
AgentCraft
agentcraft.devAI image and video prompt gallery with real outputs, plus guides on Claude Code and coding agents
- Scanned
- Category
- Directories & Aggregators
- Markdown
- Read this report
What is AgentCraft?
AgentCraft is a website that publishes AI prompts organised by the image and video models that produced them. The homepage says each entry shows its real output and credits its creator. It gives counts of 93 GPT Image 2.6 prompts and 593 Seedance 2.5 prompts, and the meta description says the site holds 5,088 Claude Code skills sorted by category. The llms.txt file describes the site as publishing developer-focused guidance on engineering AI coding agents, covering prompting, agent harness, and Claude Code and Codex skills. The homepage also links a 16-chapter course on building a code review agent with Claude Code and Codex. The input does not state prices or who operates the site beyond linked GitHub repositories.
Through a bot’s eyes
“AgentCraft is a website that shares AI image and video prompts with their real outputs and credits their creators, and publishes developer-focused guides on Claude Code, Codex and AI coding agents.”
Simulated from the response GoodForBotsBot received on , not a live answer from any AI assistant.
What it can read, and what stays hidden →Show less ↑
What a bot can read
A model reading this homepage learns that the site is built around a gallery of prompts for image and video generators. Entry titles such as isometric map posters and character montages are paired with model names including GPT Image 2, Nano Banana 2, Kling 3.0 and Seedance 2.5. Per-model prompt counts and links to the full image and video lists are present as text. Blog headlines cover measured studies of prompts and Claude Code skills, and a course teaser names a 16-chapter build of a code review agent. The llms.txt adds a site summary and a list of guides with their findings, which the homepage does not contain. The JSON-LD supplies the organisation name, URL, logo and three GitHub links. The homepage text does not describe the site's purpose in plain sentences; the title and meta description carry most of that overview.
What stays hidden
A model reading only this input cannot learn whether the prompt gallery, skills directory or course cost money, or what any plan includes. It cannot tell who runs the site or how to contact them, since no author, company or contact details appear beyond GitHub links. The Claude Code skills are known only by count and category sorting, and the blog headlines are the only description of how they are reviewed. The llms.txt file was truncated mid-entry, so the input does not show its full list of guides.
Good for Bots report
AgentCraft scores 69 of 100: the whole readable base, 19 of the 50 agent-layer points, no penalties.
robots.txt is present and valid. robots.txt is valid, with 1 user-agent group.
Sitemap is present and parses. Sitemap declared in robots.txt, listing 186 URLs.
Describes itself to machines with schema.org JSON-LD. All 4 pages read pass. The homepage describes itself to machines as Organization in valid schema.org JSON-LD.
Content is marked up so a reader can find it. 3 of 4 pages read pass; 1 partly. The page at /en/blog/claude-code-skills-rising-2026-11/ is partly structured: No <main> element and nothing carrying role="main".
Navigation links expose usable destinations. 11 of 11 assessed homepage navigation links expose usable destinations.
Links and buttons have accessible names. 122 of 122 assessed homepage links and buttons have a nonempty accessible name.
Form fields have names and recognizable semantics. 4 of 4 assessed homepage form fields meet the static naming and field-semantics checks.
Controls expose roles, states and relationships. 128 of 128 assessed homepage controls and role declarations meet the static role, state and relationship checks.
Each page declares one usable canonical URL. All 4 pages read pass. The homepage declares itself as its canonical URL.
Section links reach an element on the page. Both pages read pass. The page at /en/blog/claude-code-skills-rising-2026-11/ has 7 section links, and every one reaches an element on the page.
llms.txt is present and well formed. /llms.txt is well formed: 29 links across 3 sections.
llms.txt is descriptive, not a bare link list. /llms.txt is descriptive: a summary plus descriptions on 86% of links.
Serves markdown at its own URL. No /llms-full.txt and no markdown versions linked from llms.txt. +12.5 with its prompt.
Serves markdown when asked for it. None of the 4 pages read pass. Asking the homepage for markdown still returns text/html. +10 with its prompt.
States a position on AI crawlers. robots.txt names no AI crawlers, so every operator has to guess. +5 with its prompt.
Declares AI usage preferences. No usage declaration found in robots.txt or the final successful homepage response header. +2.5 with its prompt in Pro.
Advertises machine-readable resources with typed links. None of the 4 pages read pass. The homepage's links are all ordinary page furniture; none point an agent anywhere. +1.3 with its prompt in Pro.
Earned
The whole readable base above the line, 19 of the 50 agent-layer points below it.
3 prompts away
Still open
Behind the checks below, with prompts in Pro.
For the owner of agentcraft.dev
3 prompts, 27 points
They are in the owner view, ready to paste into Claude Code, Cursor or Codex, beside the badge and rescans.
All 36 checks
Evidence for each, in Markdown →Select a check to see what we found and how it is graded.
Readable base
50 / 50
3.3 of 3.3 points.
robots.txt is valid, with 1 user-agent group.
8.3 of 8.3 points.
Sitemap declared in robots.txt, listing 186 URLs.
8.3 of 8.3 points.
All 4 pages read pass. The homepage describes itself to machines as Organization in valid schema.org JSON-LD.
10 of 10 points.
3 of 4 pages read pass; 1 partly. The page at /en/blog/claude-code-skills-rising-2026-11/ is partly structured: No <main> element and nothing carrying role="main".
4.2 of 4.2 points.
122 of 122 assessed homepage links and buttons have a nonempty accessible name.
2.5 of 2.5 points.
4 of 4 assessed homepage form fields meet the static naming and field-semantics checks.
2.5 of 2.5 points.
128 of 128 assessed homepage controls and role declarations meet the static role, state and relationship checks.
1.7 of 1.7 points.
All 4 pages read pass. The homepage declares itself as its canonical URL.
0.8 of 0.8 points.
Both pages read pass. The page at /en/blog/claude-code-skills-rising-2026-11/ has 7 section links, and every one reaches an element on the page.
Agent layer
18.8 / 50
11.3 of 11.3 points.
/llms.txt is well formed: 29 links across 3 sections.
7.5 of 7.5 points.
/llms.txt is descriptive: a summary plus descriptions on 86% of links.
0 of 12.5 points.
No /llms-full.txt and no markdown versions linked from llms.txt.
0 of 10 points.
None of the 4 pages read pass. Asking the homepage for markdown still returns text/html.
0 of 5 points.
robots.txt names no AI crawlers, so every operator has to guess.
0 of 2.5 points.
No usage declaration found in robots.txt or the final successful homepage response header.
0 of 1.3 points.
None of the 4 pages read pass. The homepage's links are all ordinary page furniture; none point an agent anywhere.
Penalties
− 0
Penalty: 0 of 15 points.
Our crawler reaches the page without being challenged.
Penalty: 0 of 20 points.
Nothing on the homepage tells a crawler to skip it or to withhold its snippet.
Penalty: 0 of 25 points.
The captured homepage HTML contains 574 content words; completeness is not assessed.
Penalty: 0 of 25 points.
robots.txt allows every known search and user-fetch crawler to reach the homepage.
Penalty: 0 of 10 points.
The homepage is served over HTTPS and returns HTTP 200.
Penalty: 0 of 3 points.
The homepage provides both a non-empty title and meta description.
Penalty: 0 of 3 points.
The homepage is reached in 0 redirects.
Penalty: 0 of 3 points.
The homepage reached a plain crawler in 80 ms.
Agent capabilities
0 of 11 · no points
Each one found adds +1 to the badge.
Adds +1 to the badge when found.
No API catalog published.
Adds +1 to the badge when found.
No /auth.md published.
Adds +1 to the badge when found.
No AI Catalog published, so no MCP server card can be discovered.
Adds +1 to the badge when found.
No A2A Agent Card at /.well-known/agent-card.json, and none listed in an AI Catalog.
Adds +1 to the badge when found.
No usable Agent Skills index was found at either well-known path or in the AI Catalog.
Adds +1 to the badge when found.
No WebMCP declarations were found in the collected HTML or inline scripts. External scripts were not inspected.
Adds +1 to the badge when found.
No Web Bot Auth key directory at /.well-known/http-message-signatures-directory.
Adds +1 to the badge when found.
No HTTP 402 payment challenge on the homepage or robots.txt, and no manifest at /.well-known/x402.
Adds +1 to the badge when found.
No ACP discovery declaration established at /.well-known/acp.json.
Adds +1 to the badge when found.
No UCP discovery declaration established at /.well-known/ucp.
Adds +1 to the badge when found.
No public AMP discovery declaration established. Private or hidden agents may not advertise one to our crawler.
Site profile
- Login
- Not established
- API
- Not established
- Commerce
- Not established
It decides which conditional checks apply. “Not established” means no proof in what we read, not proof of absence.
Access & usage policy
Who can read this page, and for what purpose?
GoodForBotsBot received HTTP 200 via a direct request. No indexing restriction was observed in the supported homepage directives.
Sampled URL: https://agentcraft.dev/
Declared crawling rules
No matching refusal for this URL was found for the listed identities.
Sources and scope: declared crawling rules
Declared allow · Scan admission
- GoodForBotsBot/1.0 (+https://goodforbots.com/bot)
The recorded robots admission decision allowed our crawler to proceed.
GoodForBotsBot admission along the homepage request chain
https://agentcraft.dev/
Declared allow · robots.txt
- goodforbotsbot (Site diagnostics)
- * (Default robots group)
- oai-searchbot (ai-search)
- claude-searchbot (ai-search)
- perplexitybot (ai-search)
- mistralai-index (ai-search)
- kimi-searchbot (ai-search)
- amzn-searchbot (ai-search)
- meta-webindexer (ai-search)
- exasearchbot (ai-search)
- aiwebindex (ai-search)
- duckassistbot (ai-search)
- shapbot (ai-search)
- yandexadditional (ai-search)
- yandexadditionalbot (ai-search)
- gptbot (ai-training)
- claudebot (ai-training)
- mistralai-training (ai-training)
- kimibot (ai-training)
- ai2bot (Unknown)
- applebot-extended (ai-training)
- webzio-extended (ai-training)
- chatgpt-user (user-fetch)
- claude-user (user-fetch)
- perplexity-user (user-fetch)
- mistralai-user (user-fetch)
- kimi-user (user-fetch)
- amzn-user (user-fetch)
- diffbot-user (user-fetch)
- firecrawlagent (extraction)
- webzio (extraction)
- meta-externalfetcher (user-fetch, agent)
- google-extended (ai-training, grounding)
- meta-externalagent (ai-training, product-improvement)
- amazonbot (ai-training, product-improvement)
- ccbot (open-dataset, ai-training)
- googlebot (search)
- bingbot (search)
No matching refusal for this URL in the applicable robots group. This is a declaration, not an access test.
Acquisition of https://agentcraft.dev/
https://agentcraft.dev/robots.txt
User-agent: * allow: /
robots.txt is present and validStates a position on AI crawlersAllows search and user-fetch crawlers
Observed access
GoodForBotsBot received HTTP 200 via a direct request.
Sources and scope: observed access
Observed · Direct request
- GoodForBotsBot/1.0 (+https://goodforbots.com/bot)
Our request received HTTP 200. A status alone does not identify a WAF.
Our recorded direct transport attempt; intermediate attempt URLs are not retained
https://agentcraft.dev
HTTP 200
Observed · Final homepage response
- GoodForBotsBot/1.0 (+https://goodforbots.com/bot)
GoodForBotsBot received HTTP 200 via a direct request.
homepage response
https://agentcraft.dev/
HTTP 200 Transport: direct
Serves the homepage over HTTPS with HTTP 200Serves our crawler without a challenge
Indexing directives
No indexing restriction was observed in the supported homepage directives.
Sources and scope: indexing directives
Not observed · Homepage directives
No indexing restriction was observed in the supported homepage directives.
homepage response
https://agentcraft.dev/
Search snippets
No snippet restriction was observed in the supported homepage directives.
Sources and scope: search snippets
Not observed · Homepage directives
No snippet restriction was observed in the supported homepage directives.
homepage response
https://agentcraft.dev/
Search usage
No applicable preference for this use was established.
Sources and scope: search usage
Unstated · AIPREF Content-Usage
- goodforbotsbot (Site diagnostics)
- * (Default robots group)
- oai-searchbot (ai-search)
- claude-searchbot (ai-search)
- perplexitybot (ai-search)
- mistralai-index (ai-search)
- kimi-searchbot (ai-search)
- amzn-searchbot (ai-search)
- meta-webindexer (ai-search)
- exasearchbot (ai-search)
- aiwebindex (ai-search)
- duckassistbot (ai-search)
- shapbot (ai-search)
- yandexadditional (ai-search)
- yandexadditionalbot (ai-search)
- gptbot (ai-training)
- claudebot (ai-training)
- mistralai-training (ai-training)
- kimibot (ai-training)
- ai2bot (Unknown)
- applebot-extended (ai-training)
- webzio-extended (ai-training)
- chatgpt-user (user-fetch)
- claude-user (user-fetch)
- perplexity-user (user-fetch)
- mistralai-user (user-fetch)
- kimi-user (user-fetch)
- amzn-user (user-fetch)
- diffbot-user (user-fetch)
- firecrawlagent (extraction)
- webzio (extraction)
- meta-externalfetcher (user-fetch, agent)
- google-extended (ai-training, grounding)
- meta-externalagent (ai-training, product-improvement)
- amazonbot (ai-training, product-improvement)
- ccbot (open-dataset, ai-training)
- googlebot (search)
- bingbot (search)
No applicable AIPREF preference was established for this purpose.
Purpose: search; applicable robots group/path and this response's header
https://agentcraft.dev/
AI input and use
No applicable preference for this use was established.
Sources and scope: ai input and use
Unstated · AIPREF Content-Usage
- goodforbotsbot (Site diagnostics)
- * (Default robots group)
- oai-searchbot (ai-search)
- claude-searchbot (ai-search)
- perplexitybot (ai-search)
- mistralai-index (ai-search)
- kimi-searchbot (ai-search)
- amzn-searchbot (ai-search)
- meta-webindexer (ai-search)
- exasearchbot (ai-search)
- aiwebindex (ai-search)
- duckassistbot (ai-search)
- shapbot (ai-search)
- yandexadditional (ai-search)
- yandexadditionalbot (ai-search)
- gptbot (ai-training)
- claudebot (ai-training)
- mistralai-training (ai-training)
- kimibot (ai-training)
- ai2bot (Unknown)
- applebot-extended (ai-training)
- webzio-extended (ai-training)
- chatgpt-user (user-fetch)
- claude-user (user-fetch)
- perplexity-user (user-fetch)
- mistralai-user (user-fetch)
- kimi-user (user-fetch)
- amzn-user (user-fetch)
- diffbot-user (user-fetch)
- firecrawlagent (extraction)
- webzio (extraction)
- meta-externalfetcher (user-fetch, agent)
- google-extended (ai-training, grounding)
- meta-externalagent (ai-training, product-improvement)
- amazonbot (ai-training, product-improvement)
- ccbot (open-dataset, ai-training)
- googlebot (search)
- bingbot (search)
No applicable AIPREF preference was established for this purpose.
Purpose: ai-use; applicable robots group/path and this response's header
https://agentcraft.dev/
AI training
No applicable preference for this use was established.
Sources and scope: ai training
Unstated · AIPREF Content-Usage
- goodforbotsbot (Site diagnostics)
- * (Default robots group)
- oai-searchbot (ai-search)
- claude-searchbot (ai-search)
- perplexitybot (ai-search)
- mistralai-index (ai-search)
- kimi-searchbot (ai-search)
- amzn-searchbot (ai-search)
- meta-webindexer (ai-search)
- exasearchbot (ai-search)
- aiwebindex (ai-search)
- duckassistbot (ai-search)
- shapbot (ai-search)
- yandexadditional (ai-search)
- yandexadditionalbot (ai-search)
- gptbot (ai-training)
- claudebot (ai-training)
- mistralai-training (ai-training)
- kimibot (ai-training)
- ai2bot (Unknown)
- applebot-extended (ai-training)
- webzio-extended (ai-training)
- chatgpt-user (user-fetch)
- claude-user (user-fetch)
- perplexity-user (user-fetch)
- mistralai-user (user-fetch)
- kimi-user (user-fetch)
- amzn-user (user-fetch)
- diffbot-user (user-fetch)
- firecrawlagent (extraction)
- webzio (extraction)
- meta-externalfetcher (user-fetch, agent)
- google-extended (ai-training, grounding)
- meta-externalagent (ai-training, product-improvement)
- amazonbot (ai-training, product-improvement)
- ccbot (open-dataset, ai-training)
- googlebot (search)
- bingbot (search)
No applicable AIPREF preference was established for this purpose.
Purpose: train-ai; applicable robots group/path and this response's header
https://agentcraft.dev/
Coverage and interpretation
- Crawler coverage includes the registered identities and up to 50 user-agent entries from robots.txt. Source excerpts and detail groups are bounded; omissions are labelled.
- Observed access describes GoodForBotsBot only. Other crawler entries describe declared rules, not tests of those operators' access.
- Coverage is the sampled homepage response and collected robots.txt files. No indexing or citation is guaranteed. Usage preferences are declarations, not a legal determination of permission.
- This explanation adds no points or penalties. Any score effects belong to the linked checks.
- AIPREF and Content Signals keep their own definitions. AIPREF search can include processing used exclusively for search; training preferences are not a blanket ruling on every search process.
In its category
15th of 47 in Directories & Aggregators
- 1
Brivoybrivoy.com - 2
IndieLaunchindielaunch.dev - 15
AgentCraft · this reportagentcraft.dev
The prompts in the owner view would take AgentCraft to 96, 4th in its category.
To the prompts →Category average 62. The percentile appears once the catalogue holds a few hundred sites.
Scan details
Every scan is kept. Pro draws the history as a chart →
- Scanned
- 7 Nov 2026
- Rubric
- 0.9.0
- Requests
- 25
- Fetched
- 1.2 MB
- Took
- 16.1 s
- Homepage
- 80 ms · direct
- Pages read
- 4
- robots.txt
- allowed
Which pages the scan read
The homepage and up to three pages from the sitemap and homepage links. Structured data, page structure, Markdown negotiation and typed links check each page read and take the median page; other checks read the homepage.
- Homepagealways read
- /en/blog/claude-code-skills-rising-2026-11/article · linked from the homepage · from the sitemap · read
- /en/blog/collection · linked from the homepage · from the sitemap · read
- /en/prompts/image/linked from the homepage · from the sitemap · read
Do you run agentcraft.dev? Open the owner view → Removing the listing is free. Remove →