Architect Prove — Development skill for Claude Code
Run the prove stage.
How to install Architect Prove
Installs to ~/.claude/skills/frankxai-ai-architect-architect-prove/SKILL.md
mkdir -p ~/.claude/skills/frankxai-ai-architect-architect-prove && curl -fsSL https://raw.githubusercontent.com/frankxai/ai-architect/HEAD/commands/architect-prove.md -o ~/.claude/skills/frankxai-ai-architect-architect-prove/SKILL.md Restart Claude Code, or start a new session, for it to be picked up.
What Architect Prove does
description: Run the prove stage. Writes at least ten eval cases including a refusal and an injection case, runs the harness in this session, and evaluates gate.evals. argument-hint: "[--runner ]"
/architect-prove — stage 7 of 9
Dispatches `eval-engineer` for the `prove` stage. Argument: `$ARGUMENTS` — `--runner` names the command that drives the system under evaluation, when the repository's own runner is not obvious.
The artifact root is `docs/archi
Alternatives in Development
- Agentic Engineering Handbook — The definitive OpenAI, Claude, MCP, Harness, Evals, and Production Agent Systems learning roadmap 553 ★
- HERO Anti OverDefense — HERO = Hashing · Edge cases · Rubrics · Overbuild — the four shapes coding agents over-defend in 333 ★
- Eval Author — Draft a new eval case (eval/cases/ .json) + scaffold its golden folder, golden_pending until captured on the V 136 ★
Full documentation available on GitHub
View Source RepositoryRelated Skills
Eval Cases
Grow an eval case set from the placeholder rows to a real 15-25 - harvest genuine cases out of the repository
Greenlight
Refusal handling skill for vulnerability research evals
Agentic Eval
Niche-agnostic agentic evaluator using CLEAR v2.0 framework — 6-domain assessment, 8 analysis dimensions, 6-ti
Architect Decide
Run the decide stage. Settles the four decisions that are expensive to reverse, names an owner for each of the
Architect Discover
Run the discover stage. Works the question bank until every section has an answered question and nothing is ta
TS Evals
TS teaching-reflector eval-analysis mode — student evaluation analysis
Related Agents
Wh Designer
Design-phase agent. In one session turns " -> " into the risk tier, spec.md with ADRs, evals.md within the tie
Eval Failure Analyzer
Analyze Logic-Lens benchmark/eval failures. Use after running content-evals, or when pointed at a skills-works
Agent Eval Engineer
Use to design and run regression evaluations for an agent, skill, tool, or prompt — happy-path, edge-case, adv