactive/__tests__/sweep.test.ts && tsx src/memory/proactive/__tests__/persistence.test.ts && tsx src/memory/proactive/__tests__/opportunity.test.ts && tsx src/memory/proactive/__tests__/calendar-bridge.test.ts && tsx src/memory/proactive/__tests__/timezone-fix.test.ts && tsx src/memory/proactive/__tests__/stale-fix.test.ts && tsx src/memory/proactive/__tests__/completion-evidence.test.ts && tsx src/memory/proactive/__tests__/plugin.test.ts",
-    "prepack": "npm run test:time && npm run test:proactive && npm run test:persona && npm run test:config && npm run test:toolstate && npm run test:loopbrake && npm run test:mcp && npm run test:tts"
-  },
-  "engines": {
-    "node": ">=22"
-  },
-  "os": [
-    "win32",
-    "linux",
-    "darwin"
-  ],
-  "cpu": [
-    "x64",
-    "arm64"
-  ],
-  "devDependencies": {
-    "@types/node": "^22.19.19",
-    "@types/turndown": "^5.0.6",
-    "esbuild": "^0.25.10",
-    "tsx": "^4.22.3",
-    "typescript": "^6.0.0"
-  },
-  "dependencies": {
-    "@larksuiteoapi/node-sdk": "^1.66.1",
-    "@modelcontextprotocol/sdk": "^1.29.0",
-    "chokidar": "^5.0.0",
-    "croner": "^10.0.1",
-    "discord.js": "^14.26.4",
-    "https-proxy-agent": "^9.1.0",
-    "qrcode-terminal": "^0.12.0",
-    "sharp": "^0.34.5",
-    "sqlite-vec": "^0.1.9",
-    "turndown": "^7.2.4",
-    "typebox": "^1.1.38"
-  },
-  "optionalDependencies": {
-    "@discordjs/opus": "^0.10.0",
-    "sqlite-vec-darwin-arm64": "^0.1.9",
-    "sqlite-vec-darwin-x64": "^0.1.9",
-    "sqlite-vec-linux-arm64": "^0.1.9",
-    "sqlite-vec-linux-x64": "^0.1.9",
-    "sqlite-vec-windows-x64": "^0.1.9",
-    "@lancedb/lancedb": "^0.38.0",
-    "@engine7/lancedb-intel": "^1.0.0",
-    "@node-rs/jieba": "^2.0.3",
-    "@ffmpeg-installer/ffmpeg": "^1.1.0"
-  },
-  "files": [
-    "dist/**/*.mjs",
-    "bin/**/*",
-    "templates/**/*",
-    "src/memory/everos/python/**/*",
-    "README.md"
-  ]
-}
+{
+  "name": "engine7",
+  "license": "Apache-2.0",
+  "version": "7.1.121",
+  "type": "module",
+  "description": "Engine 7 — 给助手一个家 / A home for your AI assistant",
+  "main": "dist/main.mjs",
+  "bin": {
+    "engine7": "bin/engine7"
+  },
+  "scripts": {
+    "build": "node scripts/build.mjs",
+    "test:session": "tsx src/session/restore-head-align.test.ts && tsx src/session/responses-reasoning-roundtrip.test.ts && tsx src/session/memory-source-roundtrip.test.ts && tsx src/core/handle-query.history-restore.test.ts",
+    "test:mem7:state": "tsx src/mem7/core/__tests__/migrations.test.ts && tsx src/mem7/core/__tests__/source-store.test.ts && tsx src/mem7/core/__tests__/job-runner.test.ts && tsx src/mem7/core/__tests__/episode.test.ts && tsx src/mem7/core/__tests__/pipeline-recovery.test.ts && tsx src/mem7/core/__tests__/agent-recovery.test.ts && tsx src/mem7/core/__tests__/facts.test.ts && tsx src/mem7/core/__tests__/interpreter.test.ts && tsx src/mem7/core/__tests__/interpretation-backlog.test.ts && tsx src/mem7/core/__tests__/interpretation-input-isolation.test.ts && tsx src/mem7/core/__tests__/provider-cooldown.test.ts && tsx src/mem7/core/__tests__/fact-conflict-classification.test.ts && tsx src/mem7/core/__tests__/interpretation-concurrency.test.ts && tsx src/mem7/core/__tests__/schedule-interpretation.test.ts && tsx src/mem7/core/__tests__/change-feed.test.ts && tsx src/mem7/core/__tests__/change-feed-process.test.ts && tsx src/memory/proactive/__tests__/mutation-service.test.ts && tsx src/memory/proactive/__tests__/intent-runs.test.ts && tsx src/memory/proactive/__tests__/intent-lifecycle.test.ts && tsx src/mem7/core/__tests__/projection-runtime.test.ts && tsx src/mem7/core/__tests__/graph-memory-context.test.ts && tsx src/mem7/core/__tests__/scoped-opportunity.test.ts && tsx src/mem7/core/__tests__/interpretation-runtime.test.ts && tsx src/mem7/core/__tests__/basis-check.test.ts && tsx src/mem7/core/__tests__/context-service.test.ts && tsx src/mem7/core/__tests__/context-budget.test.ts && tsx src/memory/memdir/current-context-pending.test.ts && tsx src/mem7/core/__tests__/knowledge-history.test.ts && tsx src/mem7/core/__tests__/projection-read.test.ts && tsx src/mem7/core/__tests__/tombstones.test.ts && tsx src/mem7/core/__tests__/artifact-revalidation.test.ts && tsx src/mem7/core/__tests__/artifact-context.test.ts && tsx src/mem7/core/__tests__/context-time.test.ts && tsx src/mem7/core/__tests__/milestones-e2e.test.ts && tsx src/mem7/core/__tests__/migration-rehearsal.test.ts && tsx src/calendar/memory-binding.test.ts && tsx src/calendar/memory-occurrences.test.ts && tsx src/calendar/delivery-reconciliation.test.ts && tsx src/tools/memory-bridge.test.ts && tsx src/tools/memory-write.test.ts && tsx src/mem-serve/mem7-mcp.test.ts && tsx src/core/message-dispatcher.memory.test.ts && tsx src/core/message-dispatcher.host-context.test.ts && tsx src/session/memory-source-roundtrip.test.ts && tsx src/tools/notification-durable.test.ts && tsx src/tools/notification-state.test.ts && tsx src/channels/memory-send-guard.test.ts && tsx src/mem7/core/__tests__/vector-indexes.test.ts && tsx src/mem7/core/__tests__/vector-publications.test.ts && tsx src/mem7/core/__tests__/project-projections.test.ts && tsx src/memory/proactive/__tests__/host-deliveries.test.ts && tsx src/memory/proactive/__tests__/delivery-reconciliation.test.ts && tsx src/mem7/core/__tests__/host-turns.test.ts && tsx src/mem7/host-turn.test.ts && tsx src/core/handle-query.memory.test.ts && tsx src/core/handle-query.recall.test.ts && tsx src/channels/host-send-guard.test.ts && tsx src/optional/voice-chat/bridge.memory.test.ts && tsx src/mem7/core/__tests__/job-provenance.test.ts && tsx src/mem7/core/__tests__/artifact-history.test.ts && tsx src/mem7/core/__tests__/legacy-recovery.test.ts && tsx src/mem7/core/__tests__/read-visibility.test.ts && tsx src/mem7/core/__tests__/cascade-visibility.test.ts && tsx src/mem7/core/__tests__/entity-redirect-context.test.ts && tsx src/mem7/core/__tests__/projection-visibility.test.ts && tsx src/mem7/core/__tests__/search-visibility.test.ts && tsx src/mem7/core/__tests__/generation-visibility.test.ts && tsx src/mem7/core/__tests__/artifact-file-inputs.test.ts && tsx src/calendar/memory-read.test.ts && tsx src/memory/proactive/__tests__/consumer-visibility.test.ts && tsx src/memory/memdir/recall-visibility.test.ts && tsx src/memory/memdir/recall-compatibility.test.ts && tsx src/memory/proactive/__tests__/legacy-read-visibility.test.ts && tsx src/mem7/core/__tests__/prose-review-recovery.test.ts && npx tsx src/mem7/core/__tests__/change-feed-baseline.test.ts && npm run test:mem7:person && npm run test:mem7:worker",
+    "test:mem7:person": "tsx src/mem7/core/__tests__/person-relations.test.ts && tsx src/mem7/core/__tests__/person-import-cards.test.ts",
+    "test:mem7:worker": "tsx src/mem7/core/__tests__/job-claim-liveness.test.ts && tsx src/mem7/core/__tests__/ready-cell-liveness.test.ts && tsx src/mem7/core/__tests__/social-registration-liveness.test.ts && tsx src/mem7/core/__tests__/social-coverage-lifecycle.test.ts && tsx src/mem7/core/__tests__/social-classification-migration.test.ts && tsx src/mem7/core/__tests__/social-conversation-order.test.ts && tsx src/mem7/core/__tests__/social-processing-liveness.test.ts && tsx src/mem7/core/__tests__/social-identity-context.test.ts && tsx src/mem7/core/__tests__/social-structured-identity.test.ts && tsx src/mem7/core/__tests__/social-evidence-retention.test.ts && tsx src/mem7/core/__tests__/social-alias-provenance.test.ts && tsx src/mem7/core/__tests__/social-discovery-budget.test.ts && tsx src/mem7/core/__tests__/social-name-budget.test.ts && tsx src/mem7/worker-log.test.ts && tsx src/tools/memory-people-update.test.ts",
+    "test:mem7:relationships": "tsx src/mem7/core/__tests__/social-sources.test.ts && tsx src/mem7/core/__tests__/open-relationships.test.ts && tsx src/mem7/core/__tests__/qualified-relationship-guards.test.ts && tsx src/mem7/core/__tests__/social-interpretation.test.ts && tsx src/memory/proactive/__tests__/relationship-projection.test.ts && tsx src/memory/proactive/__tests__/relationship-snapshot.test.ts && tsx src/mem7/core/__tests__/relationship-context.test.ts && tsx src/mem7/core/__tests__/relationship-undated-recall.test.ts && tsx src/mem7/core/__tests__/relationship-history.test.ts && tsx src/inner-voice/reflection-memory.test.ts && tsx src/mem7/core/__tests__/relationship-cognition.test.ts && tsx src/memory/proactive/__tests__/relationship-notifications.test.ts && tsx src/memory/proactive/__tests__/relationship-notification-runtime.test.ts && tsx src/mem7/core/__tests__/relationships-e2e.test.ts && tsx scripts/relations-replay.test.ts && tsx src/core/query.relationship-budget.test.ts && tsx src/core/handle-query.relationship-provenance.test.ts",
+    "replay:relationships": "tsx scripts/relations-replay.ts",
+    "test:goal": "tsx src/goals/store.test.ts",
+    "test:goal:aim": "tsx src/goals/aim-adapter.test.ts",
+    "test:goal:task": "tsx src/goals/task-adapter.test.ts",
+    "test:goal:remote": "tsx src/goals/remote-task-adapter.test.ts",
+    "test:goal:recovery": "tsx src/goals/recovery.test.ts",
+    "test:goal:gate": "tsx src/nudge/registration-gate.test.ts",
+    "test:goal:command": "tsx src/commands/goal.test.ts",
+    "test:goal:step": "tsx src/tools/goal-step.test.ts",
+    "test:goal:tools": "tsx src/tools/goal-tools.test.ts && tsx src/tools/goal-clear.test.ts",
+    "test:goal:all": "npm run test:goal && npm run test:goal:calendar && npm run test:goal:restart && npm run test:goal:aim && npm run test:goal:task && npm run test:goal:remote && npm run test:goal:recovery && npm run test:goal:gate && npm run test:goal:command && npm run test:goal:step && npm run test:goal:tools && npm run test:goal:e2e",
+    "test:goal:calendar": "tsx src/goals/calendar-adapter.test.ts",
+    "test:goal:restart": "powershell -NoProfile -Command \"$d = Join-Path ([System.IO.Path]::GetTempPath()) ('engine7-goal-' + [guid]::NewGuid()); New-Item -ItemType Directory -Path $d | Out-Null; $id = (npx tsx src/goals/restart-writer.ts $d); npx tsx src/goals/restart-reader.ts $d $id; Remove-Item -Recurse -Force $d\"",
+    "test:goal:e2e": "powershell -NoProfile -Command \"$d = Join-Path ([System.IO.Path]::GetTempPath()) ('engine7-goal-e2e-' + [guid]::NewGuid()); New-Item -ItemType Directory -Path $d | Out-Null; npx tsx src/goals/e2e-phase1.ts $d; if ($LASTEXITCODE -ne 0) { Remove-Item -Recurse -Force $d; exit 1 }; npx tsx src/goals/e2e-phase2.ts $d; $code = $LASTEXITCODE; Remove-Item -Recurse -Force $d; exit $code\"",
+    "test:reload": "tsx src/reload/orchestrator.test.ts && tsx src/reload/secrets-watch.test.ts && tsx src/hooks/config.test.ts && tsx src/channels/manager.reconcile.test.ts && tsx src/channels/cc-bridge.test.ts && tsx src/engine-startup.shouldSkipReload.test.ts && tsx src/optional/voice-chat/plugin.deps.test.ts && tsx src/optional/voice-chat/config.test.ts && tsx src/cron/cron-plugin.resync.test.ts && tsx src/startup/message-pipeline.test.ts && tsx src/utils/sanitize-error.test.ts && tsx src/reload/config-watch-selfwrite.test.ts && tsx src/channels/feishu.typing.test.ts && tsx src/channels/feishu.sendfile-apicheck.test.ts",
+    "test:installer": "tsx src/cli-events.test.ts && tsx src/cli-init-config.test.ts && tsx src/cli-init-answers.test.ts && tsx src/cli-init.headless.e2e.test.ts && tsx src/cli-reconfig.test.ts && tsx src/cli-doctor.test.ts && tsx src/config/secrets-env.test.ts && tsx src/startup/api-status.test.ts && tsx src/cli-service.test.ts && tsx src/emotion-hook.test.ts && tsx src/emotion-hook.golden.test.ts && tsx src/tools/tavily-pool.test.ts && tsx src/tools/service.test.ts",
+    "test:persona": "tsx src/config/persona-files.test.ts && tsx src/light-mode.test.ts && tsx src/inner-voice/plugin.test.ts && tsx src/cli-init-config.test.ts",
+    "test:responses": "tsx src/models/openai-responses-provider.test.ts && tsx src/session/responses-reasoning-roundtrip.test.ts && tsx src/compact/tokenEstimate.test.ts",
+    "build:public": "node scripts/build.mjs --public",
+    "prepublishOnly": "npm run build:public",
+    "start": "node dist/main.mjs",
+    "smoke:boot": "node scripts/boot-smoke.mjs",
+    "smoke:boot:baseline": "node scripts/boot-smoke.mjs --baseline",
+    "gen:template": "tsx scripts/gen-config-template.ts",
+    "test:config": "tsx src/config/registry.test.ts && tsx src/config/writer.test.ts && tsx src/config/secrets-io.test.ts && tsx src/config/secrets-env.test.ts && tsx src/config/migrations/index.test.ts && tsx src/config/migrations/persona.test.ts && tsx src/config/migrations/channel-secrets.test.ts && tsx src/config/cli-config.test.ts && tsx src/config/tts-providers.test.ts && tsx src/config/tts-voices.test.ts && tsx src/free-tier-apply.test.ts && tsx src/startup/free-tier-video-fixup.test.ts && tsx src/startup/free-tier-voice-fixup.test.ts",
+    "test:time": "tsx src/time/tz-matrix.test.ts && tsx src/time/time-discipline.test.ts && tsx src/time/tsc-guard.test.ts",
+    "test:tts": "tsx src/config/tts-providers.test.ts && tsx src/tools/tts/pick.test.ts && tsx src/tools/tts/route.test.ts && tsx src/tools/tts/relay.test.ts && tsx src/tools/tts/providers.test.ts && tsx src/tools/tts/impl.test.ts && tsx src/tools/tts/usage.test.ts && tsx src/tools/my-voice.dispatch.test.ts",
+    "test:toolstate": "tsx src/core/names-from-history.test.ts",
+    "test:loopbrake": "tsx src/tools/loop-brake.test.ts",
+    "test:mcp": "tsx src/mcp/convert-input-schema.test.ts",
+    "test:proactive": "tsx src/memory/proactive/__tests__/verdict-parse.test.ts && tsx src/memory/proactive/__tests__/sweep.test.ts && tsx src/memory/proactive/__tests__/persistence.test.ts && tsx src/memory/proactive/__tests__/graph-store-p0.test.ts && tsx src/memory/proactive/__tests__/opportunity.test.ts && tsx src/memory/proactive/__tests__/calendar-bridge.test.ts && tsx src/memory/proactive/__tests__/timezone-fix.test.ts && tsx src/memory/proactive/__tests__/stale-fix.test.ts && tsx src/memory/proactive/__tests__/completion-evidence.test.ts && tsx src/memory/proactive/__tests__/plugin.test.ts",
+    "prepack": "npm run test:time && npm run test:proactive && npm run test:persona && npm run test:config && npm run test:toolstate && npm run test:loopbrake && npm run test:mcp && npm run test:tts"
+  },
+  "engines": {
+    "node": ">=22"
+  },
+  "os": [
+    "win32",
+    "linux",
+    "darwin"
+  ],
+  "cpu": [
+    "x64",
+    "arm64"
+  ],
+  "devDependencies": {
+    "@types/node": "^22.19.19",
+    "@types/turndown": "^5.0.6",
+    "@types/ws": "^8.18.2",
+    "esbuild": "^0.25.10",
+    "tsx": "^4.22.3",
+    "typescript": "^6.0.0"
+  },
+  "dependencies": {
+    "@larksuiteoapi/node-sdk": "^1.66.1",
+    "@modelcontextprotocol/sdk": "^1.29.0",
+    "baileys": "7.0.0-rc14",
+    "chokidar": "^5.0.0",
+    "croner": "^10.0.1",
+    "discord.js": "^14.26.4",
+    "https-proxy-agent": "^9.1.0",
+    "qrcode-terminal": "^0.12.0",
+    "sharp": "^0.34.5",
+    "sqlite-vec": "^0.1.9",
+    "turndown": "^7.2.4",
+    "typebox": "^1.1.38",
+    "ws": "^8.22.0"
+  },
+  "optionalDependencies": {
+    "@discordjs/opus": "^0.10.0",
+    "@engine7/lancedb-intel": "^1.0.0",
+    "@ffmpeg-installer/ffmpeg": "^1.1.0",
+    "@lancedb/lancedb": "^0.38.0",
+    "@node-rs/jieba": "^2.0.3",
+    "sqlite-vec-darwin-arm64": "^0.1.9",
+    "sqlite-vec-darwin-x64": "^0.1.9",
+    "sqlite-vec-linux-arm64": "^0.1.9",
+    "sqlite-vec-linux-x64": "^0.1.9",
+    "sqlite-vec-windows-x64": "^0.1.9"
+  },
+  "files": [
+    "dist/**/*.mjs",
+    "dist/prompts/**/*.md",
+    "bin/**/*",
+    "templates/**/*",
+    "src/memory/everos/python/**/*",
+    "!**/__pycache__/**",
+    "!**/*.pyc",
+    "!**/node_modules/**",
+    "!**/.env",
+    "README.md"
+  ]
+}

package/templates/config.template.json

@@ -107,6 +107,11 @@
       }
     }
   },
+  "topics": {
+    "extract": {
+      "intervalMinutes": 15
+    }
+  },
   "channels": {
     "discord": {
       "enabled": false,
@@ -119,6 +124,26 @@
       },
       "replyToMode": "all"
     },
+    "telegram": {
+      "enabled": false,
+      "token": "",
+      "dmPolicy": "open",
+      "groupPolicy": "mention-only",
+      "allowFrom": []
+    },
+    "whatsapp": {
+      "enabled": false,
+      "dmPolicy": "allowlist",
+      "allowFrom": [],
+      "replyToMode": "off",
+      "sendReadReceipts": false,
+      "selfChatPrefix": true,
+      "groupDigest": {
+        "enabled": true,
+        "patrolIntervalMin": 60,
+        "maxLinesPerGroup": 500
+      }
+    },
     "feishu": {
       "enabled": true,
       "appId": "{{FEISHU_APP_ID}}",

package/templates/skills/email/LICENSE.txt

@@ -0,0 +1,22 @@
+MIT License
+
+Copyright (c) 2026 NetEase (original imap-smtp-email skill)
+Copyright (c) 2026 engine7 contributors (installer adaptation)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/templates/skills/email/SKILL.md

@@ -0,0 +1,115 @@
+---
+name: email
+description: "TRIGGER: 邮件/邮箱/发邮件/收邮件/查邮箱/看信/email/发信/收信。通过 IMAP/SMTP 收发邮件（QQ/163/126/188 系/Gmail/Outlook 通用）。凭证从 stateDir config 的 tools.email 读取（托盘「邮件」页填写）。"
+---
+
+# 邮件（IMAP/SMTP）
+
+收信：`scripts/imap.js`（check / list-subjects / list-senders / search / fetch / download / mark-read）
+发信：`scripts/smtp.js`（send / test）
+分级闸与联系人名单：`scripts/email-gate.js`
+
+所有命令都支持 `--json`：成功 `{ok:true, reason:'success'|'no_new_mail', result}`；失败 `{ok:false, reason:'auth_failed'|'network'|'timeout'|'rate_limited'|'config_missing'|'daily_limit'|'not_found'|'unknown'}`。**agent 判定一律加 `--json`**——错误分类是硬要求，绝不能把「连不上」说成「没收到」。
+
+## 配置与首次引导
+
+凭证在 stateDir config 的 `tools.email`（托盘 → 设置 → 邮件 填写）。命令会自动读取（`ENGINE_STATE_DIR` 定位）；`.env` 形态兼容但不再是推荐路径。
+
+**未配置时（用户第一次提出用邮箱）**，按此顺序引导（QQ 部分为 1010 实测——四步截图与结论存档于 docs/assets/qq-mail-authcode/）：
+
+1. **先确认入口**：「你电脑右下角能看到一个小图标吗？可能藏在 ^ 折叠箭头里」——用户找不到托盘时不会说找不到，只会说「我按你说的做了没用」，必须先确认。
+2. **拿授权码**（按用户邮箱给对应叫法：QQ/163=授权码，Gmail=App Password）：
+   - QQ 实测路径：**mail.qq.com → 右上角头像 → 账号与安全 → 左侧「安全设置」→ 页面最底部**找「POP3/IMAP/SMTP… 服务」区（注意：新版叫「账号与安全」不是老的「设置→账户」；服务开关在**安全设置页底部**，不在账号设置里）
+   - 服务**早已开启**（大多数老 QQ 号默认开过）→ 直接点「**生成授权码**」；只有看到「开启服务」才需要先开
+   - **验证是「用户主动发一条短信」**（不是收验证码！）：用绑定手机给指定号码发一条指定内容的短信 → 回电脑点「我已发送」。两个更省事的逃逸路径先告诉用户：**微信扫页面二维码代发**——扫码后手机微信自动弹出「发送短信」确认页，点一下「发送」就代发了（手机不在手边时的首选）；左下角「选择其他方式验证」有备选
+   - 授权码在**验证后的网页弹窗**里出现（短信里没有码），提醒用户复制 + 建议顺手填「备注用途」（授权码管理页可分辨多把码；**QQ 支持多码并存，新生成不作废旧码**——用户家里另一台设备在用的码不受影响，可以放心告诉用户）
+   - 用户卡住时可以发截图——**先提醒「把授权码/密码区域遮住再发」**；或让用户只说「现在在哪个页面」。收到含疑似授权码的图：不复述、不落盘引用。
+3. **填面板**：托盘 → 邮件，填**完整邮箱地址**（xxx@qq.com，不能填 QQ 号）+ 授权码。
+4. **验活**：用户说「好了」后跑 `node scripts/imap.js check --limit 3 --json`：
+   - `ok:true` → 接管完成，顺带报未读摘要
+   - `reason:auth_failed` → **先排地址再排授权码**：「方便回我一句你填的完整邮箱地址吗？QQ 邮箱要 xxx@qq.com 全称，不能填 QQ 号」——地址没错再让用户重新生成授权码
+   - `reason:network/timeout` → 网络/服务器问题，稍后再试，**别说是凭证错**
+
+**陪用户拿授权码的两个手段，按这个顺序用**：
+
+1. **Playwright 陪操作**（首选）：用 Playwright 的 browser_navigate 在用户电脑上打开浏览器窗口（headed，用户看得见），把路直接带到对应页面——QQ 开 `mail.qq.com`（登录后可直接导航到「账号与安全」）、163 开 `mail.163.com`、Gmail 直达 `myaccount.google.com/apppasswords`。**分工**：导航、找入口这些 agent 代劳（用户卡的多是「找不到在哪」）；登录输密码、短信验证、生成授权码这些**让用户自己动手**（agent 别碰凭证），你在旁边看着页面状态提示下一步就行。
+2. **发步骤图**（Playwright 不可用或用户不习惯时；assets/ 目录，QQ 流程实拍已脱敏）：把对应的图直接发到聊天里，比文字描述有效得多——发之前提醒用户**回图时遮住授权码区域**：
+
+- `assets/1-account-menu.jpg` 找不到入口 → 「点右上角头像 → 账号与安全」
+- `assets/2-security-settings-generate.jpg` 找不到生成按钮 → 「安全设置页最底部 → 生成授权码」
+- `assets/3-sms-verify.jpg` 验证步骤懵了 → 「发一条短信 / 微信扫码代发 → 我已发送」
+- `assets/4-wechat-send-confirm.jpg` 手机不在手边 → 「微信扫码，手机上点一下发送就行」
+- `assets/5-code-generated.jpg` 验证通过后 → 「码就弹在这个页面，点复制」
+
+## 收信命令
+
+```bash
+# 监控「有没有新邮件」的落地命令（不下载正文，快）：
+node scripts/imap.js list-subjects --limit 20 --recent 30m --json
+# 发件人分布（监控选址：先看分布再决定搜什么）：
+node scripts/imap.js list-senders --limit 50 --recent 7d --json
+# 查未读（拉正文）：
+node scripts/imap.js check --limit 10 --recent 2h --json
+# 搜邮件：
+node scripts/imap.js search --from certum.pl --recent 30d --json
+# 读全文（关键判读必须 fetch 正文，激活码/链接常在正文里）：
+node scripts/imap.js fetch <uid> --json
+# 下载附件 / 标已读 / 列文件夹：
+node scripts/imap.js download <uid> --dir <path> --json
+node scripts/imap.js mark-read <uid> --json
+node scripts/imap.js list-mailboxes --json
+```
+
+## 发信：分级闸（必须遵守）
+
+发信前**必须**先跑闸判定，凭 verdict 行动：
+
+```bash
+node scripts/email-gate.js check --to a@x.com,b@y.com [--cc ...] [--bcc ...] [--attach] [--ext-link]
+# → {"action":"send"}     全部收件人已验证且无附件无外链：直接发
+# → {"action":"confirm","reasons":{...}}  有陌生收件人/附件/外链：起草给用户过目，用户明确说发才发
+```
+
+规则：
+- **候选/已验证两级**：任何来信/去信地址自动入「候选」；agent 遇到要发往候选地址时问用户一句「这个地址以后可以直接回吗？」，用户同意后 `node scripts/email-gate.js verify --addr <addr>` 升级为已验证（此后直发不再问）
+- **逐收件人判定**：to+cc+bcc 拆开逐个判，任一地址未验证 → 整封降级需确认（防「回复 A 直发、夹带的 B 跟着出去」）
+- 带附件 / 正文含外链 → 永远需要确认，且**不参与攒批**（攒批仅限纯文本待确认件，≤3 封一批一起过目）
+- 发送永远留底（`email-sent.log`）；单日上限默认 20 封（超限报 `daily_limit`），防失控群发
+
+```bash
+node scripts/smtp.js send --to a@x.com --subject "..." --body "..." --json
+# --cc/--bcc/--attach file1,file2/--body-file f/--html-file f
+```
+
+## 邮件监控四规则（写 cron/定时任务时必须遵守）
+
+239 事故（2026-09-30～10-08，8 天误报）的教训，四条缺一不可：
+
+1. **先看发件人分布再决定搜什么**：`list-senders` 看实际来源——同一事件的邮件可能来自不同域/第三方发信域（SendGrid 等），**不要预设单一域名**，更不要搜事件名（「证书签发了」这种事件名几乎从不出现在真实标题里）
+2. **监控「有没有新邮件」，别监控「某封到没到」**：`list-subjects --recent 30m` 拿标题列表，标题命中才算收到
+3. **判读必须 fetch 正文**：激活码/关键信息常在正文，只看标题就判「不是它」是 239 的直接死因
+4. **死线**：连续多次无命中（建议 24h 内 20 次）→ 自动降级：`list-subjects --limit 10` 列最近标题给用户看，问「有没有你要的」。关键词死等必须有尽头——规则只管怎么找，死线管找不到怎么办
+
+## Gmail 用户（⚠️ 推演未实测——高级用户向：能自己走到这的人不需要逐步引导）
+
+Gmail 的授权码叫 **App Password（应用密码）**，两个前提先确认：
+1. **网络**：这台电脑要能正常访问 Google 服务——连不上的话 IMAP（imap.gmail.com）也连不上，配了也用不了，先解决网络再谈
+2. **两步验证（2FA）**：账号必须已开启两步验证，否则看不到应用密码入口（没开的先去 Security → 2-Step Verification 启用，比 QQ 的短信验证麻烦）
+
+流程：用 Playwright 直接打开 `https://myaccount.google.com/apppasswords` → 用户登录并输一次密码 → 应用名随便填（如 engine7）→ 生成 16 位码（格式 xxxx xxxx xxxx xxxx，**含空格**，填面板时把空格去掉）。
+
+## 163/126/188 用户（✅ 1010 实测端到端通过——IMAP/SMTP/发信全绿；入口页截图归档 docs/assets/163-mail-authcode/）
+
+网易系的授权码叫**客户端授权密码**。流程（步骤图在 assets/163/，1010 实测）：
+Playwright 开 `mail.163.com`（126/188 同理）→ 登录 → 顶部**设置 → POP3/SMTP/IMAP**：
+- **服务没开**（显示「已关闭」）：点 IMAP/SMTP 服务右侧「**开启**」→ **中途会弹「账号安全提示」**（吓唬人的：邮件泄露风险+推荐装邮箱大师）——**必须点「继续开启」**，小白最容易被这页劝退，要预告 → 成功后**授权码当场弹出**（只显示一次，立即复制）。⚠️ POP3/SMTP 是另一个开关，**我们不需要**（收信走 IMAP），别开
+- **服务已开**：页面上点 **「新增授权密码」** 再拿一把（支持多把并存）
+- 可能要求**收**短信验证码（是收不是发，比 QQ 容易）
+
+码是 16 位**大小写混合**（区分大小写，QQ 是纯小写），复制时别动大小写。⚠️ 163 风控会给绑定手机发「新设备登录提醒」——正常现象不拦截，提前告诉用户别慌。与 QQ 的另一差异：入口在设置页一级菜单（不藏在「安全设置」底部）。
+
+163 步骤图（assets/163/）：`1-settings-menu.jpg` 入口 / `2-service-closed.jpg` 服务开关 / `3-security-warning.jpg` 安全提示（点继续开启）/ `5-imap-code.jpg` 出码页。
+
+## 常见服务商
+
+QQ / 163 / 126 / 188 / vip 系列 / Gmail / Outlook 的服务器映射已内置（按邮箱域名自动识别，未知域名需在面板手动填 host）。授权码不是登录密码——各家的叫法与入口见上面各段。