Persona Synthesizer — Development agent for Claude Code
Phase 2 of the grounded-persona-eval method — builds 3-4 personas strictly from Phase 1's sourced evidence.
How to install Persona Synthesizer
Installs to ~/.claude/agents/qte77-agentic-grounded-persona-eval-persona-synthesizer.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/qte77/agentic-grounded-persona-eval/HEAD/.claude/agents/persona-synthesizer.md -o ~/.claude/agents/qte77-agentic-grounded-persona-eval-persona-synthesizer.md Restart Claude Code, or start a new session, for it to be picked up.
What Persona Synthesizer does
name: persona-synthesizer description: Phase 2 of the grounded-persona-eval method — builds 3-4 personas strictly from Phase 1's sourced evidence. Refuses to run if Phase 1 has no real output yet.
persona-synthesizer
You run **Phase 2 (Grounded personas)** of the grounded-persona-eval method. Your only job: synthesize a small number of personas **from** Phase 1's evidence — never from assumption.
Precondition — check before starting
Read `findings/-findings.md` §Re
Alternatives in Development
- Gitnexus Synthesis Critic — GitNexus final review synthesis critic 45.8k ★
- Omd Critic — Reads the full run output, user feedback, and persona ABANDONs, then writes a root-cause critique of omd-maste 465 ★
- Probability Honesty Checker — Forces honest first-principles probability distribution + EV calculation 139 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Hcs Eval Reviewer
Reviews regression-trap quality and eval harness coverage for HCS. Ensures traps capture real failure classes
Taa Writer
Use this agent for the TAA docs track (/taa:docs). Produces grounded product documentation - user manuals, fea
Outreach Strategist
Drafts the first approach to an owner and the two follow-ups — specific, sourced, short, and honest about who
Review Synthesizer
Combines findings from 5 review personas into a unified synthesis. Deduplicates, classifies signal type, adds
Arquiteto
Meta-persona of the workbench. Invoke to plan complex tasks (>3 steps, multi-domain, irreversible action), cre
Eval Implementer
Implementer role of the D365FO agent eval loop. Runs ON THE VM (mcp-server in full mode + C# bridge) against t
Related Skills
Academic
Write academic prose - thesis chapters, papers, abstracts, literature reviews, methodology and results section
Ikigai
A Claude Code skill for authentic Japanese ikigai discovery — grounded in Kamiya, Mogi, and the Ikigai-9 scale
CR Agents
Brute-force AI method for technical work. Personas anchored to experts in their fields run as Claude Code sub-