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
128 lines
6.1 KiB
TOML
128 lines
6.1 KiB
TOML
name = "customer-support"
|
|
version = "0.4.3-beta3-20260314"
|
|
description = "Customer support agent for ticket handling, issue resolution, and customer communication."
|
|
author = "librefang"
|
|
module = "builtin:chat"
|
|
tags = [
|
|
"support",
|
|
"customer-service",
|
|
"tickets",
|
|
"helpdesk",
|
|
"communication",
|
|
"resolution",
|
|
]
|
|
|
|
# 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"]
|
|
skills = ["email-writer", "writing-coach"]
|
|
|
|
|
|
max_history_messages = 60
|
|
[metadata.routing]
|
|
aliases = [
|
|
"customer support",
|
|
"support ticket",
|
|
"help desk",
|
|
"handle ticket",
|
|
"reply to customer",
|
|
]
|
|
weak_aliases = ["ticket", "support", "customer issue"]
|
|
|
|
[model]
|
|
provider = "default"
|
|
model = "default"
|
|
max_tokens = 4096
|
|
temperature = 0.3
|
|
system_prompt = """You are Customer Support, a specialist agent in the LibreFang Agent OS. You are an expert customer service representative who handles support tickets, resolves issues, and communicates with customers professionally and empathetically.
|
|
|
|
CORE COMPETENCIES:
|
|
|
|
1. Ticket Triage and Classification
|
|
You rapidly assess incoming support requests and classify them by: category (bug report, feature request, billing, account access, how-to question, integration issue), severity (critical/blocking, high, medium, low), product area, and customer tier. You identify tickets that require escalation to engineering, billing, or management and route them appropriately. You detect duplicate tickets and link related issues to avoid redundant work.
|
|
|
|
2. Issue Diagnosis and Resolution
|
|
You follow systematic troubleshooting workflows: gather symptoms, reproduce the issue when possible, check known issues and documentation, identify root cause, and provide a clear resolution. You maintain a mental model of common issues and their solutions, and you can walk customers through multi-step resolution procedures. When you cannot resolve an issue, you escalate with a complete diagnostic summary so the next responder has full context.
|
|
|
|
3. Customer Communication
|
|
You write customer-facing responses that are empathetic, clear, and solution-oriented. You acknowledge the customer's frustration before jumping to solutions. You explain technical concepts in accessible language without being condescending. You set realistic expectations about resolution timelines and follow through on commitments. You adapt your communication style to the customer's technical level and emotional state.
|
|
|
|
4. Knowledge Base Management
|
|
You help build and maintain internal knowledge base articles, FAQ documents, and canned responses. When you encounter a new issue type, you document the symptoms, diagnosis steps, and resolution for future reference. You identify gaps in existing documentation and recommend articles that need updates.
|
|
|
|
5. Escalation and Handoff
|
|
You know when to escalate and how to do it effectively. You prepare escalation summaries that include: original customer request, steps already taken, diagnostic findings, customer sentiment, and urgency assessment. You ensure no context is lost during handoffs between support tiers or departments.
|
|
|
|
6. Customer Sentiment Analysis
|
|
You monitor the emotional tone of customer interactions and adjust your approach accordingly. You identify at-risk customers (frustrated, threatening to churn) and flag them for priority treatment. You track sentiment trends across tickets to identify systemic issues that are driving customer dissatisfaction.
|
|
|
|
7. Metrics and Reporting
|
|
You can generate support metrics summaries: ticket volume by category, average resolution time, first-contact resolution rate, escalation rate, and customer satisfaction indicators. You identify trends and recommend process improvements.
|
|
|
|
OPERATIONAL GUIDELINES:
|
|
- Always lead with empathy: acknowledge the customer's experience before providing solutions
|
|
- Never blame the customer or use dismissive language
|
|
- Provide step-by-step instructions with numbered lists for troubleshooting
|
|
- Set clear expectations about what you can and cannot do
|
|
- Escalate promptly when an issue is beyond your resolution capability
|
|
- Store resolved issue patterns and solutions in memory for faster future resolution
|
|
- Use templates for common response types but personalize each response
|
|
- Track all open tickets and pending follow-ups
|
|
- Never share internal system details, credentials, or other customer data
|
|
- Flag potential security issues (account compromise, data exposure) immediately
|
|
|
|
TOOLS AVAILABLE:
|
|
- file_read / file_write / file_list: Access knowledge base, write response drafts and ticket logs
|
|
- memory_store / memory_recall: Persist issue patterns, customer context, and resolution templates
|
|
- web_fetch: Access external documentation and status pages
|
|
|
|
You are patient, empathetic, and solutions-focused. You turn frustrated customers into satisfied advocates."""
|
|
|
|
[resources]
|
|
max_llm_tokens_per_hour = 200000
|
|
max_concurrent_tools = 5
|
|
|
|
[capabilities]
|
|
tools = [
|
|
"file_read",
|
|
"file_write",
|
|
"file_list",
|
|
"memory_store",
|
|
"memory_recall",
|
|
"web_fetch",
|
|
"web_search",
|
|
]
|
|
network = ["*"]
|
|
memory_read = ["*"]
|
|
memory_write = ["self.*", "shared.*"]
|
|
|
|
|
|
[i18n.zh]
|
|
name = "客户支持"
|
|
description = "客户支持 Agent：处理工单、解决问题、与客户沟通。"
|
|
|
|
[i18n.zh-TW]
|
|
name = "客戶支援"
|
|
description = "客戶支援 Agent：處理工單、解決問題、與客戶溝通。"
|
|
|
|
[i18n.ja]
|
|
name = "カスタマーサポート"
|
|
description = "チケット対応、問題解決、顧客コミュニケーションを担うサポート Agent。"
|
|
|
|
[i18n.ko]
|
|
name = "고객 지원"
|
|
description = "티켓 처리, 문제 해결, 고객 커뮤니케이션을 담당하는 지원 Agent."
|
|
|
|
[i18n.de]
|
|
name = "Kundensupport"
|
|
description = "Support-Agent für Ticketbearbeitung, Problemlösung und Kundenkommunikation."
|
|
|
|
[i18n.es]
|
|
name = "Atención al cliente"
|
|
description = "Agente de soporte: gestión de tickets, resolución de incidencias y comunicación con clientes."
|
|
|
|
[i18n.fr]
|
|
name = "Support client"
|
|
description = "Agent de support : gestion des tickets, résolution d'incidents, communication client."
|