dev-research

Analyzes drafted job sets against target codebases and external services to rate implementation feasibility.

Updated Jul 10, 2026
One-click install
npx skills add https://github.com/subvisual/the-a-team --skill dev-research-subvisual
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: dev-research
Source: https://github.com/subvisual/the-a-team/tree/main/.claude/skills/dev-research
Command: npx skills add https://github.com/subvisual/the-a-team --skill dev-research-subvisual

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Job sets drafted without implementation contact produce product scopes nobody can build, and the gap surfaces only after development starts. This Skill performs a planning-time reality check on a drafted job set before scope is committed. ## Core Features & Use Cases - Codebase and service sweep: Inspects the target repo for existing integrations, stack bindings, auth models, and seams, then checks every external service the jobs imply for access, auth shape, cost, and rate ceilings. - Decision-filtered findings: Keeps only findings that change a scope call, job feasibility, sequencing, or buy-vs-build, each rated cheap/moderate/expensive/unknown with a confidence level and a resolving evidence pointer. - Ledger-routed output: Returns findings through the Know/Don't Know ledger under a three-way answer rule — verifiable from the repo, covered by a Declared default, or routed as a question — with substance landing in research-plan.md and jobs carrying citations only. - Use Case: A discovery phase drafts five jobs implying a payments provider and per-user OAuth. The Skill verifies the provider's API access terms, flags the OAuth consent flow as a first-run product fact, rates one job expensive with moderate confidence, and routes a blocking partnership question to research. ## Quick Start Run the dev-research skill against this target repository and the drafted job set to produce confidence-rated feasibility findings routed through the ledger.

Frequently Asked Questions about dev-research

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I validate a job set for implementation feasibility before committing scope?▼

Dispatch this Skill over the drafted jobs and target repo. It derives the external services each job implies, sweeps the codebase and those services, and returns only findings that change a scope call, feasibility, sequencing, or buy-vs-build decision.

What does a dev research review check in external services?▼

Each implied service gets an access check (public API, partnership, waitlist), auth shape (API key, OAuth, mTLS), cost shape at order-of-magnitude level, rate ceilings where they bound a job, and whether a cheap API-shaped fake exists for v0.

Can this Skill make architecture or stack decisions?▼

No. Architecture decisions and stack picks belong to the architecture skill, which owns docs/product/adr. This Skill supplies the repo-binding evidence and service constraints those decisions rest on, but never writes ADRs or makes the decision itself.

When should dev research not be used?▼

Do not use it for architecture decisions, schema or API design, time estimates, deep code review, or any implementation work. It is a planning-time feasibility beat; depth belongs to the spec and dev phases.

What happens when a service's access model cannot be verified?▼

The finding is rated unknown and routed to a technical research activity in research-plan.md. The Skill never guesses an API contract or auth shape, and any credential-gated check that degraded is named explicitly in the return.