The cross-harness, self-improving memory layer for AI agents. Join our community: https://discord.com/invite/zdNChS2fBu
Go agent-ai-cliagent-harnessesagent-securitybrowser-extensionciclaude-codecloudcodexcursordetection-engineeringjevknowledge-basememorymobile-device-managementobservabilitysecuritysecurity-information-and-event-managementsecurity-toolstelemetrytraces
# LiteLLM proxy with beacon-gateway attached as a guardrail on every call, observing.
#
# beacon-gateway -config deploy/beacon-gateway.yaml &
# litellm --config deploy/litellm.yaml
#
# pre_call sees the whole request (system, tools, history, tool results) before it reaches the
# provider; post_call sees the model's reply, including its tool calls.
model_list:
- model_name: "anthropic/*"
litellm_params:
model: "anthropic/*"
api_key: os.environ/ANTHROPIC_API_KEY
- model_name: "openai/*"
litellm_params:
model: "openai/*"
api_key: os.environ/OPENAI_API_KEY
guardrails:
- guardrail_name: beacon-gateway-request
litellm_params:
guardrail: generic_guardrail_api
mode: pre_call
api_base: http://127.0.0.1:4019
default_on: true
# Keep the model path up when beacon-gateway is down: the call goes through unrecorded.
unreachable_fallback: fail_open
# LiteLLM withholds header values ("[present]") unless allowlisted here.
extra_headers: [x-claude-code-session-id, session_id]
- guardrail_name: beacon-gateway-response
litellm_params:
guardrail: generic_guardrail_api
mode: post_call
api_base: http://127.0.0.1:4019
default_on: true
unreachable_fallback: fail_open
extra_headers: [x-claude-code-session-id, session_id]
# Streams pass through as they arrive and are recorded at their end.
streaming_buffer_until_moderated: false