PranavNagrecha/AwesomeSalesforceSkills / skills/admin/agent-conversation-design
agent-conversation-design
VerifiedUse when writing or auditing the conversational copy layer of a Salesforce bot or Agentforce agent: utterance authoring strategy, fallback message copy, escalation-criteria phrasing, and persona-consistent dialog scripting across channels. NOT for designing the bot dialog structure, intent model, or handoff architecture — use architect/einstein-bot-architecture. NOT for encoding agent tone in Agentforce system instructions — use agentforce/agentforce-persona-design.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 16.0 KB |
| references/examples.md | markdown | 7.1 KB |
| references/gotchas.md | markdown | 6.4 KB |
| references/llm-anti-patterns.md | markdown | 8.7 KB |
| references/well-architected.md | markdown | 5.2 KB |
| scripts/check_agent_conversation_design.pyexecutable | script | 10.2 KB |
| templates/agent-conversation-design-template.md | markdown | 6.1 KB |
templates/agent-conversation-design-template.md
View on GitHub# Agent Conversation Design — Work Template
Use this template when authoring or auditing the conversation copy layer for an Einstein Bot or Agentforce deployment.
## Scope
**Skill:** `agent-conversation-design`
**Request summary:** (fill in what the practitioner asked for — e.g., "expand utterances for Returns intent", "rewrite fallback copy", "write escalation criteria for VIP handoff")
**Platform:** [ ] Einstein Bots (dialog/intent model) [ ] Agentforce (topic/action model) [ ] Both
---
## Context Gathered
**Intent / Topic inventory:**
| Intent or Topic Name | Current Utterance Count | Fallback Rate (if known) | Escalation Destination |
|---|---|---|---|
| | | | |
| | | | |
**Brand voice adjectives:** (e.g., empathetic, concise, professional)
- Adjective 1:
- Adjective 2:
- Adjective 3:
**Target channel(s):** (web chat / mobile / Slack / API / other)
- Channel 1: _________________ | Max message length: _____ chars | Markdown: Y/N | Emoji: Y/N
- Channel 2: _________________ | Max message length: _____ chars | Markdown: Y/N | Emoji: Y/N
**Escalation destinations (user-facing names):**
- Destination 1 (internal queue name): _________________ → User-facing name: _________________
- Destination 2 (internal queue name): _________________ → User-facing name: _________________
---
## Utterance Set — [Intent/Topic Name]
**Target count:** _____ utterances (minimum 20 for production; 50+ recommended for high volume)
**Case data source:** (case export date range, filter criteria used)
| # | Utterance Text | Register | Variant Type |
|---|---|---|---|
| 1 | | Formal | Original |
| 2 | | Casual | Synonym |
| 3 | | Frustrated | Register variant |
| 4 | | Abbreviated | Abbreviation |
| 5 | | Frustrated | Typo variant |
| ... | | | |
**Coverage check:**
- [ ] Formal register: _____ utterances
- [ ] Casual register: _____ utterances
- [ ] Frustrated register: _____ utterances
- [ ] Typo / error variants: _____ utterances
- [ ] Vocabulary clusters covered: _____ distinct synonym groups
---
## Fallback Copy
### Stage 1 — First Unmatched Message
```
[Bot message — stage 1 fallback]
I want to make sure I help with the right thing. Are you asking about one of these?
• [Top topic 1 in user-facing language]
• [Top topic 2 in user-facing language]
• [Top topic 3 in user-facing language]
• Something else
```
### Stage 2 — Second Consecutive Unmatched Message
```
[Bot message — stage 2 fallback]
I'm still not finding a match for what you're asking.
Would you like me to connect you with [User-facing team name]
who can [specific value: e.g., "pull up your account and help directly"]?
```
### Stage 3 — Auto-Handoff (user declines or no response after ___ seconds)
```
[Bot message — stage 3 fallback / transfer message]
Got it — I'll connect you now. [User-facing team name] will be with you shortly.
I'm passing along your message so they're ready to help.
```
---
## Escalation Intent Utterances
**Escalation intent name:** (e.g., "Escalate_To_Human", "Request_Agent")
| # | Utterance | Register |
|---|---|---|
| 1 | agent | Keyword |
| 2 | human | Keyword |
| 3 | talk to a person | Casual |
| 4 | I want to speak to a representative | Formal |
| 5 | connect me to a real person | Casual |
| 6 | get me a human now | Frustrated |
| 7 | just get me an agent | Frustrated |
| 8 | I need to talk to someone | Casual |
| 9 | agent!!!! | High-emotion |
| 10 | human pls | Abbreviated |
| ... | (add org-specific variants) | |
**Condition-based escalation triggers** (for platform configuration, not copy):
| Trigger Condition | Escalation Destination | Priority |
|---|---|---|
| (e.g., Account.VIP__c = true) | (user-facing team name) | High |
| (e.g., Case.Priority = Critical) | (user-facing team name) | High |
| (e.g., Bot fallback fired 3 times) | (user-facing team name) | Standard |
---
## Dialog Script — [Intent/Topic Name]
**Channel:** _________________
**Persona adjectives applied:** _________________
```
User: [Opening utterance — formal/casual variant]
Bot: [Turn 1 — acknowledge + clarify if needed]
[Keep under ___ characters for [channel]]
User: [User response — normal path]
Bot: [Turn 2 — resolution or next step]
[Named CTA or confirmation: "Your [X] has been [Y]."]
User: [Optional follow-up or confirmation]
Bot: [Turn 3 / close — persona-consistent sign-off]
```
**Persona review:**
- [ ] Turn 1 reflects voice adjective: _________________
- [ ] Turn 2 is within max length for [channel]
- [ ] Turn 3 closes with brand-consistent phrasing
- [ ] No internal jargon or system identifiers visible to user
- [ ] Escalation path scripted if this intent can overflow to human handoff
---
## Agentforce Topic Description (if applicable)
**Topic name:** _________________
**Description draft:**
```
Use for: [explicit subject matter — 2–4 specific scenarios]
Do NOT use for: [explicit exclusion — route to [Other Topic Name] instead]
```
**Routing boundary test:** (submit the following borderline queries 5 times each in conversation preview and confirm consistent routing)
| Borderline Query | Expected Topic | Actual Topic (run 1–5) | Pass/Fail |
|---|---|---|---|
| | | / / / / / | |
| | | / / / / / | |
---
## Review Checklist
- [ ] Every intent has 20+ utterances (50+ for high-volume production intents)
- [ ] Utterances cover formal, casual, frustrated, and typo/abbreviated registers
- [ ] Fallback copy follows 3-stage progressive clarification pattern
- [ ] Escalation copy uses user-facing team names, not internal queue identifiers
- [ ] Escalation intent includes high-emotion and frustrated-register utterances
- [ ] Agentforce topic descriptions include explicit scope and exclusion clauses
- [ ] Dialog scripts authored per channel where channel register differs
- [ ] All bot copy reviewed against brand voice adjectives
- [ ] No internal system identifiers (queue names, IDs) visible in any customer-facing message
- [ ] Fallback intent utterance list is EMPTY (Einstein Bots only)
---
## Notes
(Record any deviations from the standard patterns, org-specific constraints, or decisions made during this session.)
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |