DOCS · REFERENCE · AGENTS

Agents

Synthex ships 15 user-facing agents organised into three tiers, plus the internal helpers listed at the end. Orchestration agents coordinate. Specialist agents do domain-specific work. Research agents close feedback loops. Every agent's identity lives in plain markdown — what you see here is the directory; the prompts themselves are linked from each card.

Tier I

Orchestration

Coordinates execution · 3 agents

01 / TL

Tech Lead

Full-stack orchestrator. Breaks tasks down, delegates to specialists, rolls up the result.

View source ↗
02 / FE

Lead Frontend Engineer

Coordinates frontend delivery, enforces UX quality, accessibility, and design-system fidelity.

View source ↗
03 / PM

Product Manager

Interactive requirements gathering. Transforms PRDs into milestone-based plans.

View source ↗
Tier II

Specialists

Domain expertise · Bluminal blue · 9 agents

04

Architect

System design, ADRs, plan-review feasibility, technology trade-offs.

View source ↗
05

Code Reviewer

Craftsmanship review — correctness, maintainability, reuse, conventions.

View source ↗
06

Security Reviewer

Mandatory gate. Secrets, injection, auth, supply chain, OWASP.

View source ↗
07

Quality Engineer

Test strategy, gap analysis, behavior-driven test authoring.

View source ↗
08

Terraform Reviewer

Plan review with cost, blast-radius and security verdict for AWS.

View source ↗
09

Design System

Token ownership, component governance, compliance audit.

View source ↗
10

Performance Engineer

Quantified, full-stack performance analysis with budgets.

View source ↗
11

SRE

SLOs, observability, runbooks, deployment risk, postmortems.

View source ↗
12

Technical Writer

API docs, guides, migrations, changelogs verifiable against code.

View source ↗
Tier III

Research & Analysis

Continuous improvement · 3 agents

13

UX Researcher

Personas, journey maps, opportunity-solution trees, heuristic eval.

View source ↗
14

Metrics Analyst

DORA + HEART + AARRR + OKRs. Trend, benchmark, learn.

View source ↗
15

Retrospective Facilitator

Planned-vs-actual, blameless retros, tracked improvement loops.

View source ↗

Internal helper agents

The plugin also registers 10 helper agents that you don't call directly. Commands such as review-code/synthex:review-code, write-implementation-plan/synthex:write-implementation-plan, and the standing-pool commands invoke them for multi-model review and pool routing. See Multi-model review and Standing pools.

  • multi-model-review-orchestrator — Fans a review out to native and external CLI proposers in parallel and consolidates findings.
  • findings-consolidator — Deduplicates and merges findings from multiple reviewers into one attributed, sorted list.
  • codex-review-prompter — Adapter that invokes the OpenAI Codex CLI as an external proposer in multi-model review.
  • gemini-review-prompter — Adapter that invokes the Gemini CLI as an external code-review proposer in multi-model review.
  • ollama-review-prompter — Adapter that POSTs to a local Ollama server as an external code-review proposer.
  • claude-review-prompter — Specialty adapter invoking a second Claude CLI session as an external multi-model review proposer.
  • bedrock-review-prompter — Adapter that invokes AWS Bedrock as an external code-review proposer in multi-model review.
  • llm-review-prompter — Adapter that invokes Simon Willison's llm CLI as an external code-review proposer.
  • standing-pool-submitter — Submits review tasks to a standing pool's task list and polls for the consolidated report envelope.
  • standing-pool-cleanup — Atomically removes a stale standing pool's index entry and metadata directory under the cross-session lock.