CrewAI + Hindsight Memory

Give your CrewAI crews persistent long-term memory. Run a crew multiple times and watch it build on what it learned in previous runs.

What This Demonstrates​

• Drop-in memory backend for CrewAI via hindsight-crewai
• Persistent memory across runs - crews remember previous research
• Reflect tool - agents explicitly reason over past memories
• Bank missions - guide how Hindsight organizes memories

Architecture​

Run 1: "Research Rust benefits"
    │
    ├─ Researcher agent ──► Hindsight reflect (no prior memories)
    │                        ──► produces research findings
    ├─ Writer agent ────────► summarizes findings
    │
    └─ CrewAI auto-stores task outputs to Hindsight
                    │
Run 2: "Compare Rust with Go"
    │
    ├─ Researcher agent ──► Hindsight reflect (recalls Rust research!)
    │                        ──► builds on prior knowledge
    ├─ Writer agent ────────► writes comparative summary
    │
    └─ Memories accumulate across runs

Prerequisites​

1. Hindsight running

   export OPENAI_API_KEY=your-key
   
   docker run --rm -it --pull always -p 8888:8888 -p 9999:9999 \
     -e HINDSIGHT_API_LLM_API_KEY=$OPENAI_API_KEY \
     -e HINDSIGHT_API_LLM_MODEL=o3-mini \
     -v $HOME/.hindsight-docker:/home/hindsight/.pg0 \
     ghcr.io/vectorize-io/hindsight:latest
   

2. OpenAI API key (for CrewAI's LLM)

   export OPENAI_API_KEY=your-key
   

3. Install dependencies

   cd applications/crewai-memory
   pip install -r requirements.txt
   

   Note: hindsight-crewai is not on PyPI — it is installed directly from the Hindsight repo via git.

Quick Start​

# First run - the crew has no memories yet
python research_crew.py "What are the benefits of Rust?"

# Second run - the crew remembers the Rust research
python research_crew.py "Compare Rust with Go"

# Third run - the crew has context from both prior runs
python research_crew.py "Which language should I pick for a CLI tool?"

# Reset memory and start fresh
python research_crew.py --reset

How It Works​

1. Configure Hindsight​

from hindsight_crewai import configure, HindsightStorage

configure(hindsight_api_url="http://localhost:8888", verbose=True)

storage = HindsightStorage(
    bank_id="research-crew",
    mission="Track technology research findings, comparisons, and recommendations.",
)

2. Add the Reflect Tool​

The HindsightReflectTool lets agents explicitly query their memories with disposition-aware synthesis:

from hindsight_crewai import HindsightReflectTool

reflect_tool = HindsightReflectTool(bank_id="research-crew", budget="mid")

researcher = Agent(
    role="Researcher",
    goal="Research topics thoroughly",
    backstory="Always use hindsight_reflect to check what you already know.",
    tools=[reflect_tool],
)

3. Wire Up the Crew​

from crewai.memory.external.external_memory import ExternalMemory

crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, summary_task],
    external_memory=ExternalMemory(storage=storage),
)

crew.kickoff()

CrewAI automatically:

• Queries memories at the start of each task
• Stores task outputs after each task completes

Core Files​

Customization​

Per-Agent Memory Banks​

Give each agent isolated memory:

storage = HindsightStorage(
    bank_id="my-crew",
    per_agent_banks=True,  # "my-crew-researcher", "my-crew-writer"
)

Custom Bank Resolver​

Full control over bank naming:

storage = HindsightStorage(
    bank_id="my-crew",
    bank_resolver=lambda base, agent: f"{base}-{agent.lower()}" if agent else base,
)

Configuration Options​

See the hindsight-crewai documentation for the full API reference.

Common Issues​

"Connection refused"

• Make sure Hindsight is running on localhost:8888

"OPENAI_API_KEY not set"

export OPENAI_API_KEY=your-key

"No module named 'hindsight_crewai'"

pip install -r requirements.txt

Built with:

• CrewAI - Multi-agent orchestration
• hindsight-crewai - Hindsight storage backend for CrewAI
• Hindsight - Long-term memory for AI agents