openlatch-client 0.6.22

//! The action envelope: one normalized description of what an event records,
//! carried as `data.action` beside everything the event already says.
//!
//! A **side output**. It is built from values the envelope already carries —
//! the common keys [`super::normalize::to_common`] produced, plus `source`,
//! `id` and `time` — and it changes none of them. The shape is
//! `schemas/action-envelope.schema.json`, and the value is built through its
//! generated types, so it cannot drift from the published contract.
//!
//! Read by SHAPE, like the rest of the egress mapping. The one table keyed on
//! `source` is [`HARNESS_PROVIDERS`], which labels a maker and decides nothing;
//! the common tool vocabulary is keyed on the agent's own tool name, so a fork
//! that keeps its parent's tool names is labelled with no code change.
//!
//! Shell commands are read through the zone evaluator's existing tokenizer
//! ([`shell::classify_command`]), never a parser of our own: the action and
//! the policy see the same simple commands.

use std::sync::atomic::{AtomicBool, AtomicU64, Ordering};

use serde_json::Value;

use super::normalize;
use super::{ActionCommand, ActionEnvelope, ActionLlmParse, ActionMcpCall, ActionToken};
use crate::zone_eval::effect::path::PathEnv;
use crate::zone_eval::effect::shell;
use crate::zone_eval::facts::FactSet;
use crate::zone_eval::types::SimpleCommand;

/// The most a serialized action may weigh. Over it, the command tokens go
/// first, then the commands themselves; an action still over it is not sent.
pub const MAX_ACTION_BYTES: usize = 16 * 1024;

/// Who makes each agent, by its wire `source`. A label, never a decision: an
/// unlisted source is `null`.
const HARNESS_PROVIDERS: &[(&str, &str)] = &[
    ("claude-code", "anthropic"),
    ("codex-cli", "openai"),
    ("cursor", "cursor"),
    ("cline", "cline"),
    ("opencode", "opencode"),
];

/// The common tool vocabulary, keyed on the tool name the AGENT used. A name
/// not listed travels lowercased.
const TOOL_NAMES: &[(&str, &str)] = &[
    ("Bash", "bash"),
    ("Shell", "bash"),
    ("execute_command", "bash"),
    ("run_commands", "bash"),
    ("bash", "bash"),
    ("shell", "bash"),
    ("Read", "read"),
    ("read_file", "read"),
    ("read_files", "read"),
    ("read", "read"),
    ("Edit", "edit"),
    ("replace_in_file", "edit"),
    ("editor", "edit"),
    ("edit", "edit"),
    ("apply_patch", "edit"),
    ("Write", "write"),
    ("write_to_file", "write"),
    ("write", "write"),
    ("Grep", "search"),
    ("Glob", "search"),
    ("search_files", "search"),
    ("search_codebase", "search"),
    ("grep", "search"),
    ("glob", "search"),
    ("WebFetch", "web"),
    ("WebSearch", "web"),
    ("web_fetch", "web"),
    ("fetch_web_content", "web"),
    ("webfetch", "web"),
    ("websearch", "web"),
    ("Agent", "agent"),
    ("Task", "agent"),
    ("spawn_agent", "agent"),
    ("new_task", "agent"),
    ("task", "agent"),
];

/// The pointer every MCP call's arguments are read at.
const TOOL_INPUT_POINTER: &str = "/data/tool_input";

/// The `[normalization] action_envelope` gate, for the model relay, which holds
/// no daemon configuration of its own. Set once at daemon start from the same
/// value the hook path reads off its `AppState`.
static ENABLED: AtomicBool = AtomicBool::new(false);

/// Actions that could not be built — a panic caught around [`to_action`]. The
/// event left without one, exactly as it would with the gate off. Surfaced as
/// `action_envelope_errors` on `GET /metrics`.
static ERRORS: AtomicU64 = AtomicU64::new(0);

/// Set the model relay's copy of the gate.
pub fn set_enabled(on: bool) {
    ENABLED.store(on, Ordering::Relaxed);
}

/// Whether the model relay adds an action to its records.
pub fn enabled() -> bool {
    ENABLED.load(Ordering::Relaxed)
}

/// Record one action that could not be built.
pub fn note_error() {
    ERRORS.fetch_add(1, Ordering::Relaxed);
}

/// How many actions could not be built so far.
pub fn errors() -> u64 {
    ERRORS.load(Ordering::Relaxed)
}

/// The action a hook envelope records, once its `data` is in the common shape.
///
/// `None` when the event names no tool (a session start, a prompt, a stop) or
/// lacks a core attribute — the envelope then travels exactly as without the
/// gate. A shell command the tokenizer cannot read still yields an action, with
/// `payload_llm_cmd: null`.
pub fn to_action(envelope: &Value) -> Option<Value> {
    let data = envelope.get("data")?;
    let common_tool = text(data, "tool_name")?;
    let raw = data.get("raw_agent_payload");
    // The agent's own call: its tool name and arguments as it sent them. A
    // payload already in the common shape carries no original, and is it.
    let native_call = normalize::tool_call_of(raw.unwrap_or(data), None);
    let native = native_call
        .as_ref()
        .map_or(common_tool, |call| call.native.as_str());

    let mcp = mcp_call(native, common_tool, data);
    let tool_name = match &mcp {
        Some(call) => format!("{}.{}", call.server, call.tool),
        None => TOOL_NAMES
            .iter()
            .find(|(name, _)| *name == native)
            .map_or_else(|| native.to_lowercase(), |(_, common)| common.to_string()),
    };
    let payload_llm_cmd = (tool_name == "bash")
        .then(|| {
            let raw_input = native_call.as_ref().and_then(|call| call.input.as_ref());
            commands(raw_input, data.get("tool_input"))
        })
        .flatten();
    let payload_raw = ["raw_agent_payload", "tool_input"]
        .into_iter()
        .find(|key| data.get(key).is_some())
        .map(|key| format!("/data/{key}"));
    let source = envelope.get("source")?.as_str()?;

    let action = ActionEnvelope {
        harness_provider: harness_provider(source),
        harness_name: source.to_string(),
        session_id: text(data, "session_id").map(str::to_string),
        cwd: text(data, "cwd").map(str::to_string),
        timestamp: unix_ms(envelope.get("time")?.as_str()?)?,
        // The agent's own id only: a derived `ol_` id pairs a pre with its post
        // and identifies nothing, so it never names an action.
        action_id: normalize::tool_use_id_of(raw.unwrap_or(data))
            .or_else(|| envelope.get("id")?.as_str())?
            .to_string(),
        action_type: if mcp.is_some() { "mcp" } else { "tool" }.to_string(),
        tool_name: Some(tool_name),
        payload_raw,
        payload_llm_parse: None,
        payload_llm_mcp: mcp,
        payload_llm_cmd,
    };
    capped(action)
}

/// The action of one model relay record. `source` is the record's resolved
/// source, `unknown` included; `occurred_at` its RFC 3339 request start.
pub fn llm_action(
    source: &str,
    session_id: Option<&str>,
    occurred_at: &str,
    event_id: &str,
    parse: ActionLlmParse,
) -> Value {
    let action = ActionEnvelope {
        harness_provider: harness_provider(source),
        harness_name: source.to_string(),
        session_id: session_id.map(str::to_string),
        cwd: None,
        timestamp: unix_ms(occurred_at).unwrap_or_else(|| chrono::Utc::now().timestamp_millis()),
        action_id: event_id.to_string(),
        action_type: "llm".to_string(),
        tool_name: None,
        payload_raw: None,
        payload_llm_parse: Some(parse),
        payload_llm_mcp: None,
        payload_llm_cmd: None,
    };
    serde_json::to_value(action).unwrap_or_default()
}

fn harness_provider(source: &str) -> Option<String> {
    HARNESS_PROVIDERS
        .iter()
        .find(|(name, _)| *name == source)
        .map(|(_, provider)| provider.to_string())
}

fn text<'a>(data: &'a Value, key: &str) -> Option<&'a str> {
    data.get(key)?.as_str().filter(|s| !s.is_empty())
}

fn unix_ms(time: &str) -> Option<i64> {
    chrono::DateTime::parse_from_rfc3339(time)
        .ok()
        .map(|t| t.timestamp_millis())
}

/// The server and tool of an MCP call, from whichever MCP shape names them:
/// the common `mcp__<server>__<tool>` the mapping produced (which also covers a
/// call whose native name carries no MCP shape, like Cursor's
/// `beforeMCPExecution`), then the native `mcp__…`, `MCP:<tool>` and
/// `use_mcp_tool` spellings for a call the mapping could not name.
fn mcp_call(native: &str, common: &str, data: &Value) -> Option<ActionMcpCall> {
    let call = |server: &str, tool: &str| ActionMcpCall {
        server: server.to_string(),
        tool: tool.to_string(),
        arguments: TOOL_INPUT_POINTER.to_string(),
    };
    for name in [common, native] {
        if let Some((server, tool)) = name.strip_prefix("mcp__").and_then(|n| n.split_once("__")) {
            return Some(call(server, tool));
        }
    }
    if let Some(tool) = native.strip_prefix("MCP:") {
        return Some(call("unknown", tool));
    }
    if native == "use_mcp_tool" {
        let input = data.get("tool_input").unwrap_or(&Value::Null);
        let arg = |key| text(input, key).unwrap_or("unknown");
        return Some(call(arg("server_name"), arg("tool_name")));
    }
    None
}

/// The simple commands of a shell call, or `None` when there is no command or
/// the tokenizer could not read one of them.
///
/// A list of commands (Cline's `run_commands`) is read entry by entry from the
/// agent's own arguments — the common `command` joins them with newlines —
/// and the entries are concatenated.
fn commands(raw_input: Option<&Value>, common_input: Option<&Value>) -> Option<Vec<ActionCommand>> {
    let lines: Vec<&str> = match raw_input
        .and_then(|input| input.get("commands"))
        .and_then(Value::as_array)
    {
        Some(list) => list
            .iter()
            .filter_map(|entry| entry.as_str().or_else(|| entry.get("command")?.as_str()))
            .collect(),
        None => vec![common_input?.get("command")?.as_str()?],
    };
    let mut simple = Vec::new();
    for line in lines {
        let cls = shell::classify_command(line, &FactSet::default(), &PathEnv::default());
        if cls.simple.is_empty() {
            return None;
        }
        simple.extend(fold_wrappers(&cls.simple).into_iter().map(command_of));
    }
    if simple.is_empty() {
        return None;
    }
    let chained = simple.len() > 1;
    for command in &mut simple {
        command.is_chained = chained;
    }
    Some(simple)
}

/// The classifier lists every layer a command runs under — a wrapper (`env`,
/// `timeout`, …), an env runner (`npx`, …), an escalator (`sudo`, …) — as a
/// simple command of its own, outermost first, ahead of the command it ran.
/// Those are one command here: each run of layers is grouped with its innermost
/// command, whose tokens carry the layers' words.
///
/// A layer is a program from the classifier's own tables whose words end with
/// the next entry's — so `sudo -v && ls` stays two commands.
fn fold_wrappers(simple: &[SimpleCommand]) -> Vec<&[SimpleCommand]> {
    let wraps = |outer: &SimpleCommand, inner: &SimpleCommand| {
        let n = inner.raw_argv.len();
        let program = outer.program.as_str();
        (program == "docker"
            || [
                shell::WRAPPERS,
                shell::ENV_RUNNERS,
                shell::RECURSING_ESCALATORS,
            ]
            .iter()
            .any(|table| table.contains(&program)))
            && outer.raw_argv.len() > n
            && outer.raw_argv.ends_with(&inner.raw_argv)
            && shell::normalise(&shell::unquote(
                &outer.raw_argv[outer.raw_argv.len() - n - 1],
            )) == inner.program
    };
    let mut runs = Vec::new();
    let mut i = 0;
    while i < simple.len() {
        let mut j = i;
        while j + 1 < simple.len() && wraps(&simple[j], &simple[j + 1]) {
            j += 1;
        }
        runs.push(&simple[i..=j]);
        i = j + 1;
    }
    runs
}

/// One simple command — the last of `run`, under the layers before it — its
/// words typed.
///
/// `program`, `argv` and `redirects` are the classifier's own. The layer and
/// assignment words it stripped are not kept on any field, so they are
/// recovered from the chain part's text: the longest `raw` of the run (the
/// classifier keeps the whole part on the innermost command of a wrapper and on
/// the escalator itself), re-split by the same tokenizer, with the redirections
/// taken out. What precedes the command's own words is a layer word or, when it
/// reads `NAME=value`, an assignment.
fn command_of(run: &[SimpleCommand]) -> ActionCommand {
    let token = |t: &str, v: &str, target: Option<&str>| ActionToken {
        t: t.to_string(),
        v: v.to_string(),
        target: target.map(str::to_string),
    };
    let sc = &run[run.len() - 1];
    let part = run
        .iter()
        .map(|c| c.raw.as_str())
        .max_by_key(|raw| raw.len());
    let mut words = shell::tokenise(part.unwrap_or_default()).unwrap_or_default();
    let redirects = run.iter().find(|c| !c.redirects.is_empty()).unwrap_or(sc);
    for (op, target) in &redirects.raw_redirects {
        if let Some(at) = words
            .windows(2)
            .position(|pair| pair[0] == *op && pair[1] == *target)
        {
            words.drain(at..at + 2);
        }
    }
    let own = 1 + sc.raw_argv.len();
    let prefix = &words[..words.len().saturating_sub(own)];

    let mut tokens: Vec<ActionToken> = prefix
        .iter()
        .map(|word| {
            let kind = if is_assignment(word) {
                "assignment"
            } else {
                "wrapper"
            };
            token(kind, &shell::unquote(word), None)
        })
        .collect();
    if !sc.program.is_empty() {
        tokens.push(token("program", &sc.program, None));
    }
    tokens.extend(sc.argv.iter().map(|arg| {
        let kind = if arg.starts_with('-') { "flag" } else { "arg" };
        token(kind, arg, None)
    }));
    tokens.extend(
        redirects
            .redirects
            .iter()
            .map(|(op, target)| token("redirect", op, Some(target))),
    );
    ActionCommand {
        cmd: sc.program.clone(),
        argv: sc.argv.clone(),
        is_chained: false,
        tokens,
    }
}

/// `NAME=value` with a shell-variable name — the prefix the classifier strips.
fn is_assignment(word: &str) -> bool {
    word.split_once('=').is_some_and(|(name, _)| {
        !name.is_empty() && name.chars().all(|c| c.is_alphanumeric() || c == '_')
    })
}

/// The action as a value, within [`MAX_ACTION_BYTES`]: the command tokens are
/// dropped first, then the commands. `None` when even that is too large.
fn capped(mut action: ActionEnvelope) -> Option<Value> {
    let fits = |action: &ActionEnvelope| {
        serde_json::to_vec(action).is_ok_and(|bytes| bytes.len() <= MAX_ACTION_BYTES)
    };
    if !fits(&action) {
        for command in action.payload_llm_cmd.iter_mut().flatten() {
            command.tokens.clear();
        }
    }
    if !fits(&action) {
        action.payload_llm_cmd = None;
    }
    fits(&action).then(|| serde_json::to_value(&action).ok())?
}

#[cfg(test)]
mod tests {
    use serde_json::{json, Value};

    use super::*;
    use crate::hooks::bindings::{aliases_for, binding_id};

    const TIME: &str = "2026-10-14T09:12:15.123Z";
    const TIME_MS: i64 = 1_791_969_135_123;
    const EVENT_ID: &str = "evt_019e2a7c-3f10-7c4e-9a51-6b0f2d8e4a17";

    /// One hook event as the daemon holds it after `map_data_to_common`: the
    /// agent's payload mapped through its binding's alias table, under its
    /// hook lane's binding.
    fn hook_envelope(source: &str, event_type: &str, native: Value) -> Value {
        let binding = binding_id(source, false);
        let data =
            super::normalize::to_common(&native, aliases_for(source), &binding).unwrap_or(native);
        json!({
            "specversion": "1.0",
            "id": EVENT_ID,
            "source": source,
            "type": event_type,
            "time": TIME,
            "datacontenttype": "application/json",
            "data": data,
        })
    }

    fn action_of(source: &str, native: Value) -> Value {
        let action = to_action(&hook_envelope(source, "pre_tool_use", native)).expect("an action");
        assert_valid(&action);
        action
    }

    /// Every action is checked against the published schema.
    fn assert_valid(action: &Value) {
        let schema: Value =
            serde_json::from_str(include_str!("../../../schemas/action-envelope.schema.json"))
                .expect("schema is JSON");
        let validator = jsonschema::validator_for(&schema).expect("schema compiles");
        let errors: Vec<String> = validator
            .iter_errors(action)
            .map(|e| e.to_string())
            .collect();
        assert!(errors.is_empty(), "{action}: {errors:?}");
    }

    fn claude(tool: &str, input: Value) -> Value {
        json!({
            "session_id": "8f2c1e4a-6b1d-4c7e-9f3a-2d5b7c9e1f04",
            "transcript_path": "/Users/dev/.claude/projects/-repo/8f2c1e4a.jsonl",
            "cwd": "/repo",
            "permission_mode": "default",
            "hook_event_name": "PreToolUse",
            "tool_name": tool,
            "tool_input": input,
            "tool_use_id": "toolu_01HbQ4tWm8XJk2vF9pL3sR7n",
        })
    }

    fn codex(tool: &str, input: Value) -> Value {
        json!({
            "session_id": "s-41",
            "transcript_path": "/home/dev/.codex/sessions/2026/10/14/rollout-s-41.jsonl",
            "cwd": "/repo",
            "hook_event_name": "PreToolUse",
            "model": "gpt-5.2-codex",
            "turn_id": "turn-7",
            "tool_name": tool,
            "tool_input": input,
            "tool_use_id": "call_7Q2",
        })
    }

    fn cline_classic(tool: &str, parameters: Value) -> Value {
        json!({
            "clineVersion": "3.38.1",
            "hookName": "PreToolUse",
            "timestamp": "1791969135123",
            "taskId": "1791969100000",
            "workspaceRoots": ["/Users/dev/work/app"],
            "userId": "u-1",
            "preToolUse": {"toolName": tool, "parameters": parameters},
        })
    }

    fn cmd(command: &str) -> Value {
        action_of("claude-code", claude("Bash", json!({"command": command})))["payload_llm_cmd"]
            .clone()
    }

    fn programs(cmds: &Value) -> Vec<&str> {
        cmds.as_array()
            .expect("commands")
            .iter()
            .map(|c| c["cmd"].as_str().unwrap())
            .collect()
    }

    #[test]
    fn codex_bash_is_the_spec_example() {
        let action = action_of(
            "codex-cli",
            codex(
                "Bash",
                json!({"command": "npm test && git push origin develop"}),
            ),
        );
        assert_eq!(
            action,
            json!({
                "harness_provider": "openai",
                "harness_name": "codex-cli",
                "session_id": "s-41",
                "cwd": "/repo",
                "timestamp": TIME_MS,
                "action_id": "call_7Q2",
                "action_type": "tool",
                "tool_name": "bash",
                "payload_raw": "/data/tool_input",
                "payload_llm_parse": null,
                "payload_llm_mcp": null,
                "payload_llm_cmd": [
                    {"cmd": "npm", "argv": ["test"], "is_chained": true,
                     "tokens": [{"t": "program", "v": "npm"}, {"t": "arg", "v": "test"}]},
                    {"cmd": "git", "argv": ["push", "origin", "develop"], "is_chained": true,
                     "tokens": [{"t": "program", "v": "git"}, {"t": "arg", "v": "push"},
                                {"t": "arg", "v": "origin"}, {"t": "arg", "v": "develop"}]}
                ]
            })
        );
    }

    #[test]
    fn codex_apply_patch_is_an_edit_keyed_on_its_native_name() {
        let action = action_of(
            "codex-cli",
            codex(
                "apply_patch",
                json!({"input": "*** Begin Patch\n*** Update File: src/lib.rs\n@@\n-old\n+new\n*** End Patch\n"}),
            ),
        );
        // Unmapped in Codex's table, so the common data names it
        // `codex_cli_hook:apply_patch` and carries the original.
        assert_eq!(action["tool_name"], "edit");
        assert_eq!(action["action_type"], "tool");
        assert_eq!(action["payload_raw"], "/data/raw_agent_payload");
        assert_eq!(action["action_id"], "call_7Q2");
        assert!(action["payload_llm_cmd"].is_null());
    }

    #[test]
    fn claude_code_bash_mcp_and_read() {
        let bash = action_of("claude-code", claude("Bash", json!({"command": "ls -la"})));
        assert_eq!(bash["harness_provider"], "anthropic");
        assert_eq!(bash["action_id"], "toolu_01HbQ4tWm8XJk2vF9pL3sR7n");
        assert_eq!(bash["payload_raw"], "/data/tool_input");
        assert_eq!(
            bash["payload_llm_cmd"][0]["tokens"][1],
            json!({"t": "flag", "v": "-la"})
        );

        let mcp = action_of(
            "claude-code",
            claude(
                "mcp__github__create_issue",
                json!({"owner": "acme", "title": "x"}),
            ),
        );
        assert_eq!(mcp["action_type"], "mcp");
        assert_eq!(mcp["tool_name"], "github.create_issue");
        assert_eq!(
            mcp["payload_llm_mcp"],
            json!({"server": "github", "tool": "create_issue", "arguments": "/data/tool_input"})
        );
        assert!(mcp["payload_llm_cmd"].is_null());

        let read = action_of(
            "claude-code",
            claude("Read", json!({"file_path": "/repo/a.rs"})),
        );
        assert_eq!(
            (read["action_type"].as_str(), read["tool_name"].as_str()),
            (Some("tool"), Some("read"))
        );
        assert!(read["payload_llm_mcp"].is_null() && read["payload_llm_cmd"].is_null());

        // A tool no table lists travels lowercased.
        let other = action_of(
            "claude-code",
            claude("KillShell", json!({"shell_id": "b1"})),
        );
        assert_eq!(other["tool_name"], "killshell");
    }

    #[test]
    fn cursor_before_shell_has_no_call_id_and_uses_the_envelope_id() {
        let native: Value = serde_json::from_str(include_str!(
            "../../../tests/fixtures/cursor/before-shell-cli.json"
        ))
        .unwrap();
        let action = action_of("cursor", native);
        assert_eq!(action["action_id"], EVENT_ID);
        assert_eq!(action["harness_provider"], "cursor");
        assert_eq!(action["tool_name"], "bash");
        // `cwd` is empty, so the first workspace root.
        assert_eq!(action["cwd"], "/Users/dev/work");
        assert_eq!(action["payload_raw"], "/data/raw_agent_payload");
        assert_eq!(programs(&action["payload_llm_cmd"]), ["rm"]);
    }

    #[test]
    fn cursor_pre_tool_use_shell_and_mcp() {
        let native: Value = serde_json::from_str(include_str!(
            "../../../tests/fixtures/cursor/pre-tool-shell-ide.json"
        ))
        .unwrap();
        let shell = action_of("cursor", native);
        assert_eq!(shell["action_id"], "454c7fcf-fba9-48c8-9877-e1d8b11a351d");
        assert_eq!(shell["tool_name"], "bash");
        assert!(shell["cwd"].is_null(), "an empty cwd and no workspace root");

        let before_mcp = json!({
            "conversation_id": "c1", "generation_id": "g1", "model": "default",
            "tool_name": "create_issue", "tool_input": "{\"title\":\"x\"}",
            "mcp_server_name": "github", "url": "https://api.example.com/mcp",
            "session_id": "c1", "hook_event_name": "beforeMCPExecution",
            "cursor_version": "3.22.12", "workspace_roots": ["/Users/dev/work"]
        });
        let mcp = action_of("cursor", before_mcp);
        assert_eq!(mcp["action_type"], "mcp");
        assert_eq!(mcp["tool_name"], "github.create_issue");

        let pre_mcp = json!({
            "conversation_id": "c1", "tool_name": "MCP:create_issue",
            "tool_input": {"title": "x"}, "tool_use_id": "tu-9", "session_id": "c1",
            "hook_event_name": "preToolUse", "cursor_version": "3.22.12", "workspace_roots": []
        });
        let unnamed = action_of("cursor", pre_mcp);
        assert_eq!(unnamed["tool_name"], "unknown.create_issue");
        assert_eq!(unnamed["action_id"], "tu-9");
    }

    #[test]
    fn cline_classic_sdk_list_and_mcp() {
        let classic = action_of(
            "cline",
            cline_classic(
                "execute_command",
                json!({"command": "npm run build", "requires_approval": "false"}),
            ),
        );
        assert_eq!(classic["harness_provider"], "cline");
        assert_eq!(classic["session_id"], "1791969100000");
        assert_eq!(classic["cwd"], "/Users/dev/work/app");
        assert_eq!(classic["tool_name"], "bash");
        // The classic host sends no call id; the derived `ol_` one is never used.
        assert_eq!(classic["action_id"], EVENT_ID);
        assert_eq!(programs(&classic["payload_llm_cmd"]), ["npm"]);

        let sdk = action_of(
            "cline",
            json!({
                "taskId": "conv_1",
                "hookName": "tool_call",
                "workspaceRoots": ["/Users/dev/work/app"],
                "tool_call": {"id": "call_9", "name": "run_commands",
                              "input": {"commands": ["npm ci", "npm test -- --watch=false"]}},
                "preToolUse": {"toolName": "run_commands",
                               "parameters": {"commands": "[\"npm ci\",\"npm test -- --watch=false\"]"}}
            }),
        );
        assert_eq!(sdk["action_id"], "call_9");
        let cmds = &sdk["payload_llm_cmd"];
        assert_eq!(programs(cmds), ["npm", "npm"]);
        assert_eq!(cmds[0]["is_chained"], true);
        assert_eq!(cmds[1]["argv"], json!(["test", "--", "--watch=false"]));

        let mcp = action_of(
            "cline",
            cline_classic(
                "use_mcp_tool",
                json!({"server_name": "github", "tool_name": "create_issue",
                       "arguments": "{\"title\":\"x\"}"}),
            ),
        );
        assert_eq!(mcp["action_type"], "mcp");
        assert_eq!(mcp["tool_name"], "github.create_issue");
        assert_eq!(mcp["payload_llm_mcp"]["arguments"], "/data/tool_input");
    }

    #[test]
    fn shell_commands_are_split_and_typed() {
        let multi = cmd("cd /repo\nnpm ci\nnpm test");
        assert_eq!(programs(&multi), ["cd", "npm", "npm"]);
        assert!(multi
            .as_array()
            .unwrap()
            .iter()
            .all(|c| c["is_chained"] == true));

        let chain = cmd("cargo build && cargo test");
        assert_eq!(programs(&chain), ["cargo", "cargo"]);

        let redirect = cmd("echo hi > out.txt");
        assert_eq!(
            redirect[0]["tokens"],
            json!([{"t": "program", "v": "echo"}, {"t": "arg", "v": "hi"},
                   {"t": "redirect", "v": ">", "target": "out.txt"}])
        );
        assert_eq!(redirect[0]["argv"], json!(["hi"]));

        // A wrapper is one command with the wrapper's words in front.
        let sudo = cmd("sudo -u root rm -rf /tmp/x");
        assert_eq!(programs(&sudo), ["rm"]);
        assert_eq!(sudo[0]["is_chained"], false);
        assert_eq!(
            sudo[0]["tokens"],
            json!([{"t": "wrapper", "v": "sudo"}, {"t": "wrapper", "v": "-u"},
                   {"t": "wrapper", "v": "root"}, {"t": "program", "v": "rm"},
                   {"t": "flag", "v": "-rf"}, {"t": "arg", "v": "/tmp/x"}])
        );
        // `sudo -v && ls` is two commands, not a wrapper.
        assert_eq!(programs(&cmd("sudo -v && ls")), ["sudo", "ls"]);

        let assigned = cmd("FOO=1 make build");
        assert_eq!(programs(&assigned), ["make"]);
        assert_eq!(
            assigned[0]["tokens"][0],
            json!({"t": "assignment", "v": "FOO=1"})
        );

        // The tokenizer refuses: no commands, the action still travels.
        let broken = action_of(
            "claude-code",
            claude("Bash", json!({"command": "echo 'unbalanced"})),
        );
        assert!(broken["payload_llm_cmd"].is_null());
        assert_eq!(broken["tool_name"], "bash");
    }

    #[test]
    fn only_tool_events_get_an_action() {
        let prompt = hook_envelope(
            "claude-code",
            "user_prompt_submit",
            json!({"session_id": "s", "hook_event_name": "UserPromptSubmit", "prompt": "hi"}),
        );
        assert_eq!(to_action(&prompt), None);
    }

    #[test]
    fn an_oversized_action_drops_tokens_then_commands() {
        let many = vec!["echo aaaaaaaaaaaaaaaaaaaaaaaaaaaa"; 200].join(" && ");
        let action = cmd(&many);
        assert!(action
            .as_array()
            .unwrap()
            .iter()
            .all(|c| c.get("tokens").is_none()));
        // Within the tokenizer's 10 000 characters, over the cap without tokens.
        let huge = vec!["x"; 4_000].join(";");
        assert!(cmd(&huge).is_null());
    }

    #[test]
    fn an_llm_action_is_valid() {
        let action = llm_action(
            "unknown",
            None,
            TIME,
            "0190aaaa-bbbb-cccc-dddd-eeeeeeeeeeee",
            ActionLlmParse {
                provider: Some("anthropic".into()),
                model: Some("claude-opus-4-8".into()),
                input_tokens: Some(1),
                output_tokens: Some(2),
                cache_read_tokens: Some(3),
                cache_write_tokens: Some(4),
            },
        );
        assert_valid(&action);
        assert_eq!(action["harness_provider"], Value::Null);
        assert_eq!(action["timestamp"], TIME_MS);
    }

    /// Writes realistic input-envelope → action pairs for every agent and action
    /// type to the research workspace. Run it on purpose:
    /// `cargo test --lib action_examples_dump -- --ignored --nocapture`.
    #[test]
    #[ignore = "writes a file outside the repository"]
    fn action_examples_dump() {
        use crate::model_relay::capture::{CostBasis, Usage};
        use crate::model_relay::emit::{assemble_data, assemble_event, relay_action, Observation};
        use crate::model_relay::session::{Assurance, Resolved};
        use crate::model_relay::wire_format::WireFormat;

        let mut examples = Vec::new();
        let mut hook = |agent: &str, title: &str, source: &str, native: Value| {
            let envelope = hook_envelope(source, "pre_tool_use", native);
            let action = to_action(&envelope).expect("an action");
            assert_valid(&action);
            examples.push(json!({
                "agent": agent, "action_type": action["action_type"], "title": title,
                "input_envelope": envelope, "action": action,
            }));
        };
        hook(
            "claude-code",
            "Bash: chained npm test && git push",
            "claude-code",
            claude(
                "Bash",
                json!({"command": "npm test && git push origin develop", "description": "Run tests then push"}),
            ),
        );
        hook(
            "claude-code",
            "MCP: mcp__github__create_issue",
            "claude-code",
            claude(
                "mcp__github__create_issue",
                json!({"owner": "acme", "repo": "app", "title": "Flaky test"}),
            ),
        );
        hook(
            "claude-code",
            "Read",
            "claude-code",
            claude("Read", json!({"file_path": "/repo/src/main.rs"})),
        );
        hook(
            "claude-code",
            "Bash: multi-line script",
            "claude-code",
            claude("Bash", json!({"command": "cd /repo\nnpm ci\nnpm test"})),
        );
        hook(
            "claude-code",
            "Bash: redirect to file",
            "claude-code",
            claude(
                "Bash",
                json!({"command": "git log --oneline -20 > /tmp/log.txt 2>> /tmp/err.log"}),
            ),
        );
        hook(
            "claude-code",
            "Bash: sudo wrapper",
            "claude-code",
            claude(
                "Bash",
                json!({"command": "sudo -u root rm -rf /var/cache/app"}),
            ),
        );
        hook(
            "claude-code",
            "Bash: FOO=1 assignment prefix",
            "claude-code",
            claude(
                "Bash",
                json!({"command": "RUST_LOG=debug FOO=1 cargo test --workspace"}),
            ),
        );
        hook(
            "claude-code",
            "Bash: tokenizer failure (unbalanced quote)",
            "claude-code",
            claude("Bash", json!({"command": "echo 'unbalanced"})),
        );
        hook(
            "codex-cli",
            "Bash",
            "codex-cli",
            codex("Bash", json!({"command": "cargo test --workspace"})),
        );
        hook(
            "codex-cli",
            "apply_patch",
            "codex-cli",
            codex(
                "apply_patch",
                json!({"input": "*** Begin Patch\n*** Update File: src/lib.rs\n@@\n-old\n+new\n*** End Patch\n"}),
            ),
        );
        hook(
            "cursor",
            "beforeShellExecution (no call id)",
            "cursor",
            serde_json::from_str(include_str!(
                "../../../tests/fixtures/cursor/before-shell-cli.json"
            ))
            .unwrap(),
        );
        hook(
            "cursor",
            "preToolUse Shell",
            "cursor",
            serde_json::from_str(include_str!(
                "../../../tests/fixtures/cursor/pre-tool-shell-ide.json"
            ))
            .unwrap(),
        );
        hook(
            "cursor",
            "beforeMCPExecution",
            "cursor",
            json!({
                "conversation_id": "5fb6e06f-ba89-4692-b119-a2a49e0e3042", "generation_id": "5fb6e06f-ba89-4692-b119-a2a49e0e3042",
                "model": "default", "tool_name": "create_issue", "tool_input": "{\"title\":\"Flaky test\"}",
                "mcp_server_name": "github", "url": "https://api.githubcopilot.com/mcp/",
                "session_id": "5fb6e06f-ba89-4692-b119-a2a49e0e3042", "hook_event_name": "beforeMCPExecution",
                "cursor_version": "3.22.12", "workspace_roots": ["/Users/dev/work"]
            }),
        );
        hook(
            "cursor",
            "preToolUse MCP:<tool> (server not named on this event)",
            "cursor",
            json!({
                "conversation_id": "7dc1864a-551b-4cf8-be04-21124ba704ff", "generation_id": "ad343d7e-47aa-4949-a7c7-2098ca482044",
                "model": "grok-4.6", "tool_name": "MCP:create_issue", "tool_input": {"title": "Flaky test"},
                "tool_use_id": "9b1f0c2e-1d4a-4b7e-8c3f-5a6d7e8f9012", "cwd": "", "session_id": "7dc1864a-551b-4cf8-be04-21124ba704ff",
                "hook_event_name": "preToolUse", "cursor_version": "3.22.12", "workspace_roots": ["/Users/dev/work"]
            }),
        );
        hook(
            "cline",
            "execute_command (classic host)",
            "cline",
            cline_classic(
                "execute_command",
                json!({"command": "npm run build", "requires_approval": "false"}),
            ),
        );
        hook(
            "cline",
            "run_commands list (SDK host)",
            "cline",
            json!({
                "clineVersion": "4.1.17", "hookName": "tool_call", "taskId": "conv_1",
                "workspaceRoots": ["/Users/dev/work/app"],
                "tool_call": {"id": "call_9", "name": "run_commands", "input": {"commands": ["npm ci", "npm test -- --watch=false"]}},
                "preToolUse": {"toolName": "run_commands", "parameters": {"commands": "[\"npm ci\",\"npm test -- --watch=false\"]"}}
            }),
        );
        hook(
            "cline",
            "use_mcp_tool",
            "cline",
            cline_classic(
                "use_mcp_tool",
                json!({
                    "server_name": "github", "tool_name": "create_issue", "arguments": "{\"title\":\"Flaky test\"}"
                }),
            ),
        );

        hook(
            "codex-cli",
            "MCP: mcp__github__create_issue",
            "codex-cli",
            codex(
                "mcp__github__create_issue",
                json!({"title": "Flaky test", "repo": "acme/app"}),
            ),
        );
        hook(
            "cursor",
            "preToolUse Read",
            "cursor",
            json!({
                "conversation_id": "7dc1864a-551b-4cf8-be04-21124ba704ff", "generation_id": "ad343d7e-47aa-4949-a7c7-2098ca482044",
                "model": "claude-opus-4-8", "tool_name": "Read", "tool_input": {"file_path": "/Users/dev/work/src/app.ts"},
                "tool_use_id": "4c2d8e1f-6a7b-4c3d-9e8f-0a1b2c3d4e5f", "cwd": "/Users/dev/work", "session_id": "7dc1864a-551b-4cf8-be04-21124ba704ff",
                "hook_event_name": "preToolUse", "cursor_version": "3.22.12", "workspace_roots": ["/Users/dev/work"]
            }),
        );
        hook(
            "cline",
            "read_file (classic host)",
            "cline",
            cline_classic("read_file", json!({"path": "src/app.ts"})),
        );

        let relay = |agent: &str,
                     title: &str,
                     wire: WireFormat,
                     model: &str,
                     session: &str,
                     usage: Usage| {
            let mut obs = Observation::none();
            obs.measured = true;
            obs.event_id = "0192f3a1-7c2e-7b4d-8e9f-1a2b3c4d5e6f".to_string();
            obs.occurred_at = TIME.to_string();
            obs.model = Some(model.to_string());
            obs.model_known = true;
            obs.install_id = "agt_550e8400e29b41d4a716446655440000".to_string();
            obs.wire_format = wire;
            obs.session = Resolved {
                agent_id: Some(obs.install_id.clone()),
                source: Some(agent.to_string()),
                session_id: Some(session.to_string()),
                assurance: Assurance::Attested,
                hook_seen: Some(true),
                agent_thread_id: None,
            };
            let data = assemble_data(&obs, &usage, CostBasis::ProviderReported, None, true);
            let event = assemble_event(&obs, data, &crate::privacy::PrivacyFilter::new(&[]));
            let (provider, model) = obs.identity(model);
            let action = relay_action(&obs, &usage, &provider, &model);
            assert_valid(&action);
            json!({
                "agent": agent, "action_type": "llm", "title": title,
                "input_envelope": event.envelope, "action": action,
            })
        };
        examples.push(relay(
            "claude-code",
            "Model relay: Anthropic Messages",
            WireFormat::AnthropicMessages,
            "claude-opus-4-8",
            "8f2c1e4a-6b1d-4c7e-9f3a-2d5b7c9e1f04",
            Usage {
                input_tokens: 412,
                cache_read: 18_240,
                cache_write: 1_024,
                eph_5m: 1_024,
                eph_1h: 0,
                output_tokens: 389,
            },
        ));
        examples.push(relay(
            "codex-cli",
            "Model relay: OpenAI Responses",
            WireFormat::OpenAiResponses,
            "gpt-5.2-codex",
            "s-41",
            Usage {
                input_tokens: 2_310,
                cache_read: 9_984,
                cache_write: 0,
                eph_5m: 0,
                eph_1h: 0,
                output_tokens: 655,
            },
        ));

        examples.push(relay(
            "cursor",
            "Model relay: Cursor backend (Connect-RPC)",
            WireFormat::CursorConnectRpc,
            "claude-opus-4-8",
            "7dc1864a-551b-4cf8-be04-21124ba704ff",
            Usage {
                input_tokens: 5_120,
                cache_read: 0,
                cache_write: 0,
                eph_5m: 0,
                eph_1h: 0,
                output_tokens: 742,
            },
        ));
        examples.push(relay(
            "cline",
            "Model relay: OpenAI-compatible Chat Completions",
            WireFormat::OpenAiChatCompletions,
            "gpt-5.2",
            "conv_1",
            Usage {
                input_tokens: 3_400,
                cache_read: 1_280,
                cache_write: 0,
                eph_5m: 0,
                eph_1h: 0,
                output_tokens: 512,
            },
        ));

        // `ACTION_EXAMPLES_OUT=<file>` picks the destination; the default stays inside the build tree.
        let path = std::env::var("ACTION_EXAMPLES_OUT").unwrap_or_else(|_| {
            concat!(env!("CARGO_MANIFEST_DIR"), "/target/action-examples.json").to_string()
        });
        let body = serde_json::to_string_pretty(&examples).unwrap();
        std::fs::write(&path, body + "\n").expect("write the examples");
        println!("wrote {} examples to {path}", examples.len());
    }
}