Factory Hard — Design agent for Claude Code
Hardest code changes.
How to install Factory Hard
Installs to ~/.claude/agents/plicjo-factory-factory-hard.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/plicjo/factory/HEAD/agents/factory-hard.md -o ~/.claude/agents/plicjo-factory-factory-hard.md Restart Claude Code, or start a new session, for it to be picked up.
What Factory Hard does
name: factory-hard description: Hardest code changes. Cross-cutting design, gnarly concurrency, subtle algorithms, and precisely specified sequences that must be executed to the letter. Use when the change needs judgment on vague intent or when a mistake is expensive. model: opus effort: high background: true skills: [factory-mode]
You take the changes the parent judged too hard for the standard code delegate. Read the `factory-mode` skill in full before starting, including its Principl
Alternatives in Design
- Apex Architect — Use this agent when the user needs strategic architecture analysis, design tradeoffs, or read-only debugging 523 ★
- Mermaid Expert — Create Mermaid diagrams for flowcharts, sequences, ERDs, and architectures 7 ★
- Cognitive Engineer — Agent cognition and memory systems expert 5 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Poteto Judge
Highest-judgment delegate for poteto-mode. Cross-cutting design, gnarly concurrency, subtle algorithms, prose
Effort Xhigh
Pins XHIGH reasoning effort; pair with a per-call model param (effort-capable models only). For the hardest ve
Brief Auditor
Advisory critic that grades .design/BRIEF.md against five brief anti-patterns (vague verbs, missing audience,
System Design Expert
Tier 1 — Strategy. Owns large-scale client/server and cross-cutting design. Use for system design, client-serv
Design Reviewer Coherence
The cross-cutting reviewer of the stage-2 design review round — contradictions across the whole design; runs l
Plantuml Expert
Create PlantUML diagrams for flowcharts, sequences, ERDs, and architectures. Masters syntax for all diagram ty
Related Skills
Release Audit
Use right before cutting a release — it spawns ONE fresh sibling session via CCC that audits only the JUDGMENT
Prompting Frameworks
Four Claude/Cowork skills for structuring and QA'ing prompts: ART (refining vague asks), ICIO (structuring pro
Claude Gtm Skills
Ten Claude Agent Skills for cold email, LinkedIn, nurture sequences, and brand strategy. Not prompt templates: