chore: clean absolute paths, curasoft refs, secrets from history

- bin/chats/: env-based paths, no /mnt/ /home/ hardcodes - bin/edelweiss-pilot: remove curasoft, use DOCS_BASE env var - bin/facts/crm: use KNOWLEDGE_MESH_SEED env var - compose.edelweiss.yml: remove curasoft volumes, use DOCS_BASE - docs/chat-import-plan.md: link to Gitea issue, no secrets - bin/seed-edelweiss-facts.py: removed (curasoft-only)
2026-08-13 00:00:15 +01:00
parent 27d9521e7f
commit fec5325c7a
7 changed files with 403 additions and 262 deletions
@@ -239,7 +239,6 @@ func findOO() string {
 	}
 	candidates := []string{
 		filepath.Join(os.Getenv("HOME"), "go", "bin", "oo"),
-		"/home/ano/go/bin/oo",
 	}
 	for _, c := range candidates {
 		if _, err := os.Stat(c); err == nil {
@@ -42,28 +42,17 @@ func runSyncTelegram(args []string) int {
 		return 2
 	}

-	// Locate telegram-mcp directory
-	mcpDirs := []string{
-		envVar("TELEGRAM_MCP_DIR", ""),
-		"/mnt/8TB/projects/eslider/mcp-servers/telegram-mcp",
-		filepath.Join(os.Getenv("HOME"), "projects", "eSlider", "mcp-servers", "telegram-mcp"),
-	}
-	var mcpDir string
-	for _, d := range mcpDirs {
-		if d != "" {
-			if _, err := os.Stat(filepath.Join(d, "main.py")); err == nil {
-				mcpDir = d
-				break
-			}
-		}
-	}
+	mcpDir := envVar("TELEGRAM_MCP_DIR", "")
 	if mcpDir == "" {
-		fmt.Fprintln(os.Stderr, "chats: telegram-mcp not found; set TELEGRAM_MCP_DIR")
+		fmt.Fprintln(os.Stderr, "chats: set TELEGRAM_MCP_DIR to telegram-mcp directory")
+		return 1
+	}
+	if _, err := os.Stat(filepath.Join(mcpDir, "main.py")); err != nil {
+		fmt.Fprintf(os.Stderr, "chats: TELEGRAM_MCP_DIR=%s: main.py not found\n", mcpDir)
 		return 1
 	}

 	if sessionStr == "" {
-		// Fallback: try to read from telegram-mcp .env
 		envPath := filepath.Join(mcpDir, ".env")
 		if data, err := os.ReadFile(envPath); err == nil {
 			for _, line := range strings.Split(string(data), "\n") {
@@ -76,20 +65,10 @@ func runSyncTelegram(args []string) int {
 		}
 	}
 	if sessionStr == "" {
-		// Also try the env file at /mnt/8TB/projects/eslider/mcp-servers/telegram.env
-		envPath := "/mnt/8TB/projects/eslider/mcp-servers/telegram.env"
-		if data, err := os.ReadFile(envPath); err == nil {
-			for _, line := range strings.Split(string(data), "\n") {
-				if strings.HasPrefix(line, "TELEGRAM_SESSION_STRING=") {
-					sessionStr = strings.TrimPrefix(line, "TELEGRAM_SESSION_STRING=")
-					sessionStr = strings.Trim(sessionStr, "\"'")
-					break
-				}
-			}
-		}
+		sessionStr = envVar("TELEGRAM_SESSION_STRING", "")
 	}
 	if sessionStr == "" {
-		fmt.Fprintln(os.Stderr, "chats: TELEGRAM_SESSION_STRING not found in env or .env files")
+		fmt.Fprintln(os.Stderr, "chats: TELEGRAM_SESSION_STRING not found; set env or in TELEGRAM_MCP_DIR/.env")
 		return 1
 	}

@@ -6,32 +6,26 @@ cd "$ROOT"
 export HF_HOME="${HF_HOME:-$ROOT/var/hf}"
 export PATH="$ROOT/.venv/bin:$PATH"
 mkdir -p "$HF_HOME" "$ROOT/var"
-# Corpus policy (Andriy 2026-08-12):
-# 1 cs-lexicon, 2 vendor kbs, 3 reports only (not raw STT),
-# 4/5/6 seeded via pairing (seed-edelweiss-facts.py),
-# 7 SCHEMA/ARCHITECTURE churn — do not promote to facts,
-# 9 glossary+subjects via docs/docs
-DOCS=(
-  --corpus /home/devops/projects/docs/docs
-  --corpus /home/devops/projects/edelweiss-curasoft/docs
-  --corpus /home/devops/projects/edelweiss-ui/docs
-  --corpus /home/devops/projects/edelweiss-curasoft/docs/lexicon
-  --corpus /home/devops/projects/curasoft/docs/lexicon
-  --corpus /home/devops/projects/curasoft/docs/curasoft-de/kbs
-  --corpus /home/devops/projects/docs/docs/reports
-)
+
+# Corpus paths set via env; fallback to PROJECTS_ROOT convention.
+PROJECTS="${PROJECTS_ROOT:-/home/devops/projects}"
+DOCS_BASE="${DOCS_BASE:-$PROJECTS}"
+DOCS=()
+for d in \
+  "$DOCS_BASE/docs/docs" \
+  "$DOCS_BASE/edelweiss-ui/docs" \
+  "$DOCS_BASE/docs/docs/reports"; do
+  [ -d "$d" ] && DOCS+=(--corpus "$d")
+done
+
 cmd="${1:-stats}"; shift || true
 case "$cmd" in
-  # Write leafs without indexes → seed facts → ensure_indexes (FTS+HNSW).
-  # Never upsert under live FTS; never DROP INDEX (ghost catalog).
  index)
    bin/kb/index --rebuild --skip-indexes "${DOCS[@]}" "$@"
-    .venv/bin/python bin/seed-edelweiss-facts.py
    ;;
  search) exec bin/edelweiss-search "$@" ;;
  search-go) exec bin/kb/search "$@" ;;
  stats)  exec bin/kb/stats "$@" ;;
  audit)  exec bin/facts/audit "${1:-db}" "${@:2}" ;;
-  seed)   exec .venv/bin/python bin/seed-edelweiss-facts.py "$@" ;;
-  *) echo "usage: $0 index|search|search-go|stats|audit|seed ..." >&2; exit 2 ;;
+  *) echo "usage: $0 index|search|search-go|stats|audit ..." >&2; exit 2 ;;
 esac
@@ -19,6 +19,7 @@ Usage:
 from __future__ import annotations

 import json
+import os
 import sys
 from pathlib import Path

@@ -27,7 +28,8 @@ sys.path.insert(0, str(ROOT / "bin" / "tools"))

 from kblib import upsert_leaf, connect, leaf_id  # noqa: E402

-CORPUS_MESH = Path("/mnt/8TB/projects/eslider/cv/projects/knowledge-mesh-seed.yaml")
+MESH_ENV = os.environ.get("KNOWLEDGE_MESH_SEED", "")
+CORPUS_MESH = Path(MESH_ENV) if MESH_ENV else ROOT / "../knowledge-mesh-seed.yaml"


 def corpus_orgs(raw: str) -> dict[str, dict]:
@@ -1,187 +0,0 @@
-#!/usr/bin/env python3
-"""Seed confirmed Edelweiss facts via pairing (a x b). Never DROP INDEX."""
-from __future__ import annotations
-
-import sys
-from pathlib import Path
-
-ROOT = Path(__file__).resolve().parents[1]
-sys.path.insert(0, str(ROOT / "bin" / "tools"))
-
-import yaml
-from kblib import (  # noqa: E402
-    CONF_CONFIRMED,
-    ROOT_FACTS,
-    connect,
-    ensure_indexes,
-    leaf_index_names,
-    stats,
-    upsert_leaf,
-)
-from model2vec import StaticModel  # noqa: E402
-
-LEX = Path("/home/devops/projects/edelweiss-curasoft/docs/lexicon/edelweiss-lexicon.yml")
-CS = Path("/home/devops/projects/curasoft/docs/lexicon/cs-lexicon.yml")
-
-
-def fact(text: str, source: str, how: str, loc: str) -> tuple[str, str, str, str]:
-    return text, source, how, loc
-
-
-def from_edelweiss_lex(lex: dict) -> list[tuple[str, str, str, str]]:
-    rows: list[tuple[str, str, str, str]] = []
-    gl = lex["hosts"]["gl"]
-    rows.append(fact(
-        "CuraSoft GL server VM IP is %s on Fritz LAN (%s)." % (gl["ip"], gl["role"]),
-        gl["evidence"], "lexicon hosts.gl", "edelweiss-lexicon.yml",
-    ))
-    for key in ("pdl", "pdl2", "spdl", "gf"):
-        h = lex["hosts"][key]
-        rows.append(fact(
-            "Edelweiss host %s IP %s role=%s." % (key.upper(), h["ip"], h["role"]),
-            h["evidence"], "lexicon hosts.%s" % key, "edelweiss-lexicon.yml",
-        ))
-    pg, ms = lex["ports"]["postgres"], lex["ports"]["mssql"]
-    rows.append(fact(
-        "GL LAN DB: PostgreSQL port %s open=%s; MSSQL %s reachable=%s."
-        % (pg["port"], pg["reachable_from_lan"], ms["port"], ms["reachable_from_lan"]),
-        pg["evidence"], "lexicon ports", "edelweiss-lexicon.yml",
-    ))
-    b = lex["brain_2dph"]
-    rows.append(fact(
-        "Edelweiss 2dph pilot brain at %s; facts need >=2 sources. CLI: %s."
-        % (b["path"], b["cli"]),
-        b["evidence"], "lexicon brain_2dph", "edelweiss-lexicon.yml",
-    ))
-    v = lex["password_vault"]
-    rows.append(fact(
-        "Password vault pilot: %s; OO task #%s deadline %s."
-        % (v["choice"], v["oo_task"], v["deadline"]),
-        v["evidence"], "lexicon password_vault", "edelweiss-lexicon.yml",
-    ))
-    a = lex["assistant_pilot"]
-    rows.append(fact(
-        "Artem consent %s for local Pflege/CuraSoft assistant; scope: %s."
-        % (a["consent_date"], a["scope"]),
-        a["evidence"], "lexicon assistant_pilot", "edelweiss-lexicon.yml",
-    ))
-    return rows
-
-
-def from_cs_lex(cs: dict) -> list[tuple[str, str, str, str]]:
-    """Item 1: cs-lexicon → facts only with paired evidence."""
-    rows: list[tuple[str, str, str, str]] = []
-    src = cs.get("sources") or {}
-    default_pair = None
-    if isinstance(src, dict) and len(src) >= 2:
-        keys = list(src.keys())[:2]
-        default_pair = "%s x %s" % (keys[0], keys[1])
-    ep = cs.get("ep_typ") or {}
-    for code, meta in ep.items():
-        if not isinstance(meta, dict):
-            continue
-        ev = (meta.get("evidence") or "").replace("×", " x ")
-        if " x " not in ev:
-            if not default_pair:
-                continue
-            ev = default_pair
-        meaning = meta.get("meaning") or meta.get("name") or ""
-        rows.append(fact(
-            "CuraSoft ep_typ %s = %s (%s)." % (code, meta.get("name"), meaning),
-            ev, "cs-lexicon ep_typ", "cs-lexicon.yml",
-        ))
-    pair = "cs-lexicon.yml x curasoft-de/kbs"
-    besuch = cs.get("besuch") or {}
-    for code, label in besuch.items():
-        if code == "note" or not isinstance(label, str):
-            continue
-        rows.append(fact(
-            "CuraSoft Besuch slot %s = %s." % (code, label),
-            pair, "cs-lexicon besuch", "cs-lexicon.yml",
-        ))
-    bm = cs.get("binary_map") or {}
-    if isinstance(bm, dict) and bm:
-        rows.append(fact(
-            "CuraSoft binary_map documents RE code/status mappings for detective (not live PHI).",
-            "cs-lexicon.yml x curasoft-detective skill",
-            "cs-lexicon binary_map", "cs-lexicon.yml",
-        ))
-    return rows
-
-
-def from_oo_and_interview() -> list[tuple[str, str, str, str]]:
-    """Items 4+5+6: OO, QEMU/GL, interview bullets via pairing."""
-    return [
-        fact(
-            "OO Edelweiss Remote Work #24: Vaultwarden deploy task #431 deadline 2026-08-19; companion #432 deploy DL 2026-08-15.",
-            "OO#431 x 07-interview-2026-08-12.md",
-            "OO calendar/tasks pair", "office.produktor.io + reports/07",
-        ),
-        fact(
-            "Fahrplan pilot kickoff agreed for Tuesday after 2026-08-12 interview; OO task #434 deadline 2026-08-18.",
-            "OO#434 x 07-interview-2026-08-12.md",
-            "interview x OO", "reports/07 + OO",
-        ),
-        fact(
-            "Local Pflege/CuraSoft assistant pilot: email+docs sync consented; phone/STT ticket flow is phase 2.",
-            "07-interview-2026-08-12.md x edelweiss-lexicon.yml",
-            "interview structured bullets", "reports/07",
-        ),
-        fact(
-            "Meta/CuraSoft Wiener Hostel event tentatively 2026-08-18 ~09:30 CEST — confirm name/place with Artem.",
-            "07-interview-2026-08-12.md x OO calendar event 157",
-            "interview x calendar", "reports/07 + OO",
-        ),
-        fact(
-            "QEMU/GL safe ops: never HMP quit/reset unless asked; live disk always qemu-img -U; GL VM via dockur acronis-boot.",
-            "gl-vm-access.md x qemu-monitor-safe rule",
-            "QEMU/GL access pair", "edelweiss-curasoft/docs + .cursor/rules",
-        ),
-        fact(
-            "2dph indexes interview reports under docs/docs/reports; raw STT docs/stt is not corpus (reports only).",
-            "2dph-edelweiss-pilot.md x 00-summary.md",
-            "STT policy pair", "pilot doc + reports",
-        ),
-    ]
-
-
-def main() -> int:
-    lex = yaml.safe_load(LEX.read_text())
-    cs = yaml.safe_load(CS.read_text()) if CS.exists() else {}
-    model = StaticModel.from_pretrained("minishlab/potion-multilingual-128M")
-    db, conn = connect(read_only=False)
-
-    rows: list[tuple[str, str, str, str]] = []
-    rows.extend(from_edelweiss_lex(lex))
-    rows.extend(from_cs_lex(cs))
-    rows.extend(from_oo_and_interview())
-
-    for text, source, how, loc in rows:
-        if " x " not in source:
-            print("skip (no pair)", how, text[:60])
-            continue
-        emb = model.encode([text])[0].astype(float).tolist()
-        upsert_leaf(
-            conn,
-            text=text,
-            root=ROOT_FACTS,
-            confidence=CONF_CONFIRMED,
-            source=source,
-            source_rev="2026-08-12",
-            how=how,
-            loc=loc,
-            type_="fact",
-            embedding=emb,
-        )
-        print("ok", how)
-
-    ensure_indexes(conn)
-    print("INDEXES", sorted(leaf_index_names(conn)))
-    print(stats(conn))
-    conn.close()
-    db.close()
-    return 0
-
-
-if __name__ == "__main__":
-    raise SystemExit(main())
@@ -1,14 +1,11 @@
-# 2dph Edelweiss pilot — knowledge brain for Pflegedienst facts/info only.
+# 2dph Edelweiss pilot — knowledge brain.
 #
 #   docker compose -f compose.edelweiss.yml build
 #   docker compose -f compose.edelweiss.yml run --rm brain index --rebuild \
-#     --corpus /corpus/docs --corpus /corpus/curasoft-docs \
-#     --corpus /corpus/ui-docs --corpus /corpus/vendor-kbs \
-#     --corpus /corpus/reports
-#   docker compose -f compose.edelweiss.yml run --rm brain search "GL IP"
+#     --corpus /corpus/docs --corpus /corpus/ui-docs --corpus /corpus/reports
+#   docker compose -f compose.edelweiss.yml run --rm brain search "query"
 #
-# HNSW note: never DROP INDEX Leaf.Leaf_vec on a live DB (ghost catalog).
-# Fresh indexes = delete volume/db + --rebuild. See kblib.create_fts_and_vector.
+# Host corpus dirs set via env: DOCS_BASE (default: /home/devops/projects)
 name: 2dph-edelweiss

 services:
@@ -25,18 +22,15 @@ services:
      KB_WORKERS: "2"
      KB_PY: python3
      KB_ROOT: /app
+      DOCS_BASE: ${DOCS_BASE:-/home/devops/projects}
    volumes:
      - kb-model:/data/hf
      - kb-var:/data
      - ./:/app
-      # Edelweiss host corpus mounts (override paths on other hosts)
-      - /home/devops/projects/docs/docs:/corpus/docs:ro
-      - /home/devops/projects/edelweiss-curasoft/docs:/corpus/curasoft-docs:ro
-      - /home/devops/projects/edelweiss-ui/docs:/corpus/ui-docs:ro
-      - /home/devops/projects/curasoft/docs/curasoft-de/kbs:/corpus/vendor-kbs:ro
-      - /home/devops/projects/docs/docs/reports:/corpus/reports:ro
-      # STT raw mounted read-only for humans — do NOT pass as --corpus
-      - /home/devops/projects/docs/stt:/corpus/stt:ro
+      - ${DOCS_BASE:-/home/devops/projects}/docs/docs:/corpus/docs:ro
+      - ${DOCS_BASE:-/home/devops/projects}/edelweiss-ui/docs:/corpus/ui-docs:ro
+      - ${DOCS_BASE:-/home/devops/projects}/docs/docs/reports:/corpus/reports:ro
+      - ${DOCS_BASE:-/home/devops/projects}/docs/stt:/corpus/stt:ro
    command: ["brain", "search", "help"]
    read_only: false
    tmpfs:
@@ -1,5 +1,6 @@
 # Chat Import Pipeline

+<<<<<<< Updated upstream
 Plan and implementation details: https://git.produktor.io/eSlider/brain-chats-import/issues/1

 Progress tracked in that issue. This file kept as a local reference for agent sessions.
@@ -91,9 +92,17 @@ OO_CLI                     path to oo binary (default: $HOME/go/bin/oo)
 ### Rambox

 Не используется. IndexedDB база не существует.
+||||||| Stash base
+Created: 2026-08-12
+Author: Andriy Oblivantsev
+Purpose: One-shot handoff for a new agent session to implement `bin/chats`
+=======
+Plan and implementation details: https://git.produktor.io/eSlider/brain-chats-import/issues/1
+>>>>>>> Stashed changes

---
+Progress tracked in that issue. This file kept as a local reference for agent sessions.

+<<<<<<< Updated upstream
 ## 2. Design Decisions

 ### One binary: `bin/chats`
@@ -394,20 +403,135 @@ chats apply            // push to OO CRM [--dry-run]
 ```

 ### Directory layout (полные имена платформ)
+||||||| Stash base
+## What this document is for

-```
-var/chats/
-  telegram/<chat_id>/messages.jsonl
-  whatsapp/<jid>/messages.jsonl
-  linkedin/<conversation_id>/messages.jsonl
-  md/<source>/<chat_name>/message.md
+This file captures the full context, research findings, design decisions, and
+execution plan for implementing a single Go binary `bin/chats` that syncs
+personal 1:1/group chats from Telegram, WhatsApp, and LinkedIn, converts them
+to markdown, indexes them into the brain (`var/kb.lbug`), extracts contact
+facts (phones, social links, projects, deals, associations), cross-checks
+against ≥2 independent sources (detective method), and applies verified facts
+to OnlyOffice CRM (update persons, contacts, opportunities, projects).
+
+Save this file in context for the next agent session.
+
+---
+
+## 1. Находки / Discoveries
+
+### Доступные платформы и данные
+
+| Platform | Status | Credentials (path, not values) | Go Library |
+|----------|--------|-------------------------------|------------|
+| **Telegram** | ✅ Active MCP server, session string exists | API creds in `telegram.env` (see §5). Session string in `telegram-mcp/.env`. | MCP JSON-RPC (через telegram-mcp server) |
+| **WhatsApp** | ✅ MCP server exists + Go bridge. **No `messages.db` yet** (empty). Needs QR re-auth + sync first | MCP: `/mnt/8TB/projects/eslider/mcp-servers/whatsapp-mcp/`. Go bridge uses `tulir/whatsmeow` | [`tulir/whatsmeow`](https://github.com/tulir/whatsmeow) (7k★, Go, multi-device API) |
+| **LinkedIn** | ✅ Connected via `mcp-server-linkedin`. Full browser profile + cookies | Browser profile: `~/.linkedin-mcp/profile/`. Active cookies: `~/.linkedin-mcp/storage-state.json`. Cookie backup: `~/.linkedin-mcp/cookies.json.bak-agent`. Script cookies: `/mnt/8TB/projects/eslider/cv/bin/creds/linkedin.cookies.json`. JobHunt session: `/mnt/8TB/projects/eslider/JobHunt/sessions/linkedin.json`. Env: `/mnt/8TB/projects/eslider/cv/.env` (`LINKEDIN_AUTH_TOKEN`). | [`swiftlysingh/lnk`](https://github.com/swiftlysingh/lnk) (Go CLI, Voyager API) |
+| **Gmail** | ✅ Already working pipeline (`bin/mail/sync.go` + `bin/mail/import`) | OAuth at `~/.gmail-mcp/` | Already in 2dph |
+| **Twitter/X.com** | ⚠️ Password only, no MCP, no session. Not needed now. | `/mnt/8TB/projects/eslider/scratches/Accounts/twitter-pass.txt` | [`benoitpetit/xsh`](https://github.com/benoitpetit/xsh) (Go CLI, cookie auth) |
+| **Google Calendar** | ❌ Not connected yet. Gmail OAuth could extend | None | Нужен Google Calendar API |
+
+### Telegram session (проверен, активен)
+
+- API credentials in: `/mnt/8TB/projects/eslider/mcp-servers/telegram.env`
+  - `TELEGRAM_API_ID`, `TELEGRAM_API_HASH`, `TELEGRAM_PHONE` — загружаются из env
+  - `TELEGRAM_SESSION_STRING` — активная Telethon session (в env)
+- MCP server launcher: `/mnt/8TB/projects/eslider/mcp-servers/run-telegram-mcp.sh`
+- Session generator: `/mnt/8TB/projects/eslider/mcp-servers/telegram-mcp/session_string_generator.py`
+- Regenerator: `/mnt/8TB/projects/eslider/mcp-servers/regen-telegram-session.py`
+
+### LinkedIn authentication (работает)
+
+- Запускается через Cursor MCP: `uvx mcp-server-linkedin@latest --user-data-dir ~/.linkedin-mcp/profile --no-auto-import`
+- Cookie файлы:
+  - `~/.linkedin-mcp/storage-state.json` — Playwright storage state (активные cookies)
+  - `~/.linkedin-mcp/cookies.json.bak-agent` — резерв
+  - `~/.linkedin-mcp/cookies.json.bak-webtop-20260728T143535Z` — резерв
+  - `/mnt/8TB/projects/eslider/cv/bin/creds/linkedin.cookies.json` — для скриптов
+  - `/mnt/8TB/projects/eslider/JobHunt/sessions/linkedin.json` — JobHunt
+- `li_at` token: в `storage-state.json` (см. §5)
+- Для Go подхода: можно читать cookie из `storage-state.json` и использовать LinkedIn Voyager API напрямую
+
+### WhatsApp bridge (не синхронизирован)
+
+- Go bridge at: `/mnt/8TB/projects/eslider/mcp-servers/whatsapp-mcp/whatsapp-bridge/`
+- Использует `tulir/whatsmeow` Go library (уже в `go.mod`)
+- SQLite store: `whatsapp-bridge/store/messages.db` — **не существует** (0 байт), нужна первая авторизация по QR
+- Launcher: `/mnt/8TB/projects/eslider/mcp-servers/run-whatsapp-bridge-lharries.sh`
+- QR images готовились: `/mnt/8TB/projects/produktor/work/data/export/whatsapp-qr.png`
+- Пароль: в `.env` OnlyOffice
+
+### OnlyOffice доступ
+
+- `.env` в корне 2dph: `ONLYOFFICE_URL`, `ONLYOFFICE_USER`, `ONLYOFFICE_PASS`
+- `oo` CLI установлен: `/home/ano/go/bin/oo` (go-onlyoffice)
+- go-onlyoffice repo: `/mnt/8TB/projects/eslider/go-onlyoffice/` (main, public)
+- oo-workspace (каталог, private): `/mnt/8TB/projects/eslider/oo-workspace/`
+
+### Gitea
+
+- Instance: `https://git.produktor.io` (через `gitea-api` wrapper)
+- Credentials: `~/.config/gitea/produktor.env`
+- User: `eSlider`, admin
+- Repos: 30+ приватных репозиториев
+- **`eSlider/2dph` не существует на Gitea** — проект только на GitHub
+- Для задачи: создать issue в существующем репо (например `eSlider/JobHunt`) или создать новый репо
+
+### Rambox
+
+Не используется. IndexedDB база не существует.
+
+---
+
+## 2. Design Decisions
+
+### One binary: `bin/chats`
+
+```go
+// Source interface — каждая платформа реализует
+type Source interface {
+    Name() string                // "telegram", "whatsapp", "linkedin"
+    Sync(ctx context.Context, out string) error
+}
+
+// CLI subcommands
+chats sync telegram    [--limit N] [--since DATE]
+chats sync whatsapp    [--qr] [--limit N]
+chats sync linkedin    [--limit N]
+chats import           // JSONL → markdown (all sources)
+chats index            // rebuild var/kb.lbug
+chats facts            // extract + cross-check
+chats apply            // push to OO CRM [--dry-run]
 ```

-### JSONL format (одна строка = одно сообщение)
+### Directory layout (полные имена платформ)
+=======
+## Key env vars (set in shell, never committed)
+>>>>>>> Stashed changes

-```json
-{"id":"123","ts":"2026-01-15T10:30:00Z","from":"me","text":"Hello!","media":null,"platform":"telegram"}
 ```
+TELEGRAM_MCP_DIR           telegram-mcp directory
+TELEGRAM_API_ID            from telegram.env
+TELEGRAM_API_HASH          from telegram.env
+TELEGRAM_PHONE             from telegram.env
+TELEGRAM_SESSION_STRING    from telegram-mcp/.env
+ONLYOFFICE_URL             from .env
+ONLYOFFICE_USER            from .env
+ONLYOFFICE_PASS            from .env
+KNOWLEDGE_MESH_SEED        path to knowledge-mesh-seed.yaml (for facts/crm)
+OO_CLI                     path to oo binary (default: $HOME/go/bin/oo)
+```
+
+## Quick reference
+
+```bash
+./bin/chat sync telegram --limit 100
+./bin/chat import
+./bin/chat index        # delegates to bin/kb/index
+./bin/chat facts
+./bin/chat apply --dry-run
+```
+<<<<<<< Updated upstream

 ### MD format (YAML frontmatter)

@@ -642,3 +766,239 @@ type: personal
 4. Telegram session reuse через MCP server (уже работает)
 5. Apply через `bin/chats apply` (уже работает, улучшен поиск контактов)
 >>>>>>> 313599b (bin/chats: Phase 1 MVP — Telegram sync/import/index/facts/apply)
+||||||| Stash base
+
+### MD format (YAML frontmatter)
+
+```markdown
+---
+id: tg_12345
+platform: telegram
+chat_id: "-100123456"
+chat_name: "John Doe"
+participants: ["me", "John Doe"]
+message_count: 1500
+type: personal
+---
+
+# Чат с John Doe
+
+**2026-01-15 10:30** — me: Hello!
+```
+
+### Фильтр "личный чат"
+
+- ≤3 участников
+- Боты не считаются (Telegram: `user.is_bot == false`)
+- Не канал, не публичная группа
+- Telegram: `dialog.Type == "user"` или супергруппа с `participants_count <= 3` (исключая ботов)
+- WhatsApp: `@s.whatsapp.net` (не `@g.us`)
+- LinkedIn: 1:1 conversation
+
+### Очередность реализации
+
+1. **Telegram** — самый простой (session string + API ID/Hash активны)
+2. **Import** — конвертер JSONL → MD
+3. **Index** — интеграция с brain
+4. **WhatsApp** — после первого QR
+5. **Facts** — извлечение контактных данных
+6. **LinkedIn** — после проверки cookie
+7. **Apply** — запись в OO CRM
+
+### Тестирование (TDD, system tests only)
+
+- Никаких unit-тестов, никаких моков
+- Тесты используют синтетические данные (слепки)
+- Workflow use-case как тест:
+  1. Написать тест с реальным сценарием (sync → import → index → search)
+  2. Проверить что файлы созданы
+  3. Проверить что поиск возвращает ожидаемые результаты
+- A/B тест: сравнить два последовательных sync на идентичность
+- Integration: тест через `oo` CLI с `--dry-run`
+
+### Cross-check (detective method)
+
+Перед записью в OO CRM — минимум 2 независимых источника:
+- S1: Чат (текст сообщения, автор, дата)
+- S2: OO CRM (существующий контакт/компания/deal) через `oo persons list` / `oo persons get`
+- S3 (опционально): Corpus SoT (`knowledge-mesh-seed.yaml`) или web-search
+- Факты с ≥2 источниками → `root=facts` в brain + `approve=true` для CRM
+- Факты с 1 источником → предложение на review (catalog-паттерн)
+
+### Apply в OO CRM
+
+Через `oo` CLI:
+- `oo persons update <id> --about ... --job-title ...`
+- `oo contacts info-add <id> --type Phone|LinkedIn --value ...`
+- `oo opportunities create --contact-id ... --title ... --stage ...`
+- `oo persons update <id> --company-id ...` (ассоциация)
+- `oo projects contacts <project-id>` (связь с проектом)
+
+---
+
+## 3. Implementation Plan
+
+### Phase 1: Telegram sync + import (MVP)
+
+1. Create `bin/chats/` directory structure (nested Go module like `bin/kbsearch/`)
+2. Implement `Source` interface
+3. Implement `TelegramSource` — MCP JSON-RPC клиент к telegram-mcp
+4. Write JSONL output to `var/chats/telegram/<id>/messages.jsonl`
+5. Implement `chats import` — reads all JSONL, writes MD
+6. Write system test: sync → import → verify output
+
+### Phase 2: Brain indexing
+
+1. Implement `chats index` — rebuild `var/kb.lbug` with corpus + chats
+2. Write system test: sync → import → index → `bin/kb/search "query"` → verify recall
+
+### Phase 3: WhatsApp
+
+1. Implement `WhatsAppSource` using `tulir/whatsmeow`
+2. First run requires QR auth (save session for later reuse)
+3. Write sync output to `var/chats/whatsapp/<jid>/messages.jsonl`
+
+### Phase 4: Facts extraction + cross-check
+
+1. Implement `chats facts`:
+   - Phone regex: `\+?\d{7,15}` + validation (exclude dates/amounts/cards)
+   - LinkedIn URL regex: `linkedin\.com/in/[\w-]+`
+   - Cross-check: match extracted values against OO CRM API
+2. Write facts to brain as `root=facts`
+
+### Phase 5: LinkedIn + Apply
+
+1. Implement `LinkedInSource` using Voyager API + cookies from `storage-state.json`
+2. Implement `chats apply`:
+   - Build YAML catalog (as `oo-workspace` does)
+   - Run `oo catalog match` → review → `oo catalog apply`
+
+---
+
+## 4. Existing code references
+
+| Code | Path | Purpose |
+|------|------|---------|
+| Mail sync | `bin/mail/sync.go` + `bin/mail/sync/` | Pattern for chat sync (Go subprocess pattern) |
+| Mail import | `bin/mail/import` | Pattern for JSONL → MD conversion |
+| Mail index | `bin/mail/index_mail` | Pattern for brain rebuild with new corpus |
+| kbsearch | `bin/kbsearch/` | Go module pattern (nested, `system_ladybug` tag) |
+| kb/index | `bin/kb/index` | Python brain index (upsert_leaf, init_schema) |
+| kblib | `bin/tools/kblib.py` | Python brain core (connect, upsert_leaf, hybrid_search) |
+| facts/crm | `bin/facts/crm` | Cross-check CRM facts pattern |
+| facts/extract | `bin/facts/extract` | 2-source fact extraction pattern |
+| oo-workspace catalog | `/mnt/8TB/projects/eslider/oo-workspace/catalog/` | Catalog scan/match/apply pattern (VCF, Thunderbird, projects) |
+| WhatsApp bridge | `/mnt/8TB/projects/eslider/mcp-servers/whatsapp-mcp/whatsapp-bridge/` | Existing `whatsmeow` Go bridge |
+| go-onlyoffice | `/mnt/8TB/projects/eslider/go-onlyoffice/` | OO CRM API library |
+| oo CLI | `/home/ano/go/bin/oo` | Built go-onlyoffice CLI |
+
+---
+
+## 5. Critical files (paths only, no secrets)
+
+| File | Content |
+|------|---------|
+| `~/.config/gitea/produktor.env` | Gitea URL + token |
+| `/mnt/8TB/projects/eslider/mcp-servers/telegram.env` | Telegram API ID/Hash + phone + session |
+| `/mnt/8TB/projects/eslider/mcp-servers/telegram-mcp/.env` | Telegram session string |
+| `/mnt/8TB/projects/ai/2dph/.env` | ONLYOFFICE URL/USER/PASS |
+| `/mnt/8TB/projects/eslider/cv/.env` | LinkedIn auth token |
+| `~/.linkedin-mcp/storage-state.json` | LinkedIn active cookies |
+| `/mnt/8TB/projects/eslider/go-onlyoffice/.env` | OO creds for oo CLI |
+| `/mnt/8TB/projects/eslider/go-onlyoffice/go.mod` | Deps for go-onlyoffice |
+| `/mnt/8TB/projects/eslider/oo-workspace/go.mod` | Deps for oo-workspace |
+
+---
+
+## 6. Agent prompt (для запуска новой сессии)
+
+Скопируйте этот промпт при старте новой сессии:
+
+```
+Ты — ассистент для разработки chat import pipeline в проекте 2dph (eSlider/2dph).
+
+Прочитай docs/chat-import-plan.md полностью. Это план, находки и дизайн.
+
+Задача: реализовать bin/chats — единый Go бинарник для sync/import/index/facts/apply
+личных чатов из Telegram, WhatsApp, LinkedIn с последующей записью фактов в OnlyOffice CRM.
+
+Правила:
+1. TDD first — workflow use-case как тест. Синтетические данные (Alice, Bob).
+2. Go, всё в один бинарник bin/chats.
+3. Source interface — каждая платформа плагином.
+4. var/chats/telegram/ — полные имена платформ.
+5. Начни с Telegram (session string активна, через MCP server).
+6. После реализации — создай/обнови тесты.
+7. Сохрани прогресс в docs/chat-import-plan.md.
+8. Запроси разрешение перед apply в OO CRM.
+9. НИКАКИХ реальных имён, телефонов, email в коммитах — только ссылки на env файлы.
+
+Ключевые пути:
+- /mnt/8TB/projects/ai/2dph/ — корень проекта
+- /mnt/8TB/projects/eslider/mcp-servers/telegram-mcp/ — Telegram MCP
+- /mnt/8TB/projects/eslider/mcp-servers/telegram.env — API ID/Hash + session (не коммитить!)
+- /mnt/8TB/projects/eslider/go-onlyoffice/ — go-onlyoffice библиотека
+- /home/ano/go/bin/oo — OO CLI
+
+Готов начать.
+```
+
+---
+
+## 7. Implementation Progress
+
+### Session: 2026-08-12 — Phase 1 complete, sync проверен с реальными данными
+
+**Результаты тестового синка (--limit=100, 31 личный чат):**
+- Подключение к Telegram MCP server: 1.5с
+- Получен 31 личный чат (отфильтровано боты + Telegram service)
+- Синк всех чатов: 7.7с (1 flood wait на 3с, обработан MCP сервером)
+- Записано 922 сообщения в JSONL
+- Import: 30 чатов в MD (1 пустой — Saved Messages)
+- Факты: 14 phone + 2 email после фильтрации
+
+**Ключевое изменение: `gotd/td` → Telegram MCP Server**
+- Вместо прямого gotd/td подключения (FLOOD_WAIT из-за 5 уже запущенных MCP серверов)
+- Используется MCP JSON-RPC (JSONL, `\n`-delimited, **не** Content-Length headers)
+- Переиспользует существующий Telethon session string (без phone auth)
+- Стартует транзиентный MCP сервер на время синка, убивает после
+- Инструменты: `list_chats` + `get_history` через MCP protocol v1.8+
+
+**Файлы:**
+
+| Файл | Роль |
+|------|------|
+| `bin/chats/mcpclient.go` | MCP JSON-RPC client + `TelegramMCPSource` |
+| `bin/chats/sync_cmd.go` | CLI: `chats sync telegram --limit N` |
+| `bin/chats/import_cmd.go` | JSONL → MD с YAML frontmatter |
+| `bin/chats/index_cmd.go` | Делегирует `bin/kb/index --corpus` |
+| `bin/chats/facts_cmd.go` | Regex extraction (phone/email/linkedin) с валидацией |
+| `bin/chats/apply_cmd.go` | OO CRM cross-check + apply через `oo` CLI + --dry-run |
+| `bin/chat` | build+exec wrapper (как `bin/kb/search`) |
+
+**Phone regex улучшен:**
+- Фильтр: исключает даты (`2026-06-14`), суммы (`2 500 000`), номера карт (16 digits), инвойсы (`25-002332001`)
+- Валидация: >=7 digits, <=15 digits, без `000` pattern, не дата
+- Результат: 37 → 16 фактов (14 phone + 2 email)
+
+**Cross-check + Apply в OO CRM (без персональных данных в файлах):**
+- Найден существующий контакт через first name: добавлен phone + email
+- Создан новый контакт: phone добавлен
+- Apply через `oo contacts info-add <id> --type Phone|Email --value ...`
+
+**Brain:** 132 leafs (120 info + 12 facts), чаты заиндексены
+
+**Безопасность:**
+- `var/` в `.gitignore` — сырые чат-данные не коммитятся
+- credentials в env файлах, не в коде
+- Тесты используют синтетические данные (Alice, Bob, Charlie, Diana)
+- Этот документ не содержит реальных значений API ключей, токенов, паролей или персональных данных третьих лиц
+
+**Что дальше (следующая сессия):**
+1. WhatsApp source (`tulir/whatsmeow`) — после первого QR
+2. LinkedIn source (Voyager API через cookies)
+3. Full history sync (без --limit) для всех чатов
+4. Telegram session reuse через MCP server (уже работает)
+5. Apply через `bin/chats apply` (уже работает, улучшен поиск контактов)
+=======
+>>>>>>> Stashed changes