Skip to content

Built-in examples

OpenExpertise ships with 12 examples covering every node kind, every fan-out pattern, and three real cognitive workflows (deep research, systematic debugging, brainstorming).

Each example has a README.md, an experience.yaml, fixtures, and a mocked-LLM e2e test — no API key needed to validate the structure.

Hero examples

Review Branch ★ : The hero demo. Multi-dim code review + verifier + score + advisor evolution. Three reviewers (bugs/perf/tests) fan out, find issues, miss SQL injection — advisor proposes adding a security dimension. : Nodes: tool + agent ×3

Tri-CLI Orchestration ★ : Claude → Codex → Gemini in one DAG. The headline cross-vendor demo. : Nodes: cli-agent ×3

Deep Research : Multi-vendor research: Claude Code WebSearch + Gemini Google Search → cited synthesis. : Nodes: tool + agent + cli-agent ×2

Systematic Debugging : The superpowers systematic-debugging skill as a YAML flow. Hypothesize → verify → fix via Claude Code → re-test. : Nodes: tool ×2 + agent + cli-agent

Brainstorming : The superpowers brainstorming skill: diverge across 3 angles → cluster → critique → synthesize top 3. : Nodes: tool + agent ×2 + cli-agent ×2

Primitive demos (one node kind at a time)

Hello Tool — Smallest possible flow. No LLM. tool only.

Dataset Aggregate — Load CSV → aggregate. dataset + tool.

Agent Echo — Single agent with structured output. agent.

Oncall Runbook — Investigate an incident across 3 dimensions via for_each fan-out.

Issue Triage — Classify → search dupes → conditional dedup → route. Shows when: edges.

Release Gates — License + changelog + coverage + Claude-Code security scan → release gate.

CLI Orchestration — Two providers: Claude Code summarizes, Codex critiques.


All examples at a glance

ExampleFeaturingLLM required?CLIs required?
hello-tooltool nodenono
dataset-aggregatedataset + toolnono
agent-echoagent nodeyesno
review-branch ★for_each, when:, array_append, multi-phaseyesno
oncall-runbookfor_each fan-out, synthesis pipelineyesno
issue-triagewhen: conditional, skip-cascadeyesno
release-gatestool + cli-agent + agent mix, fan-inyesclaude
cli-orchestrationcli-agent chain, 2 providersnoclaude, codex
tri-cli-orchestration ★3-provider CLI chainnoclaude, codex, gemini
deep-researchdual for_each, fan-in, multi-vendor searchyesclaude, gemini
systematic-debuggingcli-agent for_each, 6-phase fix loopyesclaude
brainstormingcli-agent fan-out ×2, cluster + synthesizeyesclaude

Running any example

bash
# Validate YAML schema (no LLM or CLIs needed)
oe validate examples/<name>

# Run with live dashboard
oe run examples/<name> --tui

# Run with args
oe run examples/<name> --args '{"key":"value"}'

# Inspect state after run
oe state <field>

# Full event log
oe inspect <run-id>

# Trigger evolution advisor
oe evolve <run-id>

Want to share your own experience? See Sharing experiences (coming soon).

Released under the MIT License.