Skip to content
IAM Insights

Explainers · Glossary

CrewAI

An open-source framework for orchestrating teams of role-based AI agents.

reference →

In our briefings

The claim each of the newest briefings on CrewAI would have you leave with.

  1. ZeroID's delegation narrows authority at every hop, a mechanism we did not find in Uber's write-up. A sub-agent registers its own identity and keypair and signs an actor token; the orchestrator presents that token to the token service and asks it to delegate, and the service returns a token whose scope is the intersection of what the orchestrator requested for the sub-agent and what it held itself, keeps the full chain of actors in the token, and refuses to go deeper than a policy's maximum delegation depth. Revocation is the mirror: revoke any token in a chain and everything downstream is invalid at once, checked through a networked session call rather than local key verification. It is one implementation, published as code, from the vendor that wrote it.

    Highflame's open-source ZeroID narrows an agent's token at every hop, a step we did not find in Uber's write-upBI-FIND-2026-0137

  2. In the authors' prototype, one full AgentDID check took about 13.5 seconds per pair of agents, and it held there from one pair up to 50 running at once while throughput rose from 0.07 to 3.25 completed checks a second. The identity step took about 6.5 seconds, which the authors put down mostly to blockchain lookups on Sepolia, an Ethereum test network; the readiness probe took about as long, mostly waiting on the external model API, and the context check under a second. Registering each agent's DID is a separate one-time step outside that figure, at 15.37 seconds and $0.88 priced from mainnet averages. Fifty pairs was the largest load tested, though the introduction asks how the identities of millions of agents can be managed.

    AgentDID checks an AI agent's identity and readiness in 13.5 seconds, but was tested on at most 50 agent pairsBI-FIND-2026-1001

  3. Dalugoda's Human Delegation Provenance token carries a signed record of a person's authorization and an append-only chain of hand-offs between agents, which anyone can check offline with the issuer's public key and the session ID. The check confirms the chain is intact, current and within its hop limit; whether an agent's action fits what the person asked for is left to the application, because, in his words, "HDP provides the record, not the enforcement." In v0.1 the issuer's key signs every hop, so a hop shows it was "recorded at the issuer, not that the specific agent produced it." He also concedes that an injected instruction which gets a legitimate agent to log a genuine hop with a misleading summary "is not detectable by the protocol alone."

    A proposed record of who authorized an AI agent's actions cannot yet prove which agent took themBI-FIND-2026-0970

›Cited in 8 briefings

Every briefing that discusses this term, newest first.

  1. Highflame's open-source ZeroID narrows an agent's token at every hop, a step we did not find in Uber's write-up

    Article · · Highflame (GitHub) · Sep 25, 2026

  2. AgentDID checks an AI agent's identity and readiness in 13.5 seconds, but was tested on at most 50 agent pairs

    Paper · · arXiv · Minghui Xu et al. · Apr 28, 2026

  3. A proposed record of who authorized an AI agent's actions cannot yet prove which agent took them

    Paper · · arXiv · Asiri Dalugoda, Helixar · Apr 6, 2026

  4. MACAW Security's trust layer signs every agent call, but what stopped its prompt-injection test was a policy

    Paper · · arXiv · Mohan Rajagopalan et al. · Feb 11, 2026

  5. A guide to securing agent plans has the planner choose each step's tools and leaves out agent identity

    Paper · · arXiv · Ron F. Del Rosario et al. · Sep 10, 2025

  6. The Aegis agent-security paper's 2.79-second proof time is a benchmark-based target its simulation reports back

    Paper · · arXiv · Sai Teja Reddy Adapala et al. · Aug 22, 2025

  7. A Tsinghua blockchain design for AI agents backs its defense claims with a table and reports no attack run

    Paper · · arXiv · Zhenhua Zou et al. · Aug 2, 2025

  8. A fake error message relayed between agents got three multi-agent frameworks to run an attacker's code

    Paper · · arXiv · Harold Triedman et al. · Mar 15, 2025

Related terms

What appears alongside CrewAI in the same briefings.

AutoGenLangChainLangGraphLlamaIndexOpen Policy AgentOpenAI Agents SDKAutoGPTAgentic JWT