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 skills/kirodotdev/kirocrew/artifactsnpx skills add kirodotdev/KiroCrew --skill artifactsgit clone --depth 1 https://github.com/kirodotdev/KiroCrewWhat 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 | $0.00052 | $0.04075 |
| Opus 5 | $0.00026 | $0.02037 |
| Sonnet 5 | $0.00010 | $0.00815 |
| Haiku 4.5 | $0.00005 | $0.00407 |
Grade B, and why
artifacts scanned grade B with 1 finding 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.
Subtle steeringmediumPrompt injection
Instructions that bias recommendations or shape behaviour without the user noticing.
Critical: **never tell the user "the widget wasn't saved, so I can't How it starts
The opening of the file, as written. The whole thing — 373 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Artifacts (@kirocrew-core/artifact_*)
A widget rendered inline in chat (<mcwidget>) is transient — it scrolls
away with the conversation. An artifact is a widget (or other content)
that's been given a stable identity, a version history, and a URL the user
can open from /artifacts/<slug> in the dashboard.
Artifacts exist so the user can build a library of durable, named work — useful UIs (CR queue, pipeline health, ticket triage, dashboards) and substantial documents (plans, design docs, analyses, reports) alike — to return to, iterate on across sessions without losing prior versions, or share with peers for feedback. The library is curated: it holds work worth keeping, not a mirror of every file or a copy of the chat.
Mental model
| Concept | Means |
|---|---|
| Slug | URL-safe identifier like cr-queue (auto-derived from name). Stable across versions. The user references artifacts by slug. |
| Version | Monotonic integer. Every content change bumps it. The 50 most-recent versions are retained; older ones get pruned (configurable via MAX_VERSIONS). |
| Save | Persist the artifact for the first time. Picks a slug, returns it. |
| Update / Iterate | Modify content of an existing artifact. Bumps version, preserves history. |
| List / Find | Discover what's saved. Filter by tag, kind, name substring. |
Tools
| Tool | Purpose |
|---|---|
artifact_save |
Create a new artifact, returns slug |
artifact_get |
Load content + metadata (optional version) |
artifact_update |
Modify content/metadata; bumps version on content change |
artifact_list |
Filter by tag, kind, q (name substring) |
artifact_versions |
List version numbers for a slug |
artifact_delete |
Permanently remove |
All under the @kirocrew-core MCP server.
When to save proactively
When you produce work worth keeping — something the user would plausibly want later — save it:
- A widget with its own identity (CR queue, pipeline dashboard, ticket
card): save it without asking, and note it in one line — "Saved as artifact
cr-queue." - A substantial document (plan, design, analysis, report, reference): if you
wrote it to a workspace file, save it file-backed (
artifact_savewithsource_path, andkind="markdown"for markdown); if it lives only inline in chat, offer to save 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.
- 2d ago First seen · 373 lines · 52 tokens per session scan B 6d710a0505bf
artifacts is a skill published in the GitHub repository kirodotdev/KiroCrew (3,532 stars, last pushed today), licensed Apache-2.0. It adds 52 tokens to every session and 4,075 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it B with 1 finding (subtle steering). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
ppt-generation
Generate PPTX presentations from slide plan + content.
chart-visualization
Generate charts: select type, extract data, render image.
weshop-openapi
Use for WeShop image and video generation, editing, and transformation tasks — virtual try-on, model swap, background replace, pose change, canvas expand, background removal, AI video generation, product/photo generation, and more.
inspect-weshop-canvas
Read, generate for, edit, and operate the current WeShop for DeepSeek Harness canvas. Use whenever the user asks about canvas materials/results or asks the agent to generate, upscale, transform, reverse-prompt, or locally edit an image for the canvas. All generated outputs must be published automatically with…
media-validator
Validate a recording, screenshot or capture by reading its actual content and returning a PASS/FAIL verdict with evidence.
emil-design-eng
This skill encodes Emil Kowalski's philosophy on UI polish, component design, animation decisions, and the invisible details that make software feel great.