What causes "Failed to fetch experiments: {:?} Body: {}" in zed-industries/zed?

If you rely on the experiments, verify network access to the Zed API host and retry once connectivity returns.

Failed to fetch experiments: {:?} Body: {}

Error message

Failed to fetch experiments: {:?}
Body: {}

What it means

Zed edit-prediction periodically GETs its experiments endpoint (sending the Zed version header) to learn which experiments are enabled for this build; a non-2xx response bails with the status and the response body (edit_prediction.rs:1167). On failure `available_experiments` simply stays unset, so the feature degrades rather than crashes.

Source

Thrown at crates/edit_prediction/src/edit_prediction.rs:1167

                    let mut response = client
                        .authenticated_llm_request(&llm_token, organization_id, |token| {
                            Ok(http_client::Request::builder()
                                .method(Method::GET)
                                .uri(url.as_ref())
                                .header("Authorization", format!("Bearer {token}"))
                                .header(ZED_VERSION_HEADER_NAME, app_version.to_string())
                                .body(Default::default())?)
                        })
                        .await?;
                    if response.status().is_success() {
                        let mut body = Vec::new();
                        response.body_mut().read_to_end(&mut body).await?;
                        let experiments: Vec<String> = serde_json::from_slice(&body)?;
                        Ok(experiments)
                    } else {
                        let mut body = String::new();
                        response.body_mut().read_to_string(&mut body).await?;
                        anyhow::bail!(
                            "Failed to fetch experiments: {:?}\nBody: {}",
                            response.status(),
                            body
                        );
                    }
                })
                .await?;
            this.update(cx, |this, cx| {
                this.available_experiments = experiments;
                cx.notify();
            })?;
            anyhow::Ok(())
        })
        .detach_and_log_err(cx);
    }

    pub fn icons(&self, cx: &App) -> edit_prediction_types::EditPredictionIconSet {
        use ui::IconName;

View on GitHub (pinned to f4178619ac)

Solutions

If you rely on the experiments, verify network access to the Zed API host and retry once connectivity returns.
Otherwise no action is needed — the fetch is non-blocking for the rest of edit prediction and the editor.
If it persists with 4xx statuses, report it — a version-header mismatch may be involved.

Defensive patterns

Strategy: fallback

Try / catch

Wrap the fetch and treat any Err as "experiments unavailable": keep the previous (empty) experiment set, log at warn, and schedule the next periodic fetch — never propagate into the editing flow.

Prevention

Treat optional capability fetches as best-effort with their own error boundary.
Log status + body once per occurrence to catch persistent auth regressions.
Keep a cached last-known experiments list to survive transient outages.

When it happens

Trigger: The experiments endpoint returning 5xx during an incident, a 4xx from an auth/CDN layer, or the client being unable to pass a proxy that blocks the Zed API host — any non-success status on that GET.

Common situations: Zed backend incidents; corporate proxies filtering api.zed.dev; offline development — in all cases edit-prediction experiments remain disabled until a later successful fetch.

Understand the failure class

Background: "API error: {status}" and "HTTP 401/403/404/429/5xx" errors: non-2xx HTTP responses explained — this error's family across 27 libraries.

Related errors

AI-assisted analysis of zed-industries/zed@f4178619ac (2026-08-20). Data as JSON: /api/errors/e876170487b8d784. Report an issue: GitHub.