Skip to main content

agent_top_core/harness/
codex.rs

1//! OpenAI Codex CLI: `~/.codex/sessions/YYYY/MM/DD/rollout-<ts>-<id>.jsonl`.
2//!
3//! Format notes (verified on Codex CLI 0.149, 2026-09-03):
4//! * The first line is `session_meta` with `payload.cwd`, `payload.id`,
5//!   `payload.cli_version` and `payload.originator`.
6//! * `event_msg` / `token_count` carries `info.total_token_usage`, which is
7//!   cumulative for the session; `info` is null on rate-limit-only events.
8//!   `input_tokens` includes `cached_input_tokens`.
9//! * `task_started` / `task_complete` / `turn_aborted` bracket a turn.
10//! * `response_item` with `payload.type` `function_call` or
11//!   `custom_tool_call` is one tool call; the matching `*_output` item
12//!   carries the same `payload.call_id`, and the two lines' timestamps
13//!   bracket the call. That pairing is the trace.
14//!
15//! Codex model prices are not in the static table, so cost is reported as
16//! unpriced tokens.
17
18use super::{REFRESH_BUDGET_BYTES, SessionSummary, SessionTracker, parse_rfc3339_utc};
19use crate::jsonl::TailReader;
20use crate::model::{Activity, Harness, TokenUsage};
21use crate::pricing::{self, Table};
22use serde_json::Value;
23use std::path::{Path, PathBuf};
24use std::time::SystemTime;
25
26pub fn codex_dir() -> Option<PathBuf> {
27    if let Some(d) = std::env::var_os("CODEX_HOME") {
28        return Some(PathBuf::from(d));
29    }
30    std::env::var_os("HOME").map(|h| PathBuf::from(h).join(".codex"))
31}
32
33pub fn sessions_dir() -> Option<PathBuf> {
34    codex_dir().map(|d| d.join("sessions"))
35}
36
37/// Rollout files modified after `since`. Walks `YYYY/MM/DD` and prunes by
38/// directory mtime so the walk stays cheap on a long history.
39pub fn recent_rollouts(since: SystemTime) -> Vec<PathBuf> {
40    let Some(root) = sessions_dir() else { return Vec::new() };
41    let mut out = Vec::new();
42    walk(&root, 0, since, &mut out);
43    out
44}
45
46fn walk(dir: &Path, depth: usize, since: SystemTime, out: &mut Vec<PathBuf>) {
47    let Ok(rd) = std::fs::read_dir(dir) else { return };
48    for e in rd.flatten() {
49        let p = e.path();
50        let Ok(md) = e.metadata() else { continue };
51        if md.is_dir() {
52            if depth < 3 && md.modified().map(|m| m >= since).unwrap_or(true) {
53                walk(&p, depth + 1, since, out);
54            }
55        } else if p.extension().and_then(|x| x.to_str()) == Some("jsonl") && md.modified().map(|m| m >= since).unwrap_or(false) {
56            out.push(p);
57        }
58    }
59}
60
61/// Cheap header read: cwd and start time from the first line only.
62pub fn read_meta(path: &Path) -> Option<(PathBuf, SystemTime)> {
63    use std::io::{BufRead, BufReader};
64    let f = std::fs::File::open(path).ok()?;
65    let mut first = String::new();
66    BufReader::new(f).read_line(&mut first).ok()?;
67    let v: Value = serde_json::from_str(&first).ok()?;
68    if v.get("type").and_then(Value::as_str) != Some("session_meta") {
69        return None;
70    }
71    let cwd = v.pointer("/payload/cwd").and_then(Value::as_str).map(PathBuf::from)?;
72    let ts = v.get("timestamp").and_then(Value::as_str).and_then(parse_rfc3339_utc)?;
73    Some((cwd, ts))
74}
75
76pub struct CodexTranscript {
77    reader: TailReader,
78    prices: &'static Table,
79    summary: SessionSummary,
80}
81
82impl CodexTranscript {
83    pub fn new(path: impl Into<PathBuf>) -> Self {
84        CodexTranscript {
85            reader: TailReader::new(path),
86            prices: pricing::table(),
87            summary: SessionSummary { harness: Some(Harness::Codex), ..Default::default() },
88        }
89    }
90
91    /// See `ClaudeTranscript::with_prices`.
92    pub fn with_prices(mut self, prices: &'static Table) -> Self {
93        self.prices = prices;
94        self
95    }
96
97    fn ingest(&mut self, line: &str) {
98        let Ok(v) = serde_json::from_str::<Value>(line) else { return };
99        let ts = v.get("timestamp").and_then(Value::as_str).and_then(parse_rfc3339_utc);
100        if let Some(ts) = ts {
101            if self.summary.started_at.is_none() {
102                self.summary.started_at = Some(ts);
103            }
104            self.summary.last_activity = Some(ts);
105        }
106        let kind = v.get("type").and_then(Value::as_str).unwrap_or("");
107        let payload = v.get("payload");
108        let ptype = payload.and_then(|p| p.get("type")).and_then(Value::as_str).unwrap_or("");
109        match kind {
110            "session_meta" => {
111                if let Some(p) = payload {
112                    self.summary.session_id = p.get("id").or(p.get("session_id")).and_then(Value::as_str).map(str::to_string);
113                    self.summary.cwd = p.get("cwd").and_then(Value::as_str).map(PathBuf::from);
114                    self.summary.harness_version = p.get("cli_version").and_then(Value::as_str).map(str::to_string);
115                }
116            }
117            "turn_context" => {
118                if let Some(m) = payload.and_then(|p| p.get("model")).and_then(Value::as_str) {
119                    self.summary.model = Some(m.to_string());
120                }
121            }
122            "event_msg" => match ptype {
123                "token_count" => {
124                    if let Some(total) = payload.and_then(|p| p.pointer("/info/total_token_usage")) {
125                        let g = |k: &str| total.get(k).and_then(Value::as_u64).unwrap_or(0);
126                        let cached = g("cached_input_tokens");
127                        let usage = TokenUsage {
128                            input: g("input_tokens").saturating_sub(cached),
129                            cache_read: cached,
130                            output: g("output_tokens"),
131                            ..Default::default()
132                        };
133                        self.summary.usage = usage;
134                        let price = self.summary.model.as_deref().and_then(|m| self.prices.lookup(m));
135                        match price {
136                            Some(p) => {
137                                self.summary.cost_usd = p.cost(&usage);
138                                self.summary.unpriced_tokens = 0;
139                            }
140                            None => {
141                                self.summary.cost_usd = 0.0;
142                                self.summary.unpriced_tokens = usage.total();
143                            }
144                        }
145                    }
146                }
147                "task_started" | "user_message" => self.summary.activity = Activity::Working,
148                "task_complete" | "turn_aborted" | "error" => self.summary.activity = Activity::Waiting,
149                _ => {}
150            },
151            "response_item" => match ptype {
152                "function_call" | "custom_tool_call" | "local_shell_call" => {
153                    self.summary.tool_calls += 1;
154                    if let (Some(ts), Some(p)) = (ts, payload) {
155                        let id = call_id(p);
156                        let name = p.get("name").and_then(Value::as_str).unwrap_or(ptype);
157                        self.summary.spans.open(id, name.to_string(), ts, false);
158                    }
159                }
160                "function_call_output" | "custom_tool_call_output" | "local_shell_call_output" => {
161                    if let (Some(ts), Some(p)) = (ts, payload) {
162                        // Codex reports the result as an opaque string, and
163                        // agent-top does not read tool output, so a failed call
164                        // is not distinguishable from a successful one here.
165                        self.summary.spans.close(&call_id(p), ts, false);
166                    }
167                }
168                "message" if payload.and_then(|p| p.get("role")).and_then(Value::as_str) == Some("assistant") => {
169                    self.summary.turns += 1;
170                }
171                _ => {}
172            },
173            _ => {}
174        }
175    }
176}
177
178/// `call_id` on function calls, `id` on the shell-call variants.
179fn call_id(payload: &Value) -> String {
180    payload.get("call_id").or_else(|| payload.get("id")).and_then(Value::as_str).unwrap_or_default().to_string()
181}
182
183impl SessionTracker for CodexTranscript {
184    fn refresh(&mut self) -> anyhow::Result<bool> {
185        let (lines, more) = self.reader.read_new_lines(REFRESH_BUDGET_BYTES)?;
186        for l in &lines {
187            self.ingest(l);
188        }
189        Ok(more)
190    }
191
192    fn summary(&self) -> &SessionSummary {
193        &self.summary
194    }
195
196    fn path(&self) -> &Path {
197        self.reader.path()
198    }
199}
200
201#[cfg(test)]
202mod tests {
203    use super::*;
204    use std::io::Write;
205
206    #[test]
207    fn reads_cumulative_usage_and_state() {
208        let dir = std::env::temp_dir().join(format!("agent-top-codex-{}", std::process::id()));
209        std::fs::create_dir_all(&dir).unwrap();
210        let path = dir.join("rollout.jsonl");
211        let mut f = std::fs::File::create(&path).unwrap();
212        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:20.787Z","type":"session_meta","payload":{{"id":"01a0","cwd":"/tmp/p","cli_version":"0.149.1"}}}}"#).unwrap();
213        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:21.000Z","type":"turn_context","payload":{{"model":"gpt-5-codex"}}}}"#).unwrap();
214        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:22.000Z","type":"event_msg","payload":{{"type":"task_started"}}}}"#).unwrap();
215        writeln!(
216            f,
217            r#"{{"timestamp":"2026-08-28T08:53:23.000Z","type":"response_item","payload":{{"type":"function_call","call_id":"call_1","name":"shell"}}}}"#
218        )
219        .unwrap();
220        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:24.000Z","type":"event_msg","payload":{{"type":"token_count","info":{{"total_token_usage":{{"input_tokens":14778,"cached_input_tokens":12672,"output_tokens":241,"total_tokens":15019}}}}}}}}"#).unwrap();
221        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:25.000Z","type":"event_msg","payload":{{"type":"token_count","info":null}}}}"#)
222            .unwrap();
223        let mut t = CodexTranscript::new(&path);
224        t.refresh().unwrap();
225        let s = t.summary();
226        assert_eq!(s.session_id.as_deref(), Some("01a0"));
227        assert_eq!(s.model.as_deref(), Some("gpt-5-codex"));
228        assert_eq!(s.usage.input, 14778 - 12672);
229        assert_eq!(s.usage.cache_read, 12672);
230        assert_eq!(s.usage.total(), 15019);
231        assert_eq!(s.unpriced_tokens, 15019);
232        assert_eq!(s.tool_calls, 1);
233        assert_eq!(s.activity, Activity::Working);
234        assert_eq!(read_meta(&path).unwrap().0, PathBuf::from("/tmp/p"));
235        let spans = s.spans.to_vec();
236        assert_eq!(spans.len(), 1);
237        assert_eq!(spans[0].name, "shell");
238        assert!(spans[0].is_open(), "no output item yet");
239        let _ = std::fs::remove_dir_all(&dir);
240    }
241
242    #[test]
243    fn pairs_calls_with_their_outputs() {
244        let dir = std::env::temp_dir().join(format!("agent-top-codex-spans-{}", std::process::id()));
245        std::fs::create_dir_all(&dir).unwrap();
246        let path = dir.join("rollout.jsonl");
247        let mut f = std::fs::File::create(&path).unwrap();
248        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:23.000Z","type":"response_item","payload":{{"type":"function_call","call_id":"call_1","name":"exec_command"}}}}"#).unwrap();
249        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:23.100Z","type":"response_item","payload":{{"type":"custom_tool_call","call_id":"call_2","name":"apply_patch"}}}}"#).unwrap();
250        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:24.000Z","type":"response_item","payload":{{"type":"function_call_output","call_id":"call_1","output":"ok"}}}}"#).unwrap();
251        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:26.100Z","type":"response_item","payload":{{"type":"custom_tool_call_output","call_id":"call_2","output":"ok"}}}}"#).unwrap();
252        let mut t = CodexTranscript::new(&path);
253        t.refresh().unwrap();
254        let spans = t.summary().spans.to_vec();
255        assert_eq!(spans.len(), 2);
256        assert_eq!(spans[0].name, "exec_command");
257        assert_eq!(spans[0].duration_ms, Some(1_000));
258        assert_eq!(spans[1].name, "apply_patch");
259        assert_eq!(spans[1].duration_ms, Some(3_000));
260        assert_eq!(t.summary().tool_calls, 2);
261        let _ = std::fs::remove_dir_all(&dir);
262    }
263}