Skip to content

forcedotcom/sf-skills / plugins/builder/agentforce-adlc/skills/agentforce-persona-generate

agentforce-persona-generate

Verified

Use to design an AI agent persona — identity, voice, tone, behavioral style, guardrails — and encode it into Agent Script (.agent files) or Agentforce Builder field values. Runs four sequenced phases (DRAFT, REFINE, SCORE, ENCODE) plus five spokes (NAME, FILL OUT, ADJUST, CHECKLIST, WORKSHOP). Supports greenfield (brand input to a starter .agent file) and brownfield (audit an existing .agent file and re-encode persona fields only), with a 100-point evaluation rubric. Trigger when a user designs a new agent persona, retrofits persona consistency, translates brand guidelines into agent voice, defines tone/register/formality/warmth/humor/personification, encodes a persona for Agent Script or Agentforce Builder, or scores or audits an existing .agent file for persona quality. Do not trigger for building Agentforce actions/metadata, authoring Agent Script flow control without persona work, or testing agent conversations.

references/scoring-rubric.md

View on GitHub
---
version: "1.0.0"
date: 2026-07-15
---

# Persona Scoring Rubric

Reference for PHASE 3 — SCORE. The 100-point rubric, the brownfield label-mapping table, and
the per-subagent / per-action sub-scoring rules. SKILL.md keeps the SCORE phase procedure
(entry points, extraction walk, recommendations, exit); the detail lives here.

## Brownfield label mappings

When scoring existing agent instructions, map by substance, not label. Mark inferred mappings:
"inferred — [source label]." Absence of a framework concept ≠ zero score; score what's present.

| If you find... | It likely maps to... |
|---|---|
| "Personality," "communication principles," "voice guidelines" | Identity traits |
| "Communication rules," "response guidelines," "style guide" | Dimension selections |
| "Banned phrases," "forbidden phrases," "don't say" | Never-Say List |
| "Vocabulary," "terminology," "glossary" | Lexicon |
| "Tone rules," "emotional guidelines" | Tone Flex / Tone Boundaries |
| "Example responses," "sample language," "approved phrases" | Phrase Book |

## 100-Point Rubric

| Category | Points | Criteria |
|---|---|---|
| **Identity Coherence** | /11 | Traits distinct + behaviorally defined • Personification clear and consistent • Negative Identity (2-4) • Design Inputs coherent (audience → register, channel → chatting style, company → frame of reference) |
| **Dimension Consistency** | /11 | Each of 14 dimensions coherent with Identity, constraints respected • Tone Boundaries consistent with Emotional Coloring/Empathy; Tone Flex within range • Chatting Style adapted for channel |
| **Internal Consistency / Conflict-Pair Check** | /10 | No contradictions between Tone Boundaries and Phrase Book entries • Never-Say doesn't overlap Lexicon • Emotional Coloring + Humor compatible • Identity traits aligned with dimension selections • Immutable content not contradicted by lexicon |
| **Behavioral Specificity** | /9 | Concrete behavioral examples, testable rules • Never-Say ≥5 covering chatbot filler + register violations + cognitive markers + persona-specific • Lexicon populated, immutable distinguished where applicable • Brand input extraction depth |
| **Phrase Book Behavioral Coverage** | /11 | Phrases for the situations the agent will actually encounter (acknowledgement, affirmation, redirect, escalation if external, etc.) • 2-4 per applicable category • Match register and dimensions • For Talking Systems: depersonalized phrasing expected |
| **Tone Flex & Boundaries** | /9 | Triggers + shift directions + magnitudes defined • Content-sensitivity default present (or overridden) • Tone Boundaries testable • Flex range stays within boundaries |
| **Lexicon & Vocabulary** | /7 | Global lexicon populated • Per-subagent lexicon where applicable (brownfield) • Reading Level appropriate for audience • Immutable terms preserved • Lexicon format appropriate |
| **Static Messages** | /7 | Welcome reflects Identity + Register + Voice + Tone + Brevity • Error reflects Formality + Warmth + Coloring + Brevity • All static strings sound like the same agent • Telephony: AI disclosure in welcome, brevity recalibrated • SMS: first-message welcome with system identification |
| **Business Context Alignment** | /7 | Goals and metrics inform design • Persona supports objectives • Audience demographics reflected |
| **Sample Quality / Scenario Coverage** | /9 | Persona recognizable without seeing dimension table • Happy path + uncertainty + boundary + escalation scenarios • Channel-appropriate • Brand vocabulary appears naturally |
| **Completeness** | /9 | All framework sections addressed • Values only if user-provided • No gaps where the framework expects a deliberate decision |

(Total: 100. Conflict-Pair Check is new in this version. Phrase Book and Sample categories tightened to grade behavioral coverage / scenario coverage rather than category counts / aesthetics.)

**Score presentation:** Total presented as percentage — 90-100 production-ready, 75-89 strong
foundation, 50-74 needs revision, <50 significant gaps.

## Per-Subagent and Per-Action Sub-Scoring (brownfield)

For each subagent in the source `.agent` file:
- Persona reminder present in `reasoning.instructions`?
- Tone flex calibrated appropriately for the subagent's purpose (e.g., escalation subagent should drop humor)?
- Lexicon scoped correctly (subagent-specific terms in `reasoning.instructions`, not in global)?

For each action:
- `progress_indicator_message` in voice if channel is voice-bearing?
- Tone-appropriate ("Pulling the numbers..." for casual; "One moment..." for formal)?

Roll up per-layer scores into the overall total with a per-layer breakdown in the output.
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: 472kUpdated Claude CodeCodexVerifiedInstall
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: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall