benthepsychologist

Run structured market research with AI agents. Use when the user needs market mapping, competitor analysis, user voice, positioning, or go-to-market-oriented research.

Scope a research effort before execution. Use when the user has a topic, thesis, or problem but not yet a clear initiation document, research plan, or stopping condition.

Stress-test an idea, argument, strategy, or product concept. Use when the user needs strong counterarguments, assumption audits, failure modes, alternatives, and an adversarial verdict.

Investigate a repository as a source base. Use when the user needs codebase mapping, architecture findings, dependency/risk analysis, or implementation-oriented technical research.

Synthesize prior research passes into conclusions, risks, and decision-ready outputs. Use when enough evidence has been gathered and the next job is integration rather than search.

Run structured literature and evidence research. Use when the user needs a field map, core evidence review, competing models analysis, or synthesis of scientific/technical literature.

Apply a safe branching model with branch naming, commit discipline, and merge/rebase flow. Use when starting implementation, preparing commits, syncing with main, and opening pull requests.

Create re-runnable infrastructure shell scripts for gcloud and bq that safely handle existing resources and produce deterministic outcomes across repeated executions.

Start a real local HTTP server process, send curl requests to live endpoints, and assert status codes and response content for quick smoke validation.

Bootstrap registrar-managed environments using registrar CLI, verify provisioned cloud resources, and capture reproducible evidence for dev, stage, and prod setup.

Author specwright-compatible specs with complete frontmatter, bounded scope, deterministic test plans, and acceptance criteria that map directly to deliverables.