langchain-kredo 0.2.0

LangChain integration for the Kredo agent attestation protocol

langchain-kredo

LangChain integration for the Kredo agent attestation protocol.

One line of code. Signed attestation. Done.

Install

pip install langchain-kredo

One-Liner

from langchain_kredo import attest

# That's it. Resolves name, looks up skill, signs, submits.
attest("jim", "incident-triage", "Triaged 3 incidents correctly in SOC exercise")

# With a URL — auto-detected as evidence artifact
attest("jim", "code-review", "https://github.com/org/repo/pull/47")

# With explicit proficiency (1-5, default 3)
attest("jim", "threat-hunting", "Found lateral movement in 4 minutes", proficiency=5)

Set KREDO_PRIVATE_KEY env var (hex seed) and go. Subject resolved by name or pubkey. Skill resolved by reverse taxonomy lookup — just say "incident-triage", it finds the domain.

Trust Gate

Policy enforcement for agent pipelines:

from langchain_kredo import KredoSigningClient, KredoTrustGate

client = KredoSigningClient(signing_key="your-hex-seed")
gate = KredoTrustGate(client, min_score=0.3, block_warned=True)

# Check trust
result = gate.check("ed25519:agent-pubkey")
# result.passed, result.score, result.skills, result.attestor_count

# Select best agent for a task (ranks by reputation + diversity + domain proficiency)
best = gate.select_best(candidates, domain="security-operations", skill="incident-triage")

# Build-vs-buy: delegate or self-compute?
delegate = gate.should_delegate(candidates, domain="code-generation", self_proficiency=2)

# Decorator
@gate.require(min_score=0.7)
def sensitive_operation(pubkey: str):
    ...

LangChain Tools

Four tools for agent toolboxes:

from langchain_kredo import KredoCheckTrustTool, KredoSearchAttestationsTool

tools = [
    KredoCheckTrustTool(client=client),
    KredoSearchAttestationsTool(client=client),
]

Tool	Name	Purpose
KredoCheckTrustTool	kredo_check_trust	Check agent reputation + skills + warnings
KredoSearchAttestationsTool	kredo_search_attestations	Find agents by skill/domain/proficiency
KredoSubmitAttestationTool	kredo_submit_attestation	Sign and submit skill attestation
KredoGetTaxonomyTool	kredo_get_taxonomy	Browse valid domains/skills

Callback Handler

Tracks chain execution, builds attestation evidence automatically:

from langchain_kredo import KredoCallbackHandler

handler = KredoCallbackHandler()
chain.invoke(input, config={"callbacks": [handler]})

for record in handler.get_records():
    if record.success_rate >= 0.9:
        client.attest_skill(
            subject_pubkey="ed25519:...",
            domain="security-operations",
            skill="incident-triage",
            proficiency=3,
            context=record.build_evidence_context(),
            artifacts=record.build_artifacts(),
        )

Collects evidence but never auto-submits. You decide when and what to attest.

Client

Full signing-aware client for when you need more control:

client = KredoSigningClient(
    signing_key=sk,           # SigningKey, bytes, hex string, or env var
    name="my-agent",
    agent_type="agent",
)

# Read
profile = client.get_profile("ed25519:...")
my_profile = client.my_profile()  # your own profile
results = client.search(domain="security-operations")

# Write
client.register()
client.attest_skill(
    subject_pubkey="ed25519:...",
    domain="security-operations",
    skill="incident-triage",
    proficiency=4,
    context="Demonstrated expert-level triage in SOC exercise",
)

Development

cd langchain-kredo
pip install -e ".[dev]"
pytest tests/ -v  # 82 tests

License

MIT