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::price_for;
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    summary: SessionSummary,
79}
80
81impl CodexTranscript {
82    pub fn new(path: impl Into<PathBuf>) -> Self {
83        CodexTranscript { reader: TailReader::new(path), summary: SessionSummary { harness: Some(Harness::Codex), ..Default::default() } }
84    }
85
86    fn ingest(&mut self, line: &str) {
87        let Ok(v) = serde_json::from_str::<Value>(line) else { return };
88        let ts = v.get("timestamp").and_then(Value::as_str).and_then(parse_rfc3339_utc);
89        if let Some(ts) = ts {
90            if self.summary.started_at.is_none() {
91                self.summary.started_at = Some(ts);
92            }
93            self.summary.last_activity = Some(ts);
94        }
95        let kind = v.get("type").and_then(Value::as_str).unwrap_or("");
96        let payload = v.get("payload");
97        let ptype = payload.and_then(|p| p.get("type")).and_then(Value::as_str).unwrap_or("");
98        match kind {
99            "session_meta" => {
100                if let Some(p) = payload {
101                    self.summary.session_id = p.get("id").or(p.get("session_id")).and_then(Value::as_str).map(str::to_string);
102                    self.summary.cwd = p.get("cwd").and_then(Value::as_str).map(PathBuf::from);
103                    self.summary.harness_version = p.get("cli_version").and_then(Value::as_str).map(str::to_string);
104                }
105            }
106            "turn_context" => {
107                if let Some(m) = payload.and_then(|p| p.get("model")).and_then(Value::as_str) {
108                    self.summary.model = Some(m.to_string());
109                }
110            }
111            "event_msg" => match ptype {
112                "token_count" => {
113                    if let Some(total) = payload.and_then(|p| p.pointer("/info/total_token_usage")) {
114                        let g = |k: &str| total.get(k).and_then(Value::as_u64).unwrap_or(0);
115                        let cached = g("cached_input_tokens");
116                        let usage = TokenUsage {
117                            input: g("input_tokens").saturating_sub(cached),
118                            cache_read: cached,
119                            output: g("output_tokens"),
120                            ..Default::default()
121                        };
122                        self.summary.usage = usage;
123                        let price = self.summary.model.as_deref().and_then(price_for);
124                        match price {
125                            Some(p) => {
126                                self.summary.cost_usd = p.cost(&usage);
127                                self.summary.unpriced_tokens = 0;
128                            }
129                            None => {
130                                self.summary.cost_usd = 0.0;
131                                self.summary.unpriced_tokens = usage.total();
132                            }
133                        }
134                    }
135                }
136                "task_started" | "user_message" => self.summary.activity = Activity::Working,
137                "task_complete" | "turn_aborted" | "error" => self.summary.activity = Activity::Waiting,
138                _ => {}
139            },
140            "response_item" => match ptype {
141                "function_call" | "custom_tool_call" | "local_shell_call" => {
142                    self.summary.tool_calls += 1;
143                    if let (Some(ts), Some(p)) = (ts, payload) {
144                        let id = call_id(p);
145                        let name = p.get("name").and_then(Value::as_str).unwrap_or(ptype);
146                        self.summary.spans.open(id, name.to_string(), ts, false);
147                    }
148                }
149                "function_call_output" | "custom_tool_call_output" | "local_shell_call_output" => {
150                    if let (Some(ts), Some(p)) = (ts, payload) {
151                        // Codex reports the result as an opaque string, and
152                        // agent-top does not read tool output, so a failed call
153                        // is not distinguishable from a successful one here.
154                        self.summary.spans.close(&call_id(p), ts, false);
155                    }
156                }
157                "message" if payload.and_then(|p| p.get("role")).and_then(Value::as_str) == Some("assistant") => {
158                    self.summary.turns += 1;
159                }
160                _ => {}
161            },
162            _ => {}
163        }
164    }
165}
166
167/// `call_id` on function calls, `id` on the shell-call variants.
168fn call_id(payload: &Value) -> String {
169    payload.get("call_id").or_else(|| payload.get("id")).and_then(Value::as_str).unwrap_or_default().to_string()
170}
171
172impl SessionTracker for CodexTranscript {
173    fn refresh(&mut self) -> anyhow::Result<bool> {
174        let (lines, more) = self.reader.read_new_lines(REFRESH_BUDGET_BYTES)?;
175        for l in &lines {
176            self.ingest(l);
177        }
178        Ok(more)
179    }
180
181    fn summary(&self) -> &SessionSummary {
182        &self.summary
183    }
184
185    fn path(&self) -> &Path {
186        self.reader.path()
187    }
188}
189
190#[cfg(test)]
191mod tests {
192    use super::*;
193    use std::io::Write;
194
195    #[test]
196    fn reads_cumulative_usage_and_state() {
197        let dir = std::env::temp_dir().join(format!("agent-top-codex-{}", std::process::id()));
198        std::fs::create_dir_all(&dir).unwrap();
199        let path = dir.join("rollout.jsonl");
200        let mut f = std::fs::File::create(&path).unwrap();
201        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();
202        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:21.000Z","type":"turn_context","payload":{{"model":"gpt-5-codex"}}}}"#).unwrap();
203        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:22.000Z","type":"event_msg","payload":{{"type":"task_started"}}}}"#).unwrap();
204        writeln!(
205            f,
206            r#"{{"timestamp":"2026-08-28T08:53:23.000Z","type":"response_item","payload":{{"type":"function_call","call_id":"call_1","name":"shell"}}}}"#
207        )
208        .unwrap();
209        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();
210        writeln!(f, r#"{{"timestamp":"2026-08-28T08:53:25.000Z","type":"event_msg","payload":{{"type":"token_count","info":null}}}}"#)
211            .unwrap();
212        let mut t = CodexTranscript::new(&path);
213        t.refresh().unwrap();
214        let s = t.summary();
215        assert_eq!(s.session_id.as_deref(), Some("01a0"));
216        assert_eq!(s.model.as_deref(), Some("gpt-5-codex"));
217        assert_eq!(s.usage.input, 14778 - 12672);
218        assert_eq!(s.usage.cache_read, 12672);
219        assert_eq!(s.usage.total(), 15019);
220        assert_eq!(s.unpriced_tokens, 15019);
221        assert_eq!(s.tool_calls, 1);
222        assert_eq!(s.activity, Activity::Working);
223        assert_eq!(read_meta(&path).unwrap().0, PathBuf::from("/tmp/p"));
224        let spans = s.spans.to_vec();
225        assert_eq!(spans.len(), 1);
226        assert_eq!(spans[0].name, "shell");
227        assert!(spans[0].is_open(), "no output item yet");
228        let _ = std::fs::remove_dir_all(&dir);
229    }
230
231    #[test]
232    fn pairs_calls_with_their_outputs() {
233        let dir = std::env::temp_dir().join(format!("agent-top-codex-spans-{}", std::process::id()));
234        std::fs::create_dir_all(&dir).unwrap();
235        let path = dir.join("rollout.jsonl");
236        let mut f = std::fs::File::create(&path).unwrap();
237        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();
238        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();
239        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();
240        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();
241        let mut t = CodexTranscript::new(&path);
242        t.refresh().unwrap();
243        let spans = t.summary().spans.to_vec();
244        assert_eq!(spans.len(), 2);
245        assert_eq!(spans[0].name, "exec_command");
246        assert_eq!(spans[0].duration_ms, Some(1_000));
247        assert_eq!(spans[1].name, "apply_patch");
248        assert_eq!(spans[1].duration_ms, Some(3_000));
249        assert_eq!(t.summary().tool_calls, 2);
250        let _ = std::fs::remove_dir_all(&dir);
251    }
252}