AskVault skills, an overview

Written by Aashiq, Founder, AskVault · Reviewed by Aashiq

Last updated: May 15, 2026 · 6 min read

TL;DR

AskVault ships 14 built-in skills the bot can invoke during a conversation. Each has trigger phrases, policy bounds, and hard caps the LLM cannot override. Customize what skills are enabled per workspace. Available across all 13 channels with rate limits up to 50,000 queries per month.

What a skill is

A skill is a structured capability the AI agent can invoke during a conversation. Different from a simple knowledge-retrieval answer in three ways:

1. Skills can take actions. Create a Zendesk ticket, book a Calendly slot, look up an order, capture a lead. Knowledge retrieval just produces text.
2. Skills have policy bounds. Each skill has hard limits the LLM cannot override (per-visitor caps, per-workspace caps, rate limits, audience restrictions).
3. Skills are configured per workspace. You decide which skills are on and what bounds apply. Trigger phrases, response messages, integration mappings.

A conversation might invoke zero, one, or several skills. Most conversations use zero (just knowledge retrieval). Sales-intent conversations may use 3 to 5 (collect_lead, sdr_lead_qualifier, demo_scheduler, then maybe escalate_to_human).

The 14 built-in skills

Categorized by purpose:

Skill	Purpose	Plan
knowledge_search	Retrieve and cite content from indexed knowledge	Free
collect_lead	Capture name, email, phone with conversation context	Free
escalate_to_human	Hand off to live chat with full context	Growth
ticketing_router	Create Zendesk tickets routed by intent	Starter
sentiment_router	Detect frustration and route or de-escalate	Starter
sdr_lead_qualifier	BANT-style qualification before CRM sync	Growth
demo_scheduler	Book Calendly slots from chat	Growth
content_recommender	Suggest related articles after answers	Growth
wismo_order_status	Live order lookup via Shopify	Growth
wc_order_status	Live order lookup via WooCommerce	Growth
cart_recovery	Detect cart hesitation and offer help	Growth
discount_negotiator	Apply pre-approved codes within bounds	Business
subscription_manager	Live Stripe subscription lookups	Business
custom_webhook	Fire your own webhook with conversation context	Business

How the bot decides to invoke a skill

For each customer message, the bot evaluates in this order:

1. Check explicit trigger phrases. Every skill has a list of phrases like "where is my order?", "I want a demo", "talk to a human". If the customer's message matches a trigger, the skill fires.
2. Check intent classifiers. For each enabled skill with auto-trigger enabled, an intent classifier evaluates the conversation context. High-confidence matches trigger the skill.
3. Check policy preconditions. Even when a trigger fires, the skill verifies its policy preconditions are met. For example, collect_lead only fires once per conversation; discount_negotiator checks caps before offering.
4. Default to knowledge_search. If no skill triggers, the bot answers from the knowledge base.

Most messages take path 4 (just knowledge retrieval). Skills fire on specific intent-bearing messages, not every turn.

Policy bounds the LLM can't override

Each skill has bounds enforced at the AskVault layer, not in the LLM prompt. This matters because LLM-only guardrails are bypassable; AskVault's policy layer isn't.

Examples:

• discount_negotiator has per-visitor (15%), per-workspace ($5,000/mo), and per-code caps. Even if the LLM "decides" to offer 50% off, the policy layer rejects it.
• collect_lead runs once per conversation. Even if the LLM keeps trying, it won't ask again.
• custom_webhook has rate limits per workspace. Burst traffic gets a polite fallback rather than overwhelming your endpoint.

Configure bounds under AI Agents > Skills > [skill name] > Bounds.

Customizing skill behavior

Three layers of customization per skill:

• Triggers. What phrases fire the skill. Default lists work for most teams; customize for industry vocabulary.
• Messages. What the bot says when the skill fires. Customer-facing wording matches your brand voice.
• Integration mappings. For skills that call external services (Shopify, Stripe, HubSpot), map field names and routing rules.

All three configurable in the dashboard per workspace. Changes apply within seconds, no redeploy.

Enabling and disabling skills

Each skill has an on/off toggle. Disable skills you don't need:

• Free-tier customers don't need subscription_manager if they're not on Business plan.
• Internal helpdesk bots don't need cart_recovery or discount_negotiator.
• B2B SaaS without e-commerce doesn't need wismo_order_status.

Reducing skill surface area improves accuracy (fewer false-positive trigger matches) and audit clarity.

Skill execution latency

Most skills add minimal latency:

• Knowledge-only. 400 ms to 1 second per turn.
• Skill with API call. Add 300 ms to 2 seconds for the external API roundtrip.
• Multi-skill chain. A conversation that triggers collect_lead -> sdr_lead_qualifier -> demo_scheduler totals 60 to 120 seconds wall-clock, mostly waiting on the customer's responses.

Skills don't significantly affect bot response time in normal usage. Skills with slow downstream APIs (your WooCommerce host on a shared server) are the exception.

Audit logging per skill

Every skill invocation is logged:

• Skill name and trigger that fired.
• Input passed to the skill.
• Result returned.
• Policy bound check (pass/fail and reason).
• Timestamp.

Visible under Dashboard > AI Agents > Skill Audit Log. Retained 365 days on standard plans, 6 years on Enterprise.

Multi-skill conversations

Conversations can invoke multiple skills in sequence. Typical sales flow:

1. Visitor asks pricing. Bot answers via knowledge_search.
2. Visitor shows buying intent. collect_lead captures contact info.
3. sdr_lead_qualifier asks BANT questions.
4. Lead syncs to HubSpot.
5. Visitor asks for a demo. demo_scheduler offers Calendly slots.
6. Booking confirmed. Confirmation emails fire.

Each skill maintains state in the conversation context. The bot's reasoning carries across skill invocations.

Limits

• Plan availability. Each skill has a minimum plan; see the table above.
• Concurrent skill invocations. Up to 3 skills can run in parallel within one bot turn. Multi-skill chains run sequentially.
• Custom skills. Today only via custom_webhook. Native custom-skill SDK on the roadmap.

Common pitfalls

Skills fire on the wrong messages. Trigger phrases are too broad. Tighten the lists or move to intent-only auto-trigger.

Skills don't fire when they should. Trigger phrases too narrow. Add synonyms or enable auto-trigger.

Skills interrupt the conversation. Skill fires mid-explanation. Configure trigger conditions to require explicit signals or end-of-explanation markers.

Customers confused by skill output. Customer-facing messages don't match your tone. Customize messages under each skill's settings.

Related guides

• What is Retrieval-Augmented Generation (RAG)?
• Custom webhook skill
• Escalate to human skill
• Collect lead skill
• Webhooks reference

See RAG-grounded answers in action

Every AskVault answer is grounded in the source documents you upload. Never hallucinated.

Try AskVault Free → Read the API docs

Was this page helpful? Yes No