Skillget

agent-infrastructure-landscape

mcp
Security Audit
Warn
Health Warn
  • No license — Repository has no license file
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

AI agent memory & infrastructure landscape — comparative catalog of 912 systems × 68 columns covering memory layers, agent frameworks, runtimes, vector stores, knowledge graphs, MCP servers, benchmarks. Searchable with typed edges, lineages, citations.

README.md

AI Agent Infrastructure Landscape

An open, comparative catalog of the tools developers use to build, deploy,
and operate autonomous AI agents — the software systems (Claude Code,
Cursor, AutoGen, LangGraph, Mem0, Zep, etc.) that take instructions, plan,
call tools, remember things across sessions, and act in the world.

Each entry is tracked across 85 attributes (license, maturity tier,
deployment model, MCP / A2A protocol support, observability stack,
compliance posture, latency, pricing, and dozens more), with typed
relationships between systems (built-on, extends, competes-with,
cites…). Today the catalog covers 912 systems and 528 relationships.

The goal is straightforward: give builders, researchers, and analysts a
single place to compare what exists, see what's actually shipping vs.
just published, and spot the gaps.

Start here

What's in the data

Each record is a JSON object with a stable ID, display name, tier
(1 battle-tested → 5 theoretical), one or more section memberships
(e.g. "Dedicated memory layers", "Framework-embedded memory",
"Vector-database infrastructure" — 34 sections in total), and the 85
attribute cells. Every cell carries:

  • The claim itself (free-text value).
  • A source URL (where the claim came from).
  • A status (real-data, estimate, not-applicable, no-data, etc.).
  • A provenance tier (T1 auto-verifiable from a GitHub URL, T2
    resolvable source URL required, T3 estimate).
  • A last-verified date for high-volatility cells.

Sources include curated lists (Agent-Memory-Paper-List,
Awesome-GraphMemory), survey papers, benchmark leaderboards (LongMemEval,
LoCoMo, ConvoMem), vendor sites, academic venue pages, and targeted
research-agent sweeps. Claims are vendor-stated unless otherwise marked.
Honest coverage confidence is roughly 88–92% — known gaps are tracked
in PLAN.md. Full schema: docs/SCHEMA.md.

Use it in your own tools

MCP server — query the catalog from any Model Context Protocol
client (Claude Code, Claude Desktop, Continue, etc.). Nine read-only
tools: search, get-by-id, edge traversal, coverage stats, side-by-side
comparison, recent changes, eval-orphan detection, substrate
blast-radius analysis. See mcp/ and
mcp/README.md.

cd mcp && npm install && npm run build
claude mcp add landscape -- node $PWD/dist/server.js

CLI — the same nine queries as a landscape <subcommand> tool.
Text output by default, --json / --csv for machine-readable use.
See cli/ and cli/README.md.

cd cli && npm install && npm run build
./dist/landscape.js search "memory" --tier 1 --section "Dedicated memory layers"

Web app — SvelteKit static export in web/, deployed to
GitHub Pages on every push to main.

cd web && npm install
npm run dev      # local server at http://localhost:5173/
npm run build    # static export

How the catalog stays honest

Three things keep the data trustworthy:

  1. Per-cell provenance. Every claim points at the source URL it was
    sourced from, with a tier marking how verifiable the claim is. You can
    audit any entry by walking the citation.
  2. Automated freshness checks. A weekly job
    (.github/workflows/staleness.yml) flags rows whose upstream repo
    has gone quiet beyond the freshness SLA defined in
    MAINTAINER.md §2. Stale rows surface in the live
    table with a visible badge.
  3. Self-maintaining workflows. New systems submitted via the
    /submit
    form (or an intake-labelled GitHub Issue) are auto-researched into a
    draft PR (docs/INTAKE.md). Existing sections are
    periodically re-audited section-by-section
    (docs/AUDIT.md).

Editing the catalog (contributors)

data/landscape.json is the source of truth — the rendered
landscape.html is a build artefact, never edited by hand.

  1. Edit data/landscape.json directly, or let the intake-research /
    section-audit bots open a PR for you. PRs get a rendered-cell preview
    comment from .github/workflows/diff-preview.yml so reviewers see
    exactly what each touched row will look like.

  2. Run make build locally — reconciles JSON, rebuilds edges and
    citation trajectories, re-renders landscape.html.

  3. Run make validate — five offline gates, ~25 seconds:

    # Gate Catches
    1 JSON schema Records or edges violating docs/SCHEMA.md §7.
    2 Fast-step determinism Non-determinism in extract / reconcile / edge-build scripts.
    3 Render-cycle stability Markup drift in render ↔ extract round-trips beyond the documented ceiling.
    4 S2 cache integrity Corrupted Semantic Scholar cache files.
    5 Tier-provenance + freshness Cells whose tier disagrees with their citation, or rows missing required date metadata.

  4. Commit the JSON, the regenerated edges file, and the regenerated
    landscape.html together. CI re-runs validation plus a byte-identity
    check (the JSON must round-trip cleanly to the committed HTML) on
    every push.

A pre-commit hook is available via make install-hooks (idempotent;
short-circuits on changes that don't touch the pipeline).
make refresh-citations re-pulls Semantic Scholar data (~15 min); only
needed when adding research-paper rows.

Governance and license

This catalog has an explicit maintenance contract:
MAINTAINER.md. It defines what's in scope (and what
isn't), the freshness SLA per cell type, the 3-tier claim-validation
schema, the succession plan, and how to contribute or request
co-maintainer rights. The contract exists because comparative catalogs
historically die quietly — DB-Engines, State of JS, dbdb.io — and the
surviving ones all publish one.

The catalog data is released under CC-BY-4.0. The MCP server and
CLI packages are MIT (see their package.json).

Credits

This catalog cross-references, corroborates, and (where licensing
allows) imports data from the following external landscape projects.
Vendored snapshots and the full per-source licence audit live in
extraction/external/.

Reviews (0)

No results found