Graphify.md · community-built · not affiliated with Salesforce, Inc.

AgentForce as a traversable
knowledge graph

Graphify.md is a community-built project, not affiliated with Salesforce, that turns the AgentForce stack into a traversable knowledge graph for source-grounded lookup and dependency traversal. Built for Salesforce architects, platform engineers, and agent developers who need deterministic answers about billing paths, trust layer policy, action types, and grounding chains — with every result traceable to declared relationships rather than model inference.

Not a general-purpose semantic search layer — if it's not a declared edge, the graph doesn't return it.
$pip install ckg-agentforce

"The graph doesn't guess — it traverses. Every answer traces to a declared edge."

PyPI v0.2.4 40 nodes · 52 edges F1 0.471 · 4× RAG KRB v0.6.2 ELv2 data · MIT code Graphify.md
$2 per autonomous resolution. Every failed resolution is a retry, a CSAT hit, and $2 with no outcome. Wrong inference is expensive here. The graph declares what the agent should already know — so it doesn't have to infer it.

The graph

40 nodes · 52 edges · hover to explore · colored by layer

Platform Agent Trust Billing Capability Action

What developers are actually hitting

Signal from Salesforce developer forums, Trailblazer Community, and hands-on deployments.

01 — The $2 retry problem

Wrong inference = wrong billing

An agent misidentifies the resolution path. It attempts resolution, fails criteria, retries — $4 spent, zero outcome. The billing trigger lives four hops from Einstein Agent. Agents that don't traverse the full chain get it wrong.

02 — Trust layer blocking

"Which policy tier is blocking my agent?"

Einstein Trust Layer gates six downstream capabilities. Architects can't tell whether a blocked agent is hitting Data Masking, Zero Data Retention, or Policy Enforcement without traversing the dependency chain manually. The graph makes it a one-hop query.

03 — The Grounding Source Gap

Grounding fails silently

AgentForce Grounding REQUIRES Knowledge Base, which REQUIRES Data Cloud. Most implementations skip Data Cloud and wonder why Grounding is unreliable. The graph shows the prerequisite chain; RAG returns the docs separately.

What it is

The full AgentForce stack as a typed dependency graph. Pre-structured, traversable, deterministic. Served over MCP. No inference at query time.

Billing path traversal

get_prerequisites("Autonomous Resolution")

→ Autonomous Resolution  ← $2/event trigger
  ├─ [REQUIRES] Resolution Criteria
  ├─ [REQUIRES] Audit Trail
  │    └─ [REQUIRES] Einstein Trust Layer
  │         └─ [REQUIRES] AgentForce Platform
  └─ [REQUIRES] Policy Enforcement
       └─ [REQUIRES] Einstein Trust Layer

269 tokens · declared edges only
RAG: ~2,982 tokens · probabilistic

Trust layer dependency map

query_ckg("Einstein Trust Layer")

→ Dependents (what ETL gates):
  ← [REQUIRES] Data Masking
  ← [REQUIRES] Audit Trail
  ← [REQUIRES] Zero Data Retention
  ← [REQUIRES] Grounding
  ← [REQUIRES] Einstein Agent
  ← [REQUIRES] Policy Enforcement

Six capabilities gate on this one node.
RAG returns six separate docs.

Declared relationships, not confidence scores

Every edge in the graph was extracted from a Salesforce source document and given a type. There are no probabilistic weights, no cosine similarity scores, no confidence intervals. An edge either exists — declared, typed, sourced — or it doesn't. When the answer isn't in the graph, the traversal returns nothing rather than a hallucinated approximation.

REQUIRES

Hard prerequisite. A cannot function without B — e.g., Einstein Agent REQUIRES Einstein Trust Layer. The load-bearing dependency type.

ENABLES

Capability unlock. A makes B possible — e.g., Model Selection ENABLES Reasoning Engine. Used for optimization paths.

IMPLEMENTS

Concrete instantiation — e.g., NVIDIA NIM IMPLEMENTS Reasoning Engine. Useful for finding which backend serves an abstraction.

RELATES_TO

Conceptual proximity without direction — e.g., Data Masking RELATES_TO Zero Data Retention. Real but weaker than REQUIRES.

✗ RAG answer

"Einstein Trust Layer probably governs data access... (similarity: 0.79)"

Score is on the chunk, not the claim. The claim itself is unverified.

✓ CKG answer

"Einstein Trust Layer REQUIRES AgentForce Platform and gates: Data Masking · Audit Trail · Zero Data Retention · Grounding · Einstein Agent · Policy Enforcement."

No score. Declared edge. Traces to trust layer source doc.

A/B test — AgentForce domain, local models, no GPU

30 questions on the $2/resolution billing path, trust layer, and action types · CPU only · Ollama · temperature 0 · seed 42

+145%
Prereq-chain F1
+CKG
+121%
Billing path F1
+CKG
+113%
Trust layer F1
+CKG
+11pp
Key-fact
accuracy
P01 — billing prereq chain

What must resolve before AgentForce charges the $2 resolution fee?

✗ Bare model
"The agent completes the customer's request successfully..."
Vague. Misses the billing trigger chain entirely.
✓ With CKG
"Resolution Criteria must be satisfied, Audit Trail must log the event, and Policy Enforcement must clear — all gated by Einstein Trust Layer."
Exact chain. Declared edges, correct.
L03 — action type lookup

What are the four AgentForce action types?

✗ Bare model
"AgentForce supports Standard Actions, Custom Actions, and API Actions..."
Misses MuleSoft Action. Gets the count wrong.
✓ With CKG
"Flow Action · Apex Action · MuleSoft Action · External Action"
Declared edges, correct. Complete list.

Install

Add to claude.ai (no install required)

URLhttps://ckg-agentforce.onrender.com/mcp

Settings → Connectors → Add connector → paste URL above.

Local — Claude Desktop / Claude Code

$uvx ckg-agentforce
// claude_desktop_config.json
{ "mcpServers": {
  "agentforce": {
    "command": "uvx",
    "args": ["ckg-agentforce"]
  }
}}

Tools

ToolDescription
list_concepts()All 40 AgentForce concepts grouped by layer
search_concepts(query)Fuzzy search across all concepts
query_ckg(concept, depth)Typed subgraph around any concept (1–5 hops)
get_prerequisites(concept)Full upstream prerequisite chain in dependency order
resolution_path()The exact $2/resolution billing traversal — every hop declared
verify_source(concept)Source URL + SHA-256 hash · GuardrailDecisionV1 audit chain

What's in the graph

40 nodes · 52 edges · edge types: REQUIRES · ENABLES · IMPLEMENTS · RELATES_TO

Agents

Einstein Agent · Service Agent · Sales Agent · Marketing Agent

Actions

Flow Action · Apex Action · MuleSoft Action · External Action · Standard Action · Custom Action

Platform

AgentForce Platform · Data Cloud · Salesforce CRM · Knowledge Base

Reasoning

Reasoning Engine · Model Selection · NVIDIA NIM · Token Budget · Context Window

Trust

Einstein Trust Layer · Data Masking · Zero Data Retention · Policy Enforcement · Compliance Rules

Billing

Autonomous Resolution ($2/event) · Audit Trail · Agent Metrics · Handoff to Human

Workflow

Agent Topic · Agent Instruction · Resolution Criteria · Conversation State · Grounding

Routing

Omni-Channel Routing · Multi-Agent Orchestration · Prompt Template

Sources

Every node and edge traces to one of these. No probabilistic inference — declared relationships only.

TypeSourceCoverage
Officialdeveloper.salesforce.com/docs/einstein/genai/Platform, agents, actions, grounding chain
OfficialSalesforce Einstein Trust Layer docsTrust, Data Masking, ZDR, Policy Enforcement
OfficialAgentForce billing and resolution docsAutonomous Resolution, $2/event trigger, Audit Trail
OfficialData Cloud integration guideData Cloud, Knowledge Base, Grounding chain
OfficialModel selection and NIM integration docsReasoning Engine, NVIDIA NIM, Model Selection
Datasethuggingface.co/datasets/danyarm/ckg-benchmarkKRB v0.6.2 — 7,928 queries
Benchmarkgithub.com/Yarmoluk/ckg-benchmark/paper/main.pdfFull methodology, F1 0.471, RAG/GraphRAG baselines

Benchmark (KRB v0.6.2 locked)

0.471
CKG Macro F1
269 tokens · $7.81/1k
0.123
RAG baseline
2,982 tokens · $76.23/1k
0.120
GraphRAG baseline
~3,000 tokens · ~$76/1k

7,928 queries · 5-hop F1: 0.772 (CKG) vs 0.170 (RAG) · dataset public on HuggingFace · run it yourself

Try it now

Add the AgentForce graph to your agent in 30 seconds

No API key. No GPU. No inference at query time. Add the URL connector in claude.ai or run locally with one command.

97 domains available · ckg-mcp serves the full Graphify.md ecosystem · Free tier included