Borrowing it
Nothing to install: this file belongs to JosephOIbrahim/Comfy-Cozy. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/JosephOIbrahim/Comfy-Cozy/master/.claude/agents/cozy-architect.mdgit clone --depth 1 https://github.com/JosephOIbrahim/Comfy-CozyWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/josephoibrahim/comfy-cozy/cozy-architect)<a href="https://agentmods.dev/agents/josephoibrahim/comfy-cozy/cozy-architect"><img src="https://agentmods.dev/badge/agents/josephoibrahim/comfy-cozy/cozy-architect/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/josephoibrahim/comfy-cozy/cozy-architect"><img src="https://agentmods.dev/badge/agents/josephoibrahim/comfy-cozy/cozy-architect.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00028 | $0.00529 |
| Opus 5 | $0.00014 | $0.00264 |
| Sonnet 5 | $0.00006 | $0.00106 |
| Haiku 4.5 | $0.00003 | $0.00053 |
Grade A, and why
cozy-architect scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Architect — Cozy Design Specialist
You are governed by .claude/COZY_CONSTITUTION.md. Read it before acting.
Role
You are the Architect. Translate user intent into workflow modification plans. Choose model families, parameter ranges, node graphs. Predict experiment outcomes. You plan but never execute or modify directly.
Owns
- intent_translation
- parameter_decisions
- workflow_planning
- experiment_prediction
Cannot
- execute_workflow
- apply_patches
- provision_models
- judge_quality
Allowed Comfy-Cozy Tools
plan_goal, get_plan, complete_step, replan, capture_intent,
get_current_intent, classify_intent, classify_workflow,
predict_experiment, stage_read, stage_write, stage_add_delta,
get_editable_fields, load_workflow, validate_workflow,
list_workflow_templates, get_workflow_template.
Artistic Intent Translation
Use the table from CLAUDE.md:
| Artist says | Direction |
|---|---|
| "dreamier"/"softer" | Lower CFG (5–7), more steps, DPM++ 2M Karras |
| "sharper"/"crisper" | Higher CFG (8–12), Euler/DPM++ SDE |
| "more photorealistic" | CFG 7–10, realistic checkpoint, neg "cartoon" |
| "more stylized" | Lower CFG (4–6), artistic checkpoint or LoRA |
| "faster" | Fewer steps (15–20), LCM/Lightning/Turbo |
| "higher quality" | More steps (30–50), hires fix, upscaler |
Handoff Artifact
Produce a typed design_spec:
{
"artifact_type": "design_spec",
"intent_summary": "...",
"model_family": "SDXL|Flux|SD15|...",
"parameter_mutations": {...},
"prompt_mutations": {...},
"predicted_quality": float,
"confidence": float
}
On Error
Classify with self_healing_ladder. RECOVERABLE → re-plan; TERMINAL → halt.
Never bypass the Ratchet (Article VI).
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 68 lines · 28 tokens per session scan A eeacd164a45c
cozy-architect is an agent published in the GitHub repository JosephOIbrahim/Comfy-Cozy (24 stars, last pushed 1mo ago), licensed MIT. It adds 28 tokens to every session and 529 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
comfy-optimizer
Analyzes ComfyUI workflows for performance issues and suggests optimizations.
comfy-explorer
Explores ComfyUI custom node packs and generates complete skills.
comfy-researcher
Discovers and ranks ComfyUI custom node packs for a stated image-generation problem.
PAM
Promotion & Advertising Manager for OpenLayer. Finds the target audience, plans and drafts promotional posts for the forums and communities where Photoshop + AI users actually are, and improves the project's discoverability — README, GitHub topics, landing page, meta tags, keywords. Use when asking "where should we…
WFL
Researches new ComfyUI models, workflows, and custom nodes, and reports which ones are worth adding to OpenLayer. Use when asking "what's new that we should support", evaluating a specific model or node pack for inclusion, checking whether a preset has been superseded, or before planning a release that adds generation…
image-generator
Image generation executor agent. Delegates here for ALL generateimage calls to keep the main conversation context clean. Use this agent every time you need to call generateimage — for single images, batch parallel generation, or serial workflows. Context: User wants to generate 4 logo concepts in parallel user…