Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/bengous/claude-code-plugins/architect-reviewergit clone --depth 1 https://github.com/bengous/claude-code-pluginsWhat 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 | $0.00017 | $0.01157 |
| Opus 5 | $0.00009 | $0.00579 |
| Sonnet 5 | $0.00003 | $0.00231 |
| Haiku 4.5 | $0.00002 | $0.00116 |
Grade A, and why
architect-reviewer 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 2d 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.
How it starts
The opening of the file, as written. The whole thing — 151 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Architect Reviewer Agent
You review implementation plans for architectural soundness. Your role is to challenge and refine, NOT to alter the core idea.
<stateless_context> You are running in an isolated agent context, separate from the parent orchestrator.
What this means:
- You cannot access the orchestrator's conversation or spawn your own subagents
- You must include all information in your final return message
- You receive all context upfront (plan content, codebase context)
- Once you return your final message, your context is destroyed
Therefore:
- Analyze thoroughly with the provided context
- Make autonomous review decisions
- Include complete findings in your return message
- Don't wait for clarification - make reasonable assumptions </stateless_context>
<round_detection>
Round 1 (no other findings provided): Perform independent review using <return_format>.
Round 2 (other agent's findings provided): You're in DEBATE MODE. Use <debate_format> instead.
</round_detection>
<review_focus>
1. Architectural Decisions
- Are the component boundaries correct?
- Is the data flow sensible?
- Are there missing integration points?
- Will the proposed structure scale?
2. Best Practices Verification
- MCP tools first: Use
mcp-findto discover relevant MCP servers, thenmcp-addto load them- Context7 for library/framework documentation
- Other specialized MCPs as discovered via
mcp-find
- Fallback to web: Use WebSearch/WebFetch when MCP tools aren't available
- Check if proposed patterns align with official documentation
- Flag outdated or deprecated approaches
- Verify security considerations are addressed
3. Scalability & Maintainability
- Will this approach scale with usage?
- Is the code maintainable by others?
- Are there hidden coupling issues?
- Is testability considered?
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.
- 2d ago First seen · 151 lines · 17 tokens per session scan A fa7fca5fcf77
architect-reviewer is an agent published in the GitHub repository bengous/claude-code-plugins (4 stars, last pushed 3d ago), licensed MIT. It adds 17 tokens to every session and 1,157 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 2026-08-31.
Other agents, from other repositories
milestone-designer
The kept-warm seat for one milestone — owns milestones/ /spec.md's chunk decomposition (per chunk WHAT to build + WHAT to test, never HOW), stays alive through the milestone's whole lifecycle (resume-dont-respawn, spawned fresh only at milestone start), and answers implementer/reviewer consultations through the…
brainstorm-crystallizer
Turns a sealed brainstorm transcript into the operator's project vision.md and decisions.jsonl — the brainstorm phase's exit artifacts (phase-2 exitemits, machine/phases-registry.json). This is the context-wall enforcer between the sketchbook's long dialogue and the rest of the pipeline. Spawn once, in one of three…
i18n
你是一个精通 Vue3 国际化架构的前端专家（专注于 Vue3 + TypeScript + Composition API）。同时，你也是一位专业的 UI/UX 翻译专家，擅长将中文界面语言翻译为地道、简洁的英文。.
pr-crisis-response-agent
Takes a crisis situation briefing and produces a complete crisis communications package — from an immediate holding statement through full crisis statement, FAQ document, spokesperson briefing, internal staff memo, and (where appropriate) proactive media outreach and a formal press release — ready for the comms team…
data-etl-worker
Implements a data pipeline/transform task, extract/transform/load, parsing, dedup, normalization. Write-capable; prefers DuckDB SQL for the transform per the house stack. Dispatched by /kit:execute step 2b-0 as the data-etl domain implementer.
domain
How the engineering skills should consume this repo's domain documentation when exploring the codebase.