Angela Personas

Angela evaluates your documentation through 7 distinct expert lenses — each with its own priorities, blind spots, and signature questions. Together they cover the axes along which technical docs fail: unclear prose, missing user mental models, incomplete contracts, untested assumptions, narrative drift, business misalignment, and weak structure.

Personas are available in three activation modes: angela consult for a single-persona deep dive, angela polish --persona to steer a rewrite, and angela review --persona to inject lenses into corpus-wide coherence analysis.

The Seven

Icon	ID	Name	Focus
✏️	`tech-writer`	Salou	Technical writing precision and clarity
🎨	`ux-designer`	Gougou	User empathy, mental models, and accessibility
🔌	`api-designer`	Ouattara	API contracts, synthesizer-ready docs, HTTP semantics
🔍	`qa-reviewer`	Kouamé	Quality assurance and validation criteria
🏗️	`architect`	Doumbia	System design, trade-offs, and scalability
📊	`business-analyst`	Béda	Requirements traceability and business value
📖	`storyteller`	Affoué	Narrative clarity and authentic storytelling

The display names (Salou, Gougou, Ouattara, Kouamé, Doumbia, Béda, Affoué) are common Ivorian given names. Lore is built in Côte d'Ivoire and the project embraces its cultural roots rather than defaulting to generic tech-industry placeholders. The emoji keeps persona identity scannable in terminal output where names may truncate.

When to Use Which

Writing a feature doc? Start with tech-writer (Salou) for prose quality, then ux-designer (Gougou) for the reader's mental model.
Documenting an API endpoint? api-designer (Ouattara) catches missing methods, inconsistent naming, and body/header gaps that break Postman imports.
Shipping a decision? Run architect (Doumbia) for trade-off clarity, and qa-reviewer (Kouamé) to force the "what could go wrong" column.
A long-form guide or onboarding piece? storyteller (Affoué) checks the narrative doesn't drift mid-paragraph.
A product or feature spec? business-analyst (Béda) verifies requirements trace back to a named business outcome.

For single-document work, pick 1 persona. For corpus-wide review, 3–4 complementary personas give cross-lens signal (when two personas independently flag the same issue, that convergence is a high-confidence marker — Angela surfaces it via the Flagged by: attribution).

Activation modes

`angela consult <persona> <file>` — single-lens offline check

lore angela consult api-designer docs/features/login.md
lore angela consult tech-writer docs/guides/quickstart.md

No AI call, no write. Runs the persona's draft-check lens and prints suggestions. Useful after a polish or manual edit when you want one expert's take without the full draft pipeline.

Run lore angela consult (no arguments) to list all personas.

`angela polish --persona <id>` — steer the AI rewrite

lore angela polish docs/features/login.md --persona ux-designer

Biases the AI polish toward the persona's priorities — e.g. ux-designer emphasizes user flows and error-path clarity; api-designer emphasizes request/response shapes. See angela-polish.md for the full polish flow.

`angela review --persona <id>` — multi-persona corpus coherence

lore angela review --persona tech-writer --persona ux-designer --persona api-designer --persona qa-reviewer

Each persona lens is injected into the review prompt. The AI is instructed to attribute each finding to the persona(s) whose expertise flagged it. When multiple personas concur, they're listed together under Flagged by: — a strong signal that the issue matters across lenses. See angela-review.md.

Alternatively, configure personas in .lorerc and activate them via --use-configured-personas (skips the interactive confirmation prompt).

Example session

# 1. See available personas
$ lore angela consult

Available personas:
  🔌 api-designer         Ouattara
                          API contracts, synthesizer-ready docs, HTTP semantics
  # ... (7 personas)

# 2. Spot-check an API doc
$ lore angela consult api-designer docs/features/invoices.md
  warning  persona  [🔌 Ouattara] Endpoints listed without an HTTP request example — add a ```http block with method, URL, headers, and body

# 3. Review the whole corpus through multiple lenses
$ lore angela review --persona tech-writer --persona qa-reviewer

  + gap   Missing Angela persona documentation
          [abc123] commands/angela-consult.md vs commands/angela-consult.fr.md
          Flagged by: Kouamé, Gougou

Persona selection config

You can pin a default set of personas in .lorerc:

angela:
  review:
    personas:
      selection: "manual"
      manual_list:
        - tech-writer
        - api-designer
        - qa-reviewer

Then run lore angela review --use-configured-personas to skip the interactive prompt. See angela-review.md and config.md for the full cascade.

I4 — Zero hallucination

Persona injection does not relax the evidence rule. Every persona-attributed finding must carry a verifiable quote from the corpus. Findings without evidence are rejected by the post-processing validator, regardless of which persona flagged them.