Files
codex/codex-rs/core/src/apply_patch.rs
T
vishnu-oaiandAnton Panasenko 04c1782e52 OpenTelemetry events (#2103)
### Title

## otel

Codex can emit [OpenTelemetry](https://opentelemetry.io/) **log events**
that
describe each run: outbound API requests, streamed responses, user
input,
tool-approval decisions, and the result of every tool invocation. Export
is
**disabled by default** so local runs remain self-contained. Opt in by
adding an
`[otel]` table and choosing an exporter.

```toml
[otel]
environment = "staging"   # defaults to "dev"
exporter = "none"          # defaults to "none"; set to otlp-http or otlp-grpc to send events
log_user_prompt = false    # defaults to false; redact prompt text unless explicitly enabled
```

Codex tags every exported event with `service.name = "codex-cli"`, the
CLI
version, and an `env` attribute so downstream collectors can distinguish
dev/staging/prod traffic. Only telemetry produced inside the
`codex_otel`
crate—the events listed below—is forwarded to the exporter.

### Event catalog

Every event shares a common set of metadata fields: `event.timestamp`,
`conversation.id`, `app.version`, `auth_mode` (when available),
`user.account_id` (when available), `terminal.type`, `model`, and
`slug`.

With OTEL enabled Codex emits the following event types (in addition to
the
metadata above):

- `codex.api_request`
  - `cf_ray` (optional)
  - `attempt`
  - `duration_ms`
  - `http.response.status_code` (optional)
  - `error.message` (failures)
- `codex.sse_event`
  - `event.kind`
  - `duration_ms`
  - `error.message` (failures)
  - `input_token_count` (completion only)
  - `output_token_count` (completion only)
  - `cached_token_count` (completion only, optional)
  - `reasoning_token_count` (completion only, optional)
  - `tool_token_count` (completion only)
- `codex.user_prompt`
  - `prompt_length`
  - `prompt` (redacted unless `log_user_prompt = true`)
- `codex.tool_decision`
  - `tool_name`
  - `call_id`
- `decision` (`approved`, `approved_for_session`, `denied`, or `abort`)
  - `source` (`config` or `user`)
- `codex.tool_result`
  - `tool_name`
  - `call_id`
  - `arguments`
  - `duration_ms` (execution time for the tool)
  - `success` (`"true"` or `"false"`)
  - `output`

### Choosing an exporter

Set `otel.exporter` to control where events go:

- `none` – leaves instrumentation active but skips exporting. This is
the
  default.
- `otlp-http` – posts OTLP log records to an OTLP/HTTP collector.
Specify the
  endpoint, protocol, and headers your collector expects:

  ```toml
  [otel]
  exporter = { otlp-http = {
    endpoint = "https://otel.example.com/v1/logs",
    protocol = "binary",
    headers = { "x-otlp-api-key" = "${OTLP_TOKEN}" }
  }}
  ```

- `otlp-grpc` – streams OTLP log records over gRPC. Provide the endpoint
and any
  metadata headers:

  ```toml
  [otel]
  exporter = { otlp-grpc = {
    endpoint = "https://otel.example.com:4317",
    headers = { "x-otlp-meta" = "abc123" }
  }}
  ```

If the exporter is `none` nothing is written anywhere; otherwise you
must run or point to your
own collector. All exporters run on a background batch worker that is
flushed on
shutdown.

If you build Codex from source the OTEL crate is still behind an `otel`
feature
flag; the official prebuilt binaries ship with the feature enabled. When
the
feature is disabled the telemetry hooks become no-ops so the CLI
continues to
function without the extra dependencies.

---------

Co-authored-by: Anton Panasenko <apanasenko@openai.com>
2025-09-29 11:30:55 -07:00

112 lines
4.3 KiB
Rust

use crate::codex::Session;
use crate::codex::TurnContext;
use crate::function_tool::FunctionCallError;
use crate::protocol::FileChange;
use crate::protocol::ReviewDecision;
use crate::safety::SafetyCheck;
use crate::safety::assess_patch_safety;
use codex_apply_patch::ApplyPatchAction;
use codex_apply_patch::ApplyPatchFileChange;
use std::collections::HashMap;
use std::path::PathBuf;
pub const CODEX_APPLY_PATCH_ARG1: &str = "--codex-run-as-apply-patch";
pub(crate) enum InternalApplyPatchInvocation {
/// The `apply_patch` call was handled programmatically, without any sort
/// of sandbox, because the user explicitly approved it. This is the
/// result to use with the `shell` function call that contained `apply_patch`.
Output(Result<String, FunctionCallError>),
/// The `apply_patch` call was approved, either automatically because it
/// appears that it should be allowed based on the user's sandbox policy
/// *or* because the user explicitly approved it. In either case, we use
/// exec with [`CODEX_APPLY_PATCH_ARG1`] to realize the `apply_patch` call,
/// but [`ApplyPatchExec::auto_approved`] is used to determine the sandbox
/// used with the `exec()`.
DelegateToExec(ApplyPatchExec),
}
pub(crate) struct ApplyPatchExec {
pub(crate) action: ApplyPatchAction,
pub(crate) user_explicitly_approved_this_action: bool,
}
pub(crate) async fn apply_patch(
sess: &Session,
turn_context: &TurnContext,
sub_id: &str,
call_id: &str,
action: ApplyPatchAction,
) -> InternalApplyPatchInvocation {
match assess_patch_safety(
&action,
turn_context.approval_policy,
&turn_context.sandbox_policy,
&turn_context.cwd,
) {
SafetyCheck::AutoApprove {
user_explicitly_approved,
..
} => InternalApplyPatchInvocation::DelegateToExec(ApplyPatchExec {
action,
user_explicitly_approved_this_action: user_explicitly_approved,
}),
SafetyCheck::AskUser => {
// Compute a readable summary of path changes to include in the
// approval request so the user can make an informed decision.
//
// Note that it might be worth expanding this approval request to
// give the user the option to expand the set of writable roots so
// that similar patches can be auto-approved in the future during
// this session.
let rx_approve = sess
.request_patch_approval(sub_id.to_owned(), call_id.to_owned(), &action, None, None)
.await;
match rx_approve.await.unwrap_or_default() {
ReviewDecision::Approved | ReviewDecision::ApprovedForSession => {
InternalApplyPatchInvocation::DelegateToExec(ApplyPatchExec {
action,
user_explicitly_approved_this_action: true,
})
}
ReviewDecision::Denied | ReviewDecision::Abort => {
InternalApplyPatchInvocation::Output(Err(FunctionCallError::RespondToModel(
"patch rejected by user".to_string(),
)))
}
}
}
SafetyCheck::Reject { reason } => InternalApplyPatchInvocation::Output(Err(
FunctionCallError::RespondToModel(format!("patch rejected: {reason}")),
)),
}
}
pub(crate) fn convert_apply_patch_to_protocol(
action: &ApplyPatchAction,
) -> HashMap<PathBuf, FileChange> {
let changes = action.changes();
let mut result = HashMap::with_capacity(changes.len());
for (path, change) in changes {
let protocol_change = match change {
ApplyPatchFileChange::Add { content } => FileChange::Add {
content: content.clone(),
},
ApplyPatchFileChange::Delete { content } => FileChange::Delete {
content: content.clone(),
},
ApplyPatchFileChange::Update {
unified_diff,
move_path,
new_content: _new_content,
} => FileChange::Update {
unified_diff: unified_diff.clone(),
move_path: move_path.clone(),
},
};
result.insert(path.clone(), protocol_change);
}
result
}