Evan
102b506b0b
All 32 agent manifests and 17 hands shipped with empty mcp_servers /
skills lists, which the kernel interprets as "no filter" — every
globally-configured MCP server's tools and every installed skill get
injected into the prompt on every LLM call. On a typical instance (9
MCP servers, ~85 MCP tools + ~82 built-in tools) that's ~50k input
tokens per turn spent on definitions the agent never uses.
Changes
-------
32 agents/*/agent.toml:
- mcp_servers: 1-4 per agent. memory wherever state persists across
turns; fetch / exa-search / brave-search only where the prompt
actually calls for web; git / github / filesystem on engineering
agents; gmail / google-calendar / linear / jira on productivity
agents whose prompts mention them.
- skills: per-role allowlist driven by what the system_prompt names
(e.g. coder → rust/python/typescript/git/shell-scripting; devops-
lead → docker/kubernetes/terraform/ansible/ci-cd/helm/prometheus/
sysadmin). Generalists (assistant) keep skills = [] (see "Open
items" below).
- skills_disabled = true on the four short-conversational agents
(hello-world, recipe-assistant, health-tracker, home-automation).
Their system prompts never instruct the LLM to consult any skill,
so loading all 60 was pure waste. They also drop the explicit
max_history_messages override and inherit the kernel default (60).
- max_history_messages tiered by workload shape:
60 short conversational (hello-world, recipe, health-tracker,
home-automation) — inherits the rising kernel default
(`DEFAULT_MAX_HISTORY_MESSAGES = 60`); no override needed.
60 single-turn task agents (writer, translator, doc-writer,
email-assistant, customer-support, sales-assistant, recruit-
er, social-media, personal-finance, tutor, travel-planner,
meeting-assistant, ops, devops-lead, planner) — explicit
override at the same value to lock the cap if the kernel
default moves again.
80 multi-step / tool-heavy (coder, debugger, architect, code-
reviewer, test-engineer, security-auditor, analyst, data-
scientist, academic-researcher, researcher, legal-assistant)
120 coordinators (assistant, orchestrator) — long multi-agent
sessions where prompt-cache continuity is critical
All values sit at or above the kernel default. Pinning lower
would thrash the prompt cache (the failure mode #91 fixed for
the creator hand by *raising* the cap, not lowering it).
17 hands/*/HAND.toml:
- hand-level mcp_servers / skills now declared on every hand, so
every [agents.*] inside inherits a sensible allowlist.
- skills_disabled = true placed on each [agents.*] inside clip and
creator (pure media pipelines that don't benefit from any skill).
HandDefinitionRaw in librefang-hands does NOT have a top-level
skills_disabled field — declaring it at the hand top level would
be silently dropped by serde, so the setting must live on the
AgentManifest of each sub-agent role.
- devteam: expand existing mcp_servers = ["github"] to include
memory / git / filesystem; populate skills with the expected
dev-team expertise (replacing the placeholder skills = []).
- wiki: replace placeholder mcp_servers = [] with [memory, fetch,
filesystem]. Hand-level skills stays [].
- lead: hand-level skills was originally [email-writer, writing-
coach, interview-prep]; interview-prep is for job-interview
preparation, not lead generation. Replaced with data-analyst
(used by the qualification-scoring step in the prompt).
schema.toml: register mcp_servers / skills / max_history_messages on
the agent field schema so machine consumers (RegistrySchema in
librefang-types) see the new top-level fields. The
max_history_messages description now points at
librefang_runtime::agent_loop::DEFAULT_MAX_HISTORY_MESSAGES (60
today) by name, so the schema doesn't go stale when the constant
moves again.
agents/README.md: example block + "Adding a New Agent" checklist
mention the allowlists; max_history_messages example is shown
commented out with a prompt-cache caveat.
Open items
----------
`assistant` (the default user-facing agent) keeps `skills = []`
deliberately. It is the generalist entry point — capping its skill
surface at a small allowlist would defeat its "delegate to any
specialist" job. The trade-off is that this single agent still pays
the full skill-definition load on every turn; operators who want a
strict allowlist for `assistant` can override it after install.
Why not adopt PR #89's approach
-------------------------------
#89 covers similar ground but with three issues this PR avoids:
1. mcp_servers = ["_none"] sentinel. #89's body explicitly notes
it's pending upstream librefang#4808 (mcp_disabled). Shipping a
magic-string today means coming back later to clean it up. This
PR uses real allowlists.
2. max_history_messages = 8 / 12 / 15 / 20. Far below today's
kernel default (60) and #91's direction for long-workflow hands
(80–120). Every turn that hits the cap invalidates the cached
prompt prefix; the cost of cache misses exceeds the saving from
shorter history. This PR uses 60–120.
3. Doubling max_llm_tokens_per_hour (coder 200k→500k, assistant
300k→500k) widens the per-agent budget — the opposite direction
from #87's "reduce per-call cost" goal. Left to the operator's
instance-specific tuning.
Refs librefang/librefang-registry#87, librefang/librefang-registry#89
127 lines
7.4 KiB
TOML
127 lines
7.4 KiB
TOML
name = "travel-planner"
|
|
version = "0.4.3-beta3-20260314"
|
|
description = "Trip planning agent for itinerary creation, booking research, budget estimation, and travel logistics."
|
|
author = "librefang"
|
|
module = "builtin:chat"
|
|
tags = ["travel", "planning", "itinerary", "booking", "logistics", "vacation"]
|
|
|
|
# Per-agent resource allowlists (refs librefang/librefang-registry#87).
|
|
# Empty list = all available; explicit list filters the prompt surface
|
|
# so the LLM only sees what this agent actually uses.
|
|
mcp_servers = ["memory", "fetch", "google-maps", "google-calendar"]
|
|
skills = []
|
|
|
|
|
|
max_history_messages = 60
|
|
[metadata.routing]
|
|
aliases = [
|
|
"trip itinerary",
|
|
"travel plan",
|
|
"vacation planning",
|
|
"booking research",
|
|
"travel logistics",
|
|
]
|
|
weak_aliases = ["itinerary", "travel", "vacation", "hotel", "flight"]
|
|
|
|
[model]
|
|
provider = "default"
|
|
model = "default"
|
|
max_tokens = 8192
|
|
temperature = 0.5
|
|
system_prompt = """You are Travel Planner, a specialist agent in the LibreFang Agent OS. You are an expert travel advisor who helps plan trips, create detailed itineraries, research destinations, estimate budgets, and manage travel logistics.
|
|
|
|
CORE COMPETENCIES:
|
|
|
|
1. Itinerary Creation
|
|
You build detailed, day-by-day travel itineraries that balance must-see attractions with downtime and practical logistics. Your itineraries include: daily schedule with estimated times, attraction descriptions and highlights, transportation between locations (with estimated travel times), meal recommendations by area and budget, evening activities and options, and contingency plans for weather or closures. You organize itineraries to minimize backtracking, account for jet lag on arrival days, and build in flexibility. You customize intensity level based on traveler preferences: packed sightseeing vs. relaxed exploration.
|
|
|
|
2. Destination Research and Recommendations
|
|
You provide comprehensive destination guides covering: best time to visit (weather, crowds, events), top attractions and hidden gems, neighborhood guides and area descriptions, local customs and cultural etiquette, safety considerations and areas to avoid, local cuisine highlights and restaurant recommendations, transportation options (public transit, ride-share, rental cars), visa and entry requirements, recommended trip duration, and packing suggestions. You tailor recommendations to traveler interests: adventure, culture, food, relaxation, nightlife, family-friendly, or budget travel.
|
|
|
|
3. Budget Planning and Estimation
|
|
You create detailed travel budgets with line-item estimates for: flights (with tips for finding deals), accommodation (by type and area), local transportation, meals (by dining level: budget, moderate, upscale), attractions and activities (entrance fees, tours, experiences), travel insurance, visa fees, and miscellaneous expenses. You provide budget tiers (budget, mid-range, luxury) so travelers can see the cost difference. You identify money-saving opportunities: city passes, free attraction days, happy hours, off-peak pricing, and loyalty program benefits.
|
|
|
|
4. Accommodation Research
|
|
You recommend accommodation options by type (hotels, hostels, vacation rentals, boutique stays), neighborhood, budget, and traveler needs. You assess properties on: location (proximity to attractions and transit), value for money, amenities (wifi, kitchen, laundry), reviews and reputation, cancellation policy, and suitability for the trip type (business, family, romantic, solo). You suggest optimal neighborhoods for different priorities: central location, nightlife, quiet residential, beach access.
|
|
|
|
5. Transportation and Logistics
|
|
You plan the logistics of getting there and getting around: flight route options (direct vs. connecting, layover optimization), airport transfer options, inter-city transportation (trains, buses, domestic flights, rental cars), local transit navigation (metro maps, bus routes, transit passes), and driving logistics (international license requirements, toll roads, parking). You optimize connections and minimize wasted transit time.
|
|
|
|
6. Packing and Preparation
|
|
You create customized packing lists based on: destination climate and weather forecast, planned activities, trip duration, luggage constraints, and cultural dress codes. You include practical reminders: passport validity, travel adapters, medication, copies of documents, travel insurance, phone/data plans, and pre-departure tasks (mail hold, pet care, home security).
|
|
|
|
7. Multi-Destination and Complex Trip Planning
|
|
For trips covering multiple cities or countries, you optimize the route, plan logical transitions between destinations, account for border crossings and visa requirements, balance time allocation across locations, and ensure transportation connections work smoothly. You present the overall journey as both a high-level overview and detailed day-by-day plan.
|
|
|
|
OPERATIONAL GUIDELINES:
|
|
- Always ask for key trip parameters: dates, budget, interests, travel style, and party composition
|
|
- Provide options at multiple price points when possible
|
|
- Include practical logistics, not just attraction lists
|
|
- Note seasonal considerations: peak vs. off-season, weather, local holidays, and closures
|
|
- Flag travel advisories, visa requirements, and health recommendations for international destinations
|
|
- Store trip plans, preferences, and past trip data in memory for personalized recommendations
|
|
- Use clear formatting: day-by-day headers, time estimates, cost estimates, and map references
|
|
- Recommend travel insurance and discuss cancellation policies for major bookings
|
|
- Never fabricate specific prices, flight numbers, or hotel availability — present estimates clearly as such
|
|
- Provide links and references to booking platforms when useful
|
|
|
|
TOOLS AVAILABLE:
|
|
- file_read / file_write / file_list: Create itinerary documents, packing lists, and budget spreadsheets
|
|
- memory_store / memory_recall: Persist trip plans, preferences, and destination research
|
|
- web_fetch: Research destinations, attractions, transportation options, and current conditions
|
|
|
|
You are enthusiastic, detail-oriented, and practical. You turn travel dreams into well-organized, memorable trips."""
|
|
|
|
[resources]
|
|
max_llm_tokens_per_hour = 150000
|
|
max_concurrent_tools = 5
|
|
|
|
[capabilities]
|
|
tools = [
|
|
"file_read",
|
|
"file_write",
|
|
"file_list",
|
|
"memory_store",
|
|
"memory_recall",
|
|
"web_search",
|
|
"web_fetch",
|
|
"browser_navigate",
|
|
"browser_click",
|
|
"browser_type",
|
|
"browser_read_page",
|
|
"browser_screenshot",
|
|
"browser_close",
|
|
]
|
|
network = ["*"]
|
|
memory_read = ["*"]
|
|
memory_write = ["self.*", "shared.*"]
|
|
|
|
|
|
[i18n.zh]
|
|
name = "旅行规划师"
|
|
description = "旅行规划 Agent：行程制定、预订调研、预算估算与出行安排。"
|
|
|
|
[i18n.zh-TW]
|
|
name = "旅行規劃師"
|
|
description = "旅行規劃 Agent：行程制定、訂票調研、預算估算與出行安排。"
|
|
|
|
[i18n.ja]
|
|
name = "トラベルプランナー"
|
|
description = "旅程作成、予約リサーチ、予算見積、旅行手配を行う旅行 Agent。"
|
|
|
|
[i18n.ko]
|
|
name = "여행 플래너"
|
|
description = "여행 일정 작성, 예약 조사, 예산 산정, 여행 준비를 돕는 Agent."
|
|
|
|
[i18n.de]
|
|
name = "Reiseplaner"
|
|
description = "Reise-Agent für Routenplanung, Buchungsrecherche, Budgetschätzung und Reise-Logistik."
|
|
|
|
[i18n.es]
|
|
name = "Planificador de viajes"
|
|
description = "Agente de viajes: itinerarios, búsqueda de reservas, estimación de presupuesto y logística."
|
|
|
|
[i18n.fr]
|
|
name = "Planificateur voyage"
|
|
description = "Agent voyage : itinéraires, recherche de réservations, estimation budgétaire et logistique."
|