Skip to main content

Recomputable verification adapter for CrewAI — 6-dim verification, drift detection, and verified failover

Project description

correctover-crewai

Recomputable verification adapter for CrewAI agents.

What This Does

correctover-crewai plugs into CrewAI's tool-call hook system to provide:

  • 6-dimension verification — structure, schema, latency, cost, identity, integrity
  • Recomputable proof hashes — anyone can re-derive the verification result from the raw bytes, no trust required
  • Drift detection — catch when output quality degrades across providers or over time
  • Verified failover — automatic fallback to backup providers when verification fails

Quick Start

pip install correctover-crewai
from crewai import Agent, Crew, Task
from correctover_crewai import CorrectoverCrewAIAdapter, VerifierDepth

# Initialize with your Correctover engine
adapter = CorrectoverCrewAIAdapter(
    api_key="your-correctover-key",
    verifier_depth=VerifierDepth.D2,
    enable_failover=True,
)

# Register hooks globally
adapter.register()

# Use CrewAI normally — verification happens automatically
agent = Agent(role="Researcher", goal="Find accurate info")
task = Task(description="Research AI safety", agent=agent)
crew = Crew(agents=[agent], tasks=[task])
result = crew.kickoff()

# Inspect verdicts
for v in adapter.get_verdicts():
    print(f"{v.tool_name}: {v.verdict} (confidence={v.confidence:.2f})")

Recomputable vs Tamper-Evident

Feature correctover-crewai asqav-crewai
Verification Recomputable from bytes Tamper-evident signatures
Trust model No trust needed Trust the signer
Drift detection Built-in Not available
Failover Automatic Not available
6-dim verification Structure, Schema, Latency, Cost, Identity, Integrity Audit logging only

Verifier Depth Levels

  • D0 (audit_only): Record events, no intervention
  • D1 (bounded_repair): Low-risk automatic retries
  • D2 (structured_replan): Replan with structured output constraints
  • D3 (public_conformance): Generate compliance reports

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

correctover_crewai-0.1.0.tar.gz (13.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

correctover_crewai-0.1.0-py3-none-any.whl (13.4 kB view details)

Uploaded Python 3

File details

Details for the file correctover_crewai-0.1.0.tar.gz.

File metadata

  • Download URL: correctover_crewai-0.1.0.tar.gz
  • Upload date:
  • Size: 13.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.12

File hashes

Hashes for correctover_crewai-0.1.0.tar.gz
Algorithm Hash digest
SHA256 e2e476fb6cb646373147d960f2cd4c5d19f21e736acc8288b2b684be71c6275f
MD5 d4f2ede64ed294c869535a43d4cd05c1
BLAKE2b-256 db381879e779eb3e0a530ebd243527b208294232ee490fdbeeef3680d9cd28c7

See more details on using hashes here.

File details

Details for the file correctover_crewai-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for correctover_crewai-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 569eb32bf8162c7bf53031d7391bf535cc5abed46cfaafd4b482e87d05a79176
MD5 31a03e380d5d3fb22fb46ddfbc3c28b7
BLAKE2b-256 a7d1dd62f41efea9d2cdf42296e1a23d7abeafb0108b179204a35370f0c9e434

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page