Basic Usage#

The nemo-flow binary observes coding agents that do not expose every LLM call site directly. It combines agent-specific hook endpoints with a passthrough LLM gateway so NeMo Flow owns both the agent lifecycle and the model request lifecycle.

Use the gateway when you need one observability boundary for OpenAI Codex, Claude Code, Cursor, and Hermes without replacing each agent’s canonical hook payload.

Hook Endpoints#

Each hook endpoint accepts the agent’s native hook JSON directly. Do not wrap the payload in a shared gateway envelope.

POST /hooks/codex accepts Codex hook JSON and returns the Codex-compatible hook response object.
POST /hooks/claude-code accepts Claude Code hook JSON and returns Claude-compatible fields such as continue and permission decisions when the hook event supports them.
POST /hooks/cursor accepts Cursor hook JSON and returns Cursor-compatible fields such as continue, permission, user_message, and agent_message when the hook event supports them.
POST /hooks/hermes accepts Hermes shell hook JSON and returns the empty JSON object expected by Hermes hook commands.

The adapters preserve vendor fields such as session IDs, working directories, transcript paths, model names, tool payloads, shell payloads, MCP payloads, file payloads, user identity, and subagent metadata in NeMo Flow event metadata.

Gateway Routes#

Route all coding-agent LLM traffic through the gateway when full LLM lifecycle observability is required.

POST /v1/responses
POST /v1/chat/completions
POST /v1/messages
POST /v1/messages/count_tokens
GET /v1/models

The gateway forwards raw provider JSON without rewriting OpenAI or Anthropic payload schemas. It removes only hop-by-hop transport headers, forwards streaming responses as streams, and emits NeMo Flow LLM start and end events under the active session scope.

Transparent Run#

Use the agent shortcuts for no-install local observability. The wrapper starts a gateway on a dynamic 127.0.0.1 port, injects the resolved hook and gateway configuration into the launched coding agent, and stops the gateway when the agent exits.

nemo-flow codex
nemo-flow claude
nemo-flow cursor
nemo-flow hermes

Use nemo-flow run -- <command> when you want to launch an explicit command instead of the built-in shortcut:

nemo-flow run -- codex

If a launcher or wrapper hides the real agent name, set that wrapper as the configured command and pass --agent. The same pattern applies to Claude Code, Codex, Cursor, and Hermes:

[agents.codex]
command = "my-codex-wrapper"

nemo-flow run --agent codex

Hermes is different from the other transparent modes: run --agent hermes starts the gateway and exports the dynamic NEMO_FLOW_GATEWAY_URL, but Hermes shell hooks still need to be installed or otherwise approved in Hermes config.

Use --dry-run --print to inspect the generated hook config, gateway environment, gateway URL, and final command without launching the agent.

Shared Configuration#

Shared TOML config is optional. The gateway loads defaults, then system config, then project config, then user config. User config takes priority over system and project config. CLI flags and environment variables override file config.

Config file locations are:

/etc/nemo-flow/config.toml
.nemo-flow/config.toml
$XDG_CONFIG_HOME/nemo-flow/config.toml
~/.config/nemo-flow/config.toml

Example:

[upstream]
openai_base_url = "https://api.openai.com/v1"
anthropic_base_url = "https://api.anthropic.com"

[agents.claude]
command = "claude"

[agents.codex]
command = "codex"

[agents.cursor]
command = "cursor-agent"
patch_restore_hooks = true

[agents.hermes]
command = "hermes"

Observability exporters are configured in plugins.toml. Use nemo-flow plugins edit for the user file, nemo-flow plugins edit --project for .nemo-flow/plugins.toml, or write the plugin config directly:

version = 1

[[components]]
kind = "observability"
enabled = true

[components.config.atif]
enabled = true
output_directory = ".nemo-flow/atif"

[components.config.openinference]
enabled = true
endpoint = "http://127.0.0.1:4318/v1/traces"

Transparent runs always bind the managed gateway to 127.0.0.1:0. The selected port is discovered by the wrapper and exposed to hooks through NEMO_FLOW_GATEWAY_URL.

Common environment variables for direct gateway server use are:

NEMO_FLOW_GATEWAY_BIND
NEMO_FLOW_OPENAI_BASE_URL
NEMO_FLOW_ANTHROPIC_BASE_URL

Plugin configuration controls process-level Observability exporters. Per-session configuration controls structured metadata on the top-level agent begin event and the plugin configuration metadata associated with the session.

hook-forward can also pass per-session configuration through headers:

x-nemo-flow-config-profile
x-nemo-flow-session-metadata
x-nemo-flow-plugin-config
x-nemo-flow-gateway-mode

The accepted gateway mode values are hook-only, passthrough, and required. The gateway records this value as session metadata so downstream exporters and review tooling can distinguish hook-only traces from sessions where provider traffic was expected to pass through the gateway.

Runtime Mapping#

The gateway normalizes vendor hook payloads into private internal events before calling NeMo Flow APIs.

Agent start opens a top-level ScopeType::Agent scope on a dedicated ScopeStackHandle.
Subagent start opens a child ScopeType::Agent scope. Subagent stop closes that scope when it is still active.
Tool pre-use starts a NeMo Flow tool span. Tool post-use, denial, or failure closes it.
Prompt, response, agent-thought, and Hermes LLM hooks are retained as private correlation hints. They are not emitted as NeMo Flow events.
Compaction, notification, and unknown hook events become mark events under the active session scope.
Gateway requests emit NeMo Flow LLM start and end events under the active session scope. Before each LLM start, the gateway uses explicit subagent headers, pending hints, shared conversation/generation/request identifiers, and the previous correlated owner to choose the parent scope.
LLM responses that contain future tool-use suggestions are retained as private tool-call hints. The next matching tool hook can then inherit the subagent scope that owned the LLM response, even when the hook payload does not include a subagent id.

Gateway requests can provide explicit correlation identifiers with these headers:

x-nemo-flow-session-id
x-nemo-flow-subagent-id
x-nemo-flow-conversation-id
x-nemo-flow-generation-id
x-nemo-flow-request-id

When those headers are absent, the gateway also looks for conversation_id/conversationId/conversation.id, generation_id/generationId/generation.id, and request_id/requestId/request.id fields in the provider request body. Correlation hints expire after five minutes. If the gateway cannot select one unambiguous hint, it falls back to the previous LLM owner, then to the only active subagent, then to the top-level agent scope.

Every gateway LLM event includes llm_correlation_status metadata. Possible values are explicit, single_hint, matched_hint, sticky_last_owner, active_subagent, agent_fallback, and ambiguous_fallback. Matched hints can also add llm_correlation_source, llm_correlation_subagent_id, llm_correlation_conversation_id, llm_correlation_generation_id, llm_correlation_request_id, and llm_correlation_agent_type.

Generated hook bundles subscribe to the events needed for that mapping:

Agent	Correlation hint hooks	Scope, tool, and mark hooks
Claude Code	`UserPromptSubmit`, `AfterAgentResponse`, `AfterAgentThought`, `Stop`	`SessionStart`, `SessionEnd`, `SubagentStart`, `SubagentStop`, `PreToolUse`, `PostToolUse`, `PostToolUseFailure`, `Notification`, `PreCompact`
Codex	`UserPromptSubmit`, `AfterAgentResponse`, `AfterAgentThought`, `Stop`	`SessionStart`, `SessionEnd`, `SubagentStart`, `SubagentStop`, `PreToolUse`, `PostToolUse`, `PostToolUseFailure`, `Notification`, `PreCompact`
Cursor	`beforeSubmitPrompt`, `afterAgentResponse`, `afterAgentThought`	`sessionStart`, `sessionEnd`, `subagentStart`, `subagentStop`, `preToolUse`, `postToolUse`, `beforeShellExecution`, `afterShellExecution`, `beforeMCPExecution`, `afterMCPExecution`, `preCompact`, `stop`
Hermes	`pre_llm_call`, `post_llm_call`	`on_session_start`, `on_session_end`, `on_session_finalize`, `on_session_reset`, `subagent_start`, `subagent_stop`, `pre_tool_call`, `post_tool_call`

Cursor hook-only mode observes agent, subagent, and tool lifecycle. To observe Cursor LLM lifecycle completely, configure Cursor model traffic to use the gateway.

Hook Forwarding#

Hooks generated by the wrapper (Claude/Codex/Cursor ephemeral, Hermes via setup) invoke nemo-flow hook-forward <agent> from stdin. Inside the wrapper the gateway URL comes from NEMO_FLOW_GATEWAY_URL injected on every run; outside the wrapper (Hermes standalone, IDE-launched Claude/Codex) the hook command falls back to its embedded --gateway-url.

hook-forward reads the canonical hook payload from standard input, sends it to the matching endpoint, and prints the endpoint response. It fails open by default so observability outages do not block the coding agent. Add --fail-closed only when policy requires hook delivery to block the agent.

Optional flags map to gateway headers:

--session-metadata sets x-nemo-flow-session-metadata.
--plugin-config sets x-nemo-flow-plugin-config.
--profile sets x-nemo-flow-config-profile.
--gateway-mode sets x-nemo-flow-gateway-mode.

Agent Guides#

Use the per-agent guide for end-to-end setup, smoke tests, and GUI or application-mode caveats.

Each guide covers transparent run setup, gateway routing, hook smoke tests, Agent Trajectory Interchange Format (ATIF) export verification on session end, and troubleshooting missing LLM lifecycle data.