multi-websearch-mcp 0.3.0rc2__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- multi_websearch_mcp-0.3.0rc2/.dockerignore +22 -0
- multi_websearch_mcp-0.3.0rc2/.env.example +126 -0
- multi_websearch_mcp-0.3.0rc2/.env.example.de +126 -0
- multi_websearch_mcp-0.3.0rc2/.gitignore +15 -0
- multi_websearch_mcp-0.3.0rc2/ANALYSE.md +507 -0
- multi_websearch_mcp-0.3.0rc2/CHANGELOG.md +77 -0
- multi_websearch_mcp-0.3.0rc2/Dockerfile +32 -0
- multi_websearch_mcp-0.3.0rc2/LICENSE +21 -0
- multi_websearch_mcp-0.3.0rc2/PKG-INFO +191 -0
- multi_websearch_mcp-0.3.0rc2/README.de.md +602 -0
- multi_websearch_mcp-0.3.0rc2/README.md +145 -0
- multi_websearch_mcp-0.3.0rc2/build-requirements.in +2 -0
- multi_websearch_mcp-0.3.0rc2/build-requirements.lock +16 -0
- multi_websearch_mcp-0.3.0rc2/deploy/bwrap/searxng-bwrap +186 -0
- multi_websearch_mcp-0.3.0rc2/deploy/bwrap/websearch-bwrap +306 -0
- multi_websearch_mcp-0.3.0rc2/deploy/lib/deploytool.py +310 -0
- multi_websearch_mcp-0.3.0rc2/deploy/lib/sandbox.sh +230 -0
- multi_websearch_mcp-0.3.0rc2/deploy/podman/install.sh +181 -0
- multi_websearch_mcp-0.3.0rc2/deploy/podman/searxng-cache.volume +2 -0
- multi_websearch_mcp-0.3.0rc2/deploy/podman/websearch-data.volume +3 -0
- multi_websearch_mcp-0.3.0rc2/deploy/podman/websearch-mcp.container +39 -0
- multi_websearch_mcp-0.3.0rc2/deploy/podman/websearch-monitor.container +37 -0
- multi_websearch_mcp-0.3.0rc2/deploy/podman/websearch-searxng.container +43 -0
- multi_websearch_mcp-0.3.0rc2/deploy/podman/websearch.network +6 -0
- multi_websearch_mcp-0.3.0rc2/deploy/systemd/install.sh +78 -0
- multi_websearch_mcp-0.3.0rc2/deploy/systemd/websearch-mcp.service +16 -0
- multi_websearch_mcp-0.3.0rc2/deploy/systemd/websearch-monitor.service +17 -0
- multi_websearch_mcp-0.3.0rc2/deploy/systemd/websearch-searxng.service +15 -0
- multi_websearch_mcp-0.3.0rc2/docker-compose.yml +120 -0
- multi_websearch_mcp-0.3.0rc2/docs/QUICKSTART.md +143 -0
- multi_websearch_mcp-0.3.0rc2/docs/USAGE.md +556 -0
- multi_websearch_mcp-0.3.0rc2/pyproject.toml +89 -0
- multi_websearch_mcp-0.3.0rc2/requirements.lock +196 -0
- multi_websearch_mcp-0.3.0rc2/searxng/settings.yml +95 -0
- multi_websearch_mcp-0.3.0rc2/server.json +43 -0
- multi_websearch_mcp-0.3.0rc2/tests/conftest.py +10 -0
- multi_websearch_mcp-0.3.0rc2/tests/mcp_smoke.py +81 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_cli.py +146 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_core.py +1125 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_deploy.py +918 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_envfile.py +94 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_features.py +713 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_fetch.py +287 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_fetch_pinning.py +315 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_gnews.py +339 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_i18n.py +63 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_mcp.py +450 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_monitor.py +1166 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_notify.py +457 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_providers.py +620 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_reddit.py +246 -0
- multi_websearch_mcp-0.3.0rc2/tests/test_setup_cmds.py +711 -0
- multi_websearch_mcp-0.3.0rc2/watches.example.json +6 -0
- multi_websearch_mcp-0.3.0rc2/websearch/__init__.py +21 -0
- multi_websearch_mcp-0.3.0rc2/websearch/__main__.py +546 -0
- multi_websearch_mcp-0.3.0rc2/websearch/core.py +856 -0
- multi_websearch_mcp-0.3.0rc2/websearch/doctor.py +554 -0
- multi_websearch_mcp-0.3.0rc2/websearch/engine.py +839 -0
- multi_websearch_mcp-0.3.0rc2/websearch/envfile.py +146 -0
- multi_websearch_mcp-0.3.0rc2/websearch/fetch.py +537 -0
- multi_websearch_mcp-0.3.0rc2/websearch/gnews.py +404 -0
- multi_websearch_mcp-0.3.0rc2/websearch/i18n.py +48 -0
- multi_websearch_mcp-0.3.0rc2/websearch/mcp_server.py +475 -0
- multi_websearch_mcp-0.3.0rc2/websearch/metrics.py +72 -0
- multi_websearch_mcp-0.3.0rc2/websearch/monitor.py +926 -0
- multi_websearch_mcp-0.3.0rc2/websearch/notify.py +666 -0
- multi_websearch_mcp-0.3.0rc2/websearch/providers.py +1013 -0
- multi_websearch_mcp-0.3.0rc2/websearch/py.typed +0 -0
- multi_websearch_mcp-0.3.0rc2/websearch/reddit.py +222 -0
- multi_websearch_mcp-0.3.0rc2/websearch/wizard.py +527 -0
|
@@ -0,0 +1,22 @@
|
|
|
1
|
+
# Nur das Nötige in den Build-Kontext (schneller Build, keine Geheimnisse im Image-Kontext)
|
|
2
|
+
.env
|
|
3
|
+
.env.*
|
|
4
|
+
!.env.example
|
|
5
|
+
!.env.example.de
|
|
6
|
+
.venv
|
|
7
|
+
venv
|
|
8
|
+
**/__pycache__
|
|
9
|
+
**/*.pyc
|
|
10
|
+
.pytest_cache
|
|
11
|
+
.mypy_cache
|
|
12
|
+
.git
|
|
13
|
+
.github
|
|
14
|
+
dist
|
|
15
|
+
build
|
|
16
|
+
*.egg-info
|
|
17
|
+
watches
|
|
18
|
+
searxng
|
|
19
|
+
tests
|
|
20
|
+
deploy
|
|
21
|
+
*.db
|
|
22
|
+
*.db-*
|
|
@@ -0,0 +1,126 @@
|
|
|
1
|
+
# Copy to .env and adjust - or simply run: websearch init --contact you@your-company.com
|
|
2
|
+
# docker compose reads this file itself; for local runs load it with: set -a; . ./.env; set +a
|
|
3
|
+
# Values: $VAR / ${VAR} are expanded in unquoted and "double-quoted" values (like compose and the shell) -
|
|
4
|
+
# put passwords or tokens containing $ in 'single quotes'.
|
|
5
|
+
# German version of this file: .env.example.de
|
|
6
|
+
SEARXNG_SECRET=
|
|
7
|
+
# SearXNG image tag (default: the tested version; newer tags: hub.docker.com/r/searxng/searxng/tags, or "latest")
|
|
8
|
+
# SEARXNG_VERSION=2026.11.5-3e95f6cd2
|
|
9
|
+
# Podman Quadlet only: host port of SearXNG (default 8888)
|
|
10
|
+
# SEARXNG_PORT=8888
|
|
11
|
+
# Wikimedia/GDELT require a contact in the User-Agent
|
|
12
|
+
WEBSEARCH_UA="websearch/0.3.0 (contact@your-company.example)"
|
|
13
|
+
# Language of CLI output, logs and notifications: de | en (default: system language)
|
|
14
|
+
WEBSEARCH_LANG=en
|
|
15
|
+
# Own free Marginalia key: mail to contact@marginalia-search.com
|
|
16
|
+
MARGINALIA_KEY=public
|
|
17
|
+
# Disable sources, comma-separated (e.g. bing_rss for commercial use)
|
|
18
|
+
WEBSEARCH_DISABLE=
|
|
19
|
+
# Log level: DEBUG, INFO, WARNING (default), ERROR
|
|
20
|
+
WEBSEARCH_LOG_LEVEL=
|
|
21
|
+
|
|
22
|
+
# --- Paths and services: Docker sets these itself - uncomment only for local runs ----------------
|
|
23
|
+
# SearXNG instance (Docker: http://searxng:8080, local: http://127.0.0.1:8888)
|
|
24
|
+
# SEARXNG_URL=http://127.0.0.1:8888
|
|
25
|
+
# SQLite files: cache + host-wide pauses/rate slots/paid counters/metrics, and monitoring state
|
|
26
|
+
# (all processes of a host should share the same cache file)
|
|
27
|
+
# WEBSEARCH_CACHE=
|
|
28
|
+
# WEBSEARCH_MONITOR_DB=
|
|
29
|
+
# MCP server over HTTP: bind address, port, allowed Host headers (DNS rebinding protection)
|
|
30
|
+
# MCP_HOST=127.0.0.1
|
|
31
|
+
# MCP_PORT=8765
|
|
32
|
+
# MCP_ALLOWED_HOSTS=127.0.0.1:*,localhost:*,[::1]:*
|
|
33
|
+
|
|
34
|
+
# --- MCP server over HTTP: access protection -------------------------------------------------------
|
|
35
|
+
# If set, every request to /mcp and /metrics needs "Authorization: Bearer <token>" (/healthz stays open).
|
|
36
|
+
# Required as soon as the port is reachable beyond 127.0.0.1. Generate:
|
|
37
|
+
# sed -i "s/^MCP_AUTH_TOKEN=.*/MCP_AUTH_TOKEN=$(openssl rand -hex 32)/" .env
|
|
38
|
+
MCP_AUTH_TOKEN=
|
|
39
|
+
# Remote access additionally: allow the public host name (DNS rebinding protection stays on), with and without port
|
|
40
|
+
# MCP_ALLOWED_HOSTS=127.0.0.1:*,localhost:*,[::1]:*,search.example.org,search.example.org:*
|
|
41
|
+
# 1 = allow listening on 0.0.0.0 without a token inside a container (default 0: without MCP_AUTH_TOKEN the
|
|
42
|
+
# containerized MCP server does not start). Only if the port is published on 127.0.0.1 on a single-user
|
|
43
|
+
# machine; a public name in MCP_ALLOWED_HOSTS always needs a token
|
|
44
|
+
# MCP_ALLOW_NO_AUTH=
|
|
45
|
+
# force container detection (1/0); otherwise /.dockerenv (Docker) or /run/.containerenv (Podman)
|
|
46
|
+
# WEBSEARCH_IN_CONTAINER=
|
|
47
|
+
# 0 = MCP agents may never use paid APIs
|
|
48
|
+
WEBSEARCH_MCP_PAID=1
|
|
49
|
+
|
|
50
|
+
# --- Docker image -----------------------------------------------------------------------------------
|
|
51
|
+
# Prebuilt image instead of a local build (docker compose pull && docker compose up -d --no-build)
|
|
52
|
+
# WEBSEARCH_IMAGE=ghcr.io/c0decave/multi-websearch-mcp:0.3.0rc2
|
|
53
|
+
|
|
54
|
+
# --- Proxies (optional, see ANALYSE.md section 3.1) -------------------------------------------------
|
|
55
|
+
# for the own public HTTP sources (Bing RSS, Google News, GDELT, HN, Wikipedia, Marginalia) and fetch_url -
|
|
56
|
+
# never for your SearXNG instance, the paid APIs and Reddit (keys/credentials do not go through a proxy)
|
|
57
|
+
# WEBSEARCH_PROXY=
|
|
58
|
+
# for the ddgs sources (evaluated by ddgs itself); SearXNG: outgoing.proxies in searxng/settings.yml
|
|
59
|
+
# DDGS_PROXY=
|
|
60
|
+
|
|
61
|
+
# --- Reddit (official API, free for "script" apps, ~100 requests/min) ------------------------------
|
|
62
|
+
# Create an app at https://www.reddit.com/prefs/apps (type "script"); the source is active only with both values
|
|
63
|
+
REDDIT_CLIENT_ID=
|
|
64
|
+
REDDIT_CLIENT_SECRET=
|
|
65
|
+
# optional: Reddit asks for "<platform>:<app-id>:<version> (by /u/<name>)" (default: WEBSEARCH_UA)
|
|
66
|
+
# REDDIT_USER_AGENT=
|
|
67
|
+
|
|
68
|
+
# --- Optional paid/free-tier APIs: last resort only ------------------------------------------------
|
|
69
|
+
# Order: exa -> linkup -> tavily -> brave_api -> perplexity -> serper; used until ONE answers.
|
|
70
|
+
# Paid sources are always asked for at most 10 results (more results cost extra with some).
|
|
71
|
+
# Exa: 10 $ free credit/month, no payment method; mode "fast" (Lite tier) is the default
|
|
72
|
+
EXA_API_KEY=
|
|
73
|
+
EXA_SEARCH_TYPE=fast
|
|
74
|
+
# Linkup: monthly free quota, then about 5 $/1,000
|
|
75
|
+
LINKUP_API_KEY=
|
|
76
|
+
# Tavily: 1,000 credits/month free, no card
|
|
77
|
+
TAVILY_API_KEY=
|
|
78
|
+
# Brave: 5 $ free credit/month (~1,000 queries), credit card required
|
|
79
|
+
BRAVE_API_KEY=
|
|
80
|
+
# Perplexity Search API: mode "fast" = 1 $/1,000 (default), "pro" = standard mode 5 $/1,000
|
|
81
|
+
PERPLEXITY_API_KEY=
|
|
82
|
+
PERPLEXITY_SEARCH_TYPE=fast
|
|
83
|
+
# Serper (Google results): 2,500 free, then prepaid from ~1 $/1,000
|
|
84
|
+
SERPER_API_KEY=
|
|
85
|
+
# Hard cost limit: max. paid queries per day / month (UTC, counted host-wide, see "websearch status").
|
|
86
|
+
# Also set a spending limit in the provider's dashboard (Brave keeps billing the credit card after the
|
|
87
|
+
# free credit without warning).
|
|
88
|
+
WEBSEARCH_PAID_DAILY_LIMIT=100
|
|
89
|
+
WEBSEARCH_PAID_MONTHLY_LIMIT=1000
|
|
90
|
+
|
|
91
|
+
# --- fetch_url (page retrieval for agents) ---------------------------------------------------------
|
|
92
|
+
# 1 = also allow targets in the internal network (intranet use only!)
|
|
93
|
+
WEBSEARCH_FETCH_ALLOW_PRIVATE=0
|
|
94
|
+
# 1 = ignore robots.txt (not recommended)
|
|
95
|
+
WEBSEARCH_FETCH_IGNORE_ROBOTS=0
|
|
96
|
+
# 0 = do not resolve Google News redirect links (fetch_url then refuses them with a hint)
|
|
97
|
+
# WEBSEARCH_GNEWS_DECODE=1
|
|
98
|
+
|
|
99
|
+
# --- Monitoring ------------------------------------------------------------------------------------
|
|
100
|
+
# Target for new results and blind warnings: self-hosted ntfy, n8n, Slack, Discord, Teams, Telegram ...
|
|
101
|
+
WEBSEARCH_WEBHOOK=
|
|
102
|
+
# json (default) | text (ntfy) | slack | discord | teams (Workflows webhook) | telegram | email
|
|
103
|
+
WEBSEARCH_WEBHOOK_FORMAT=json
|
|
104
|
+
# telegram only: chat id (number, groups negative) or @channelname; WEBSEARCH_WEBHOOK = https://api.telegram.org/bot<TOKEN>/sendMessage
|
|
105
|
+
# WEBSEARCH_TELEGRAM_CHAT_ID=
|
|
106
|
+
# email only (WEBSEARCH_WEBHOOK is not needed then): SMTP server
|
|
107
|
+
# WEBSEARCH_SMTP_HOST=
|
|
108
|
+
# port (default 587, 465 with WEBSEARCH_SMTP_TLS=ssl)
|
|
109
|
+
# WEBSEARCH_SMTP_PORT=587
|
|
110
|
+
# starttls (default) | ssl | none
|
|
111
|
+
# WEBSEARCH_SMTP_TLS=starttls
|
|
112
|
+
# login (optional, only together with the password)
|
|
113
|
+
# WEBSEARCH_SMTP_USER=
|
|
114
|
+
# WEBSEARCH_SMTP_PASSWORD=
|
|
115
|
+
# sender (default: WEBSEARCH_SMTP_USER if that is an address)
|
|
116
|
+
# WEBSEARCH_SMTP_FROM=
|
|
117
|
+
# recipients, comma-separated
|
|
118
|
+
# WEBSEARCH_SMTP_TO=
|
|
119
|
+
# send a warning after this many "blind" runs in a row (0 results because of errors)
|
|
120
|
+
WEBSEARCH_BLIND_AFTER=3
|
|
121
|
+
# at most this many watches run at the same time (spreads the load over the search engines)
|
|
122
|
+
WEBSEARCH_MONITOR_CONCURRENCY=2
|
|
123
|
+
# optional: additionally append every alert (results and blind/recovered warnings) as a JSON line to this file
|
|
124
|
+
# WEBSEARCH_MONITOR_OUT=
|
|
125
|
+
# optional: file updated after each monitor pass (Docker healthcheck)
|
|
126
|
+
# WEBSEARCH_HEARTBEAT=
|
|
@@ -0,0 +1,126 @@
|
|
|
1
|
+
# Kopieren nach .env und anpassen - oder einfach: websearch init --contact du@eure-firma.de
|
|
2
|
+
# docker compose liest die Datei selbst; lokal laden mit: set -a; . ./.env; set +a
|
|
3
|
+
# Werte: $VAR / ${VAR} werden in unquotierten und "doppelt" quotierten Werten ersetzt (wie compose und die Shell) -
|
|
4
|
+
# Passwörter oder Tokens mit $ in 'einfache Anführungszeichen' setzen.
|
|
5
|
+
# Englische Fassung dieser Datei: .env.example
|
|
6
|
+
SEARXNG_SECRET=
|
|
7
|
+
# SearXNG-Image-Tag (Standard: die getestete Version; neuere Tags: hub.docker.com/r/searxng/searxng/tags, oder "latest")
|
|
8
|
+
# SEARXNG_VERSION=2026.11.5-3e95f6cd2
|
|
9
|
+
# nur Podman-Quadlet: Host-Port von SearXNG (Standard 8888)
|
|
10
|
+
# SEARXNG_PORT=8888
|
|
11
|
+
# Wikimedia/GDELT verlangen einen Kontakt im User-Agent
|
|
12
|
+
WEBSEARCH_UA="websearch/0.3.0 (kontakt@eure-firma.example)"
|
|
13
|
+
# Sprache von Kommandozeile, Logs und Benachrichtigungen: de | en (Standard: Systemsprache)
|
|
14
|
+
WEBSEARCH_LANG=de
|
|
15
|
+
# Eigener kostenloser Marginalia-Key: Mail an contact@marginalia-search.com
|
|
16
|
+
MARGINALIA_KEY=public
|
|
17
|
+
# Quellen abschalten, kommagetrennt (z.B. bing_rss bei kommerzieller Nutzung)
|
|
18
|
+
WEBSEARCH_DISABLE=
|
|
19
|
+
# Log-Level: DEBUG, INFO, WARNING (Standard), ERROR
|
|
20
|
+
WEBSEARCH_LOG_LEVEL=
|
|
21
|
+
|
|
22
|
+
# --- Pfade und Dienste: Docker setzt diese selbst - nur für lokale Läufe einkommentieren -------------
|
|
23
|
+
# SearXNG-Instanz (Docker: http://searxng:8080, lokal: http://127.0.0.1:8888)
|
|
24
|
+
# SEARXNG_URL=http://127.0.0.1:8888
|
|
25
|
+
# SQLite-Dateien: Cache + hostweite Pausen/Tempo/Bezahl-Zähler/Metriken bzw. Monitoring-Zustand
|
|
26
|
+
# (alle Prozesse eines Hosts sollten dieselbe Cache-Datei nutzen)
|
|
27
|
+
# WEBSEARCH_CACHE=
|
|
28
|
+
# WEBSEARCH_MONITOR_DB=
|
|
29
|
+
# MCP-Server über HTTP: Bind-Adresse, Port, erlaubte Host-Header (DNS-Rebinding-Schutz)
|
|
30
|
+
# MCP_HOST=127.0.0.1
|
|
31
|
+
# MCP_PORT=8765
|
|
32
|
+
# MCP_ALLOWED_HOSTS=127.0.0.1:*,localhost:*,[::1]:*
|
|
33
|
+
|
|
34
|
+
# --- MCP-Server über HTTP: Zugriffsschutz ---------------------------------------------------------
|
|
35
|
+
# Gesetzt -> jede Anfrage an /mcp und /metrics braucht "Authorization: Bearer <token>" (/healthz bleibt offen).
|
|
36
|
+
# Pflicht, sobald der Port über 127.0.0.1 hinaus erreichbar ist. Erzeugen:
|
|
37
|
+
# sed -i "s/^MCP_AUTH_TOKEN=.*/MCP_AUTH_TOKEN=$(openssl rand -hex 32)/" .env
|
|
38
|
+
MCP_AUTH_TOKEN=
|
|
39
|
+
# Fernzugriff zusätzlich: öffentlichen Hostnamen erlauben (DNS-Rebinding-Schutz bleibt aktiv), mit und ohne Port
|
|
40
|
+
# MCP_ALLOWED_HOSTS=127.0.0.1:*,localhost:*,[::1]:*,suche.example.org,suche.example.org:*
|
|
41
|
+
# 1 = im Container ohne Token auf 0.0.0.0 lauschen erlauben (Standard 0: ohne MCP_AUTH_TOKEN startet der
|
|
42
|
+
# MCP-Server im Container nicht). Nur, wenn der Port auf 127.0.0.1 eines Einzelplatzrechners veröffentlicht
|
|
43
|
+
# ist; ein öffentlicher Name in MCP_ALLOWED_HOSTS braucht immer ein Token
|
|
44
|
+
# MCP_ALLOW_NO_AUTH=
|
|
45
|
+
# Container-Erkennung erzwingen (1/0); sonst /.dockerenv (Docker) bzw. /run/.containerenv (Podman)
|
|
46
|
+
# WEBSEARCH_IN_CONTAINER=
|
|
47
|
+
# 0 = MCP-Agents dürfen Bezahl-APIs nie nutzen
|
|
48
|
+
WEBSEARCH_MCP_PAID=1
|
|
49
|
+
|
|
50
|
+
# --- Docker-Image ---------------------------------------------------------------------------------
|
|
51
|
+
# Fertiges Image statt lokalem Build (docker compose pull && docker compose up -d --no-build)
|
|
52
|
+
# WEBSEARCH_IMAGE=ghcr.io/c0decave/multi-websearch-mcp:0.3.0rc2
|
|
53
|
+
|
|
54
|
+
# --- Proxys (optional, siehe ANALYSE.md Abschnitt 3.1) --------------------------------------------
|
|
55
|
+
# für die eigenen öffentlichen HTTP-Quellen (Bing-RSS, Google News, GDELT, HN, Wikipedia, Marginalia) und
|
|
56
|
+
# fetch_url - nie für die eigene SearXNG-Instanz, die Bezahl-APIs und Reddit (Keys/Zugangsdaten nicht über Proxys)
|
|
57
|
+
# WEBSEARCH_PROXY=
|
|
58
|
+
# für die ddgs-Quellen (wertet ddgs selbst aus); SearXNG: outgoing.proxies in searxng/settings.yml
|
|
59
|
+
# DDGS_PROXY=
|
|
60
|
+
|
|
61
|
+
# --- Reddit (offizielle API, kostenlos für "script"-Apps, ~100 Anfragen/min) ---------------------
|
|
62
|
+
# App anlegen unter https://www.reddit.com/prefs/apps (Typ "script"); Quelle ist nur mit beiden Werten aktiv
|
|
63
|
+
REDDIT_CLIENT_ID=
|
|
64
|
+
REDDIT_CLIENT_SECRET=
|
|
65
|
+
# optional: Reddit wünscht "<plattform>:<app-id>:<version> (by /u/<name>)" (Standard: WEBSEARCH_UA)
|
|
66
|
+
# REDDIT_USER_AGENT=
|
|
67
|
+
|
|
68
|
+
# --- Optionale Bezahl-/Freikontingent-APIs: nur letzte Reserve -----------------------------------
|
|
69
|
+
# Reihenfolge: exa -> linkup -> tavily -> brave_api -> perplexity -> serper; genutzt wird, bis EINE liefert.
|
|
70
|
+
# Bezahlquellen werden immer mit max. 10 Treffern gefragt (weitere Treffer kosten bei manchen extra).
|
|
71
|
+
# Exa: 10 $ Freiguthaben/Monat, ohne Zahlungsmittel; Modus "fast" (Lite-Tarif) ist Standard
|
|
72
|
+
EXA_API_KEY=
|
|
73
|
+
EXA_SEARCH_TYPE=fast
|
|
74
|
+
# Linkup: monatliches Gratiskontingent, danach ca. 5 $/1.000
|
|
75
|
+
LINKUP_API_KEY=
|
|
76
|
+
# Tavily: 1.000 Credits/Monat gratis, ohne Karte
|
|
77
|
+
TAVILY_API_KEY=
|
|
78
|
+
# Brave: 5 $ Freiguthaben/Monat (~1.000 Anfragen), Kreditkarte nötig
|
|
79
|
+
BRAVE_API_KEY=
|
|
80
|
+
# Perplexity Search API: Modus "fast" = 1 $/1.000 (Standard), "pro" = Standardmodus 5 $/1.000
|
|
81
|
+
PERPLEXITY_API_KEY=
|
|
82
|
+
PERPLEXITY_SEARCH_TYPE=fast
|
|
83
|
+
# Serper (Google-Ergebnisse): 2.500 gratis, danach Prepaid ab ~1 $/1.000
|
|
84
|
+
SERPER_API_KEY=
|
|
85
|
+
# Harte Kostengrenze: max. Bezahl-Abfragen pro Tag / Monat (UTC, hostweit gezählt, siehe "websearch status").
|
|
86
|
+
# Zusätzlich im Dashboard des Anbieters ein Ausgabelimit setzen (Brave bucht nach dem Freiguthaben
|
|
87
|
+
# ohne Warnung per Kreditkarte weiter).
|
|
88
|
+
WEBSEARCH_PAID_DAILY_LIMIT=100
|
|
89
|
+
WEBSEARCH_PAID_MONTHLY_LIMIT=1000
|
|
90
|
+
|
|
91
|
+
# --- fetch_url (Seitenabruf für Agents) ----------------------------------------------------------
|
|
92
|
+
# 1 = auch Ziele im internen Netz erlauben (nur für Intranet-Nutzung!)
|
|
93
|
+
WEBSEARCH_FETCH_ALLOW_PRIVATE=0
|
|
94
|
+
# 1 = robots.txt ignorieren (nicht empfohlen)
|
|
95
|
+
WEBSEARCH_FETCH_IGNORE_ROBOTS=0
|
|
96
|
+
# 0 = Google-News-Weiterleitungen nicht auflösen (fetch_url lehnt sie dann mit Hinweis ab)
|
|
97
|
+
# WEBSEARCH_GNEWS_DECODE=1
|
|
98
|
+
|
|
99
|
+
# --- Monitoring ------------------------------------------------------------------------------------
|
|
100
|
+
# Ziel für neue Treffer und Blind-Warnungen: selbst gehostetes ntfy, n8n, Slack, Discord, Teams, Telegram ...
|
|
101
|
+
WEBSEARCH_WEBHOOK=
|
|
102
|
+
# json (Standard) | text (ntfy) | slack | discord | teams (Workflows-Webhook) | telegram | email
|
|
103
|
+
WEBSEARCH_WEBHOOK_FORMAT=json
|
|
104
|
+
# nur telegram: Chat-ID (Zahl, Gruppen negativ) oder @kanalname; WEBSEARCH_WEBHOOK = https://api.telegram.org/bot<TOKEN>/sendMessage
|
|
105
|
+
# WEBSEARCH_TELEGRAM_CHAT_ID=
|
|
106
|
+
# nur email (WEBSEARCH_WEBHOOK wird dann nicht gebraucht): SMTP-Server
|
|
107
|
+
# WEBSEARCH_SMTP_HOST=
|
|
108
|
+
# Port (Standard 587, bei WEBSEARCH_SMTP_TLS=ssl 465)
|
|
109
|
+
# WEBSEARCH_SMTP_PORT=587
|
|
110
|
+
# starttls (Standard) | ssl | none
|
|
111
|
+
# WEBSEARCH_SMTP_TLS=starttls
|
|
112
|
+
# Login (optional, nur zusammen mit dem Passwort)
|
|
113
|
+
# WEBSEARCH_SMTP_USER=
|
|
114
|
+
# WEBSEARCH_SMTP_PASSWORD=
|
|
115
|
+
# Absender (Standard: WEBSEARCH_SMTP_USER, falls das eine Adresse ist)
|
|
116
|
+
# WEBSEARCH_SMTP_FROM=
|
|
117
|
+
# Empfänger, kommagetrennt
|
|
118
|
+
# WEBSEARCH_SMTP_TO=
|
|
119
|
+
# nach so vielen "blinden" Läufen in Folge (0 Treffer wegen Fehlern) eine Warnung senden
|
|
120
|
+
WEBSEARCH_BLIND_AFTER=3
|
|
121
|
+
# so viele Watches laufen höchstens gleichzeitig (verteilt die Last auf die Suchmaschinen)
|
|
122
|
+
WEBSEARCH_MONITOR_CONCURRENCY=2
|
|
123
|
+
# optional: jede Meldung (Treffer und Blind-/Recovered-Warnungen) zusätzlich als JSON-Zeile in diese Datei
|
|
124
|
+
# WEBSEARCH_MONITOR_OUT=
|
|
125
|
+
# optional: Datei, die nach jedem Monitor-Durchlauf aktualisiert wird (Docker-Healthcheck)
|
|
126
|
+
# WEBSEARCH_HEARTBEAT=
|