Agent Quickstart

Use Git.Top as the Knowledge Graph of Open Source.

A short, production-oriented path for agents and developers to move from health checks to workflow, Atlas journeys, project knowledge, recommendations, comparison, MCP, GRP, and trust evidence.

Open Quickstart JSON Open Recipes Check Health Try Workflow Agent Map Quality
RESThttps://git.top/api/*

Structured project knowledge, recommendations, graph, compare, quality, and roadmap.

MCP corehttps://git.top/mcp/core

Five-tool first-use profile for project search and selection.

MCPhttps://git.top/mcp

JSON-RPC discovery and tools for agent runtimes.

GRPhttps://git.top/api/grp/query

Goal-level graph reasoning for plans, stacks, comparisons, and project sets.

check-trust-gate

Check Trust Gate

GET /api/trustMCP get_trust_gate

Get a single production-readiness decision before making a high-confidence recommendation.

curl https://git.top/api/trust
decisionproduction_readychecks[].statusquality.release_scorequality.data_trust_scoremetadata.source

Use the gate decision to decide whether to answer directly, disclose caveats, or fail closed.

check-data-source

Check Data Source

GET /api/health

Verify production data is D1-backed before making a high-confidence recommendation.

curl https://git.top/api/health
db=availablemetadata.source=d1metadata.reason=d1_querysync_freshness

Use require_d1=true on critical REST calls when an agent should fail closed instead of accepting seed fallback.

choose-workflow

Choose A Workflow

GET /api/workflowMCP get_agent_workflow

Use a guided path when the task spans trends, recommendations, graph, alternatives, score, compare, and trust checks.

curl "https://git.top/api/workflow?intent=choose%20a%20Cloudflare-ready%20agent%20framework&deployment=cloudflare&category=agent_framework&cloudflare_ready=true&limit=5&require_d1=true"
recommended_sequenceshortlisttrend_contextagent_maptrust_policy

Start here for broad or ambiguous project-selection tasks after confirming the trust gate.

inspect-surface-map

Inspect The Surface Map

GET /api/agent-map

Map each Git.Top product concept to the human page, REST endpoint, MCP tool, output fields, and trust fields.

curl https://git.top/api/agent-map
surfaces[].conceptsurfaces[].restsurfaces[].mcp_toolssurfaces[].trust_fields

Use this before guessing whether to call recommendations, alternatives, graph, compare, score, Atlas, GRP, or quality endpoints. Treat short_path as the default route and reference_path as the expansion route.

search-projects

Search Projects

GET /api/searchMCP search_projects

Retrieve candidate projects by query, category, deployment, language, project kind, and Cloudflare readiness.

curl "https://git.top/api/search?category=agent_framework&deployment=cloudflare&ranking=browse&limit=10&require_d1=true"
projects[].repoprojects[].project_kindprojects[].collection_metadatametadata.source

Use browse ranking for category discovery and query search for direct retrieval.

explore-atlas-journeys

Explore Atlas Journeys

GET /api/journeysMCP get_atlas

Start from an ecosystem map and move into ordered journeys plus comparison paths before choosing projects.

curl https://git.top/api/journeys?limit=8&require_d1=true
journeys[].stepscomparison_pathscomparison_paths[].api_hrefstats.comparison_path_count

Use comparison_paths when the user asks for ecosystem-specific shortlists or map-to-compare decisions.

inspect-project

Inspect One Project

GET /api/project/:projectMCP get_project

Fetch structured project knowledge and the compact agent summary before recommending, comparing, or citing a repository.

curl https://git.top/api/project/cloudflare/agents?require_d1=true
summary.purposesummary.good_forsummary.not_good_forquality_signal_confidencemetadata.source

Use summary for first-pass agent reasoning, then check classification evidence before saying a project is Cloudflare-ready or production-fit.

recommend

Ask For A Recommendation

GET /api/recommendMCP recommend_project

Turn a concrete use case and constraints into an explainable shortlist.

curl "https://git.top/api/recommend?use_case=build%20a%20browser%20automation%20agent&deployment=docker&limit=5&require_d1=true"
recommendations[].decision_summaryfit_profileadoption_planrisk_flagsconfidence

Present the top project, one alternative, caveats, and metadata source.

compare

Compare Candidates

GET /api/compareMCP compare_projects

Use decision matrices instead of raw star counts when the user already has candidates.

curl "https://git.top/api/compare?repos=cloudflare/agents,langchain-ai/langchain,run-llama/llama_index&deployment=cloudflare&require_d1=true"
summarydecision_matrixwinnerreasoningnext_actions

Move from comparison into graph, alternatives, score, or project pages before finalizing.

quality

Check Trust And Coverage

GET /api/qualityMCP get_quality_report

Inspect release score, corpus data trust, coverage, stale data, and review risk before high-confidence answers.

curl https://git.top/api/quality?require_d1=true
release_scoredata_trust_scorerisk_levelimprovement_plancoverageissuesmetadata.source

Use the trust gate first, then disclose degraded sync, seed fallback, low classification confidence, or partial quality signals.

grp

Use GRP For Goal-Level Planning

POST /api/grp/queryMCP git_top_grp_query

Use graph reasoning when the user asks for a plan, stack, comparison, or project set.

curl -X POST "https://git.top/api/grp/query?require_d1=true" -H "content-type: application/json" -d '{"goal":"build a Cloudflare-ready coding agent stack","mode":"plan","constraints":{"deploy":["cloudflare"],"agent_ready":true}}'
solution_pathsrecommended_stacknodesedgesmetadata.data_source

Use GRP output as a plan, then fetch project, graph, score, and compare details for the final answer.