> ajan_yetkinlikleri_rehberi
Ajan Yetkinlikleri & Şartname Rehberi
Otonom ajan sistemleri için RFC standartlarına tam uyumlu, makine tarafından çalıştırılabilir SKILL.md paketleri ve mimari rehberler.
Publish RFC 9309 Compliant robots.txt for AI Crawlers
Architect and deploy an RFC 9309 compliant /robots.txt with explicit crawler permissions for AI agents (ClaudeBot, GPTBot, PerplexityBot) and sitemap declarations.
▸ Define explicit User-agent rules for general search engines and dedicated AI training/search bots.
Generate Token-Optimized XML Sitemaps
Produce dynamic, XXE-safe XML sitemaps advertising canonical multilingual routes, last-modified timestamps, and change frequencies.
▸ Expose all public indexable URLs with canonical HTTPS domains.
Advertise Machine Surfaces via RFC 8288 Link Headers
Implement RFC 8288 HTTP Link response headers to advertise machine-readable endpoints (MCP server cards, API catalogs, llms.txt) on every HTML response.
▸ Include rel='mcp-server-card' linking to /.well-known/mcp/server-card.json.
Publish DNS-AID Discovery Records (RFC 9460)
Configure DNS-AID SVCB/TXT records enabling autonomous agents to discover capabilities at the network transport layer via DNS.
▸ Publish _index._agents TXT record on the domain zone apex.
Implement HTTP Markdown Content Negotiation
Support 'Accept: text/markdown' content negotiation so AI agents receive high-density token-efficient Markdown representations of web pages.
▸ Detect 'Accept: text/markdown' in request headers.
Configure AI Bot Crawler Rules & Disclosures
Establish transparent crawling and training guidelines for OpenAI, Anthropic, Google, and Perplexity autonomous agents.
▸ Permit AI search agents to index public documentation and failure catalogs.
Publish Content Signals HTTP Headers
Emit Content-Signals HTTP headers declaring AI training, search, and input usage rights per industry open standards.
▸ Set Content-Signals: ai-train=yes|no, search=yes, ai-input=yes.
Implement Web Bot Authentication Signatures
Authenticate autonomous bots via HTTP signatures and verifiable cryptographic credentials.
▸ Verify bot public keys and signature headers.
Deploy RFC 9727 API Catalog Linkset
Publish /.well-known/api-catalog using application/linkset+json to expose machine-readable OpenAPI and MCP service descriptions.
▸ Publish linkset at /.well-known/api-catalog.
Configure OAuth 2.0 Authorization Server Metadata (RFC 8414)
Expose RFC 8414 metadata at /.well-known/oauth-authorization-server declaring token endpoints, grant types, and agent_auth extensions.
▸ Publish /.well-known/oauth-authorization-server with issuer, token_endpoint, and jwks_uri.
Expose OAuth Protected Resource Metadata (RFC 9728)
Serve RFC 9728 metadata at /.well-known/oauth-protected-resource declaring authorization server associations and supported bearer methods.
▸ Publish /.well-known/oauth-protected-resource.
Author Agentic Registration Manual (auth.md)
Publish /auth.md providing step-by-step instructions for AI agents on how to discover, register, claim, and exchange tokens.
▸ Publish /auth.md with '# auth.md' top-level title.
Publish SEP-1649 MCP Server Card Manifest
Deploy /.well-known/mcp/server-card.json declaring Model Context Protocol server capabilities, transport endpoints, and callable tool schemas.
▸ Publish /.well-known/mcp/server-card.json.
Deploy Agent Skills Discovery Index (RFC v0.2.0)
Serve /.well-known/agent-skills/index.json exposing on-demand tool and workflow skills with SHA-256 integrity digests.
▸ Publish index.json listing available skills with name, description, and skill URL.
Integrate Client-Side WebMCP (navigator.modelContext)
Expose client-side browser tools and state to AI browser agents via the emerging WebMCP navigator.modelContext standard.
▸ Detect window.navigator.modelContext in browser runtime.
Publish ARD Capability Manifest (ai-catalog.json)
Serve /.well-known/ai-catalog.json declaring host capabilities, representative user queries, and OpenAPI schemas.
▸ Publish /.well-known/ai-catalog.json with host details and capabilities.
Prepare Incident Evidence
A structured workflow to gather, sanitize, and format system state evidence following an incident.
▸ Learn how to secure logs without exposing PII.
agentskills.io Standart Formatı İle Uyumlu
Tüm yetkinlikler Claude Code, Cursor, Windsurf ve Antigravity araçlarıyla doğrudan entegre edilebilir. Toplu keşif için RFC 9309 ve index.json kataloglarını kullanın.