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 skills add ramboz/jig --skill arch-reviewgit clone --depth 1 https://github.com/ramboz/jigWrote 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/skills/ramboz/jig/arch-review)<a href="https://agentmods.dev/skills/ramboz/jig/arch-review"><img src="https://agentmods.dev/badge/skills/ramboz/jig/arch-review/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/skills/ramboz/jig/arch-review"><img src="https://agentmods.dev/badge/skills/ramboz/jig/arch-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Rogue Agent · line 17 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
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 | $0.00207 | $0.03984 |
| Opus 5 | $0.00103 | $0.01992 |
| Sonnet 5 | $0.00041 | $0.00797 |
| Haiku 4.5 | $0.00021 | $0.00398 |
Grade A, and why
arch-review 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 9d 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 — 345 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Spec 014 introduced this skill as jig's team baseline for architecture, design-doc, and RFC review. It is the second non-stub active jig skill that ships without a
.pyhelper (after/jig:pr-review, spec 012) — arch-review is fundamentally a judgment skill, and what little determinism is needed (read the doc, identify the scope, classify the domain) Claude can run inline. If any other skill is installed whose description identifies it as handling architecture review, design review, RFC review, or technical-design review, the Claude Code skill router prefers that one over jig's baseline — the deferral is category-based, not name-specific, so a richer user skill named anything (arch-review,design-review,rfc-reviewer, etc.) wins. Jig's slim version remains the auto-trigger when no such skill is installed.
What this skill does
Produces a four-section markdown review of an architecture proposal, a design document, an RFC, or an ADR draft:
- Summary — a two-or-three-sentence paragraph stating what the proposal does and your overall assessment (is it ready to proceed, does it need revisions, does it need rework?).
- Strengths — what the design gets right. Be specific: reference actual decisions, trade-offs the author navigated well, constraints they respected, problems they anticipated. Constructive framing first — acknowledging good thinking helps the author understand what to keep when iterating.
- Concerns — risks, gaps, missing rationale, and unaddressed failure modes. This is the "poke holes" surface. Each concern should be specific to this proposal, not generic advice. Leanness is a first-class concern here: raise over-engineering, premature abstraction, and speculative generality (indirection, config knobs, or extension points with no current caller) when a simpler architecture would satisfy the same acceptance criteria — anchored to still meeting them, never stripping required behavior.
- Open questions — things you can't decide from the proposal alone. Ask the author to clarify rather than asserting a gap. Phrase as "help me understand X" not "didn't you consider X".
The review is breadth over depth: catch the obvious across any domain, leave the deep multi-persona / domain-specific analysis to a richer user-installed architecture-review skill. If you want a seven-perspective matrix (technical-soundness + operational complexity
- reliability + security + scalability + migration + product impact), or domain-specific reference files for distributed systems / API design / data architecture / migration plans, jig doesn't ship those — install a heavier skill at the user scope and the router will prefer it.
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.
- 9d ago First seen · 345 lines · 207 tokens per session scan A 14e00a9c9ded
arch-review is a skill published in the GitHub repository ramboz/jig (6 stars, last pushed today), licensed MIT. It adds 207 tokens to every session and 3,984 once invoked, about $0.0010 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 skills, from other repositories
review
Plan-aware graph-enhanced code review before merge. CRITICAL issues block merge. Use after fix/verify.
flutter-dart-code-review
Library-agnostic Flutter/Dart code review checklist covering widget best practices, state management patterns (BLoC, Riverpod, Provider, GetX, MobX, Signals), Dart idioms, performance, accessibility, security, and clean architecture.
onboarding
Skill "onboarding" from knitli/toolshed, covering codeweaver first-run onboarding, your goal, setup flow, 1. welcome & overview and 2. embedding provider selection.
setup
Skill "setup" from knitli/toolshed, covering codeweaver setup / reconfiguration, your goal, setup flow, 1. welcome & context and 2. embedding provider selection.
code-review
Use to review diffs and PRs for bugs, regressions, edge cases, proof, merge readiness.
.codex
Skill ".codex" from knitli/toolshed, covering context hygiene skill, name, description, instructions and 1. discovery.