What causes "Error encoding custom model: %v" in plandex-ai/plandex?

Check client-side logs: the request was likely aborted by the caller — retry with a stable connection.

Error encoding custom model: %v

Error message

Error encoding custom model: %v

What it means

GetCustomModelHandler writes the fetched custom model to the HTTP response via json.NewEncoder(w).Encode(res.ToApi()). If encoding fails — the response writer has broken/failed (client disconnected, connection reset) or the value cannot be marshaled — the handler logs 'Error encoding custom model' and returns 500 with the same message. In practice this is almost always a broken HTTP pipe, not a data problem.

Source

Thrown at app/server/handlers/models.go:366

	id := mux.Vars(r)["modelId"]

	res, err := db.GetCustomModel(auth.OrgId, id)
	if err != nil {
		log.Printf("Error fetching custom model: %v\n", err)
		http.Error(w, "Failed to fetch custom model: "+err.Error(), http.StatusInternalServerError)
		return
	}

	if res == nil {
		http.Error(w, "Custom model not found", http.StatusNotFound)
		return
	}

	err = json.NewEncoder(w).Encode(res.ToApi())
	if err != nil {
		log.Printf("Error encoding custom model: %v\n", err)
		http.Error(w, fmt.Sprintf("Error encoding custom model: %v", err), http.StatusInternalServerError)
		return
	}

	log.Println("Successfully fetched custom model")
}

func ListCustomModelsHandler(w http.ResponseWriter, r *http.Request) {
	log.Println("Received request for ListCustomModelsHandler")

	auth := Authenticate(w, r, true)
	if auth == nil {
		return
	}

	if !requireMinClientVersion(w, r, CustomModelsMinClientVersion) {
		return
	}

View on GitHub (pinned to e2d772072e)

Solutions

Check client-side logs: the request was likely aborted by the caller — retry with a stable connection.
Verify any reverse proxy (nginx, ALB) timeout/read-buffer settings are not closing the connection mid-write.
If it persists with a live client, inspect CustomModel.ToApi() output for unmarshalable values (channels, funcs, invalid UTF-8) added in recent changes.
Ensure the handler is not writing after the response has already been partially committed elsewhere.

Example fix

// before
err = json.NewEncoder(w).Encode(res.ToApi())
if err != nil {
    http.Error(w, fmt.Sprintf("Error encoding custom model: %v", err), http.StatusInternalServerError)
}
// after
buf, err := json.Marshal(res.ToApi())
if err != nil {
    http.Error(w, "encoding failed", http.StatusInternalServerError)
    return
}
w.Header().Set("Content-Type", "application/json")
w.Write(buf) // encode to buffer first; transport errors on w are not client-fixable

Defensive patterns

Strategy: try-catch

Validate before calling

// Go: verify the request context is still alive before doing work
if r.Context().Err() != nil { return } // client already gone; skip the DB/encode work

Try / catch

if err := json.NewEncoder(w).Encode(res.ToApi()); err != nil {
    // after headers are sent you cannot change the status; log and return
    log.Printf("encode/transport error: %v", err)
    return
}

Prevention

Marshal to a buffer with json.Marshal first, then write once, so data errors differ from transport errors
Set Content-Type and status before the first write so errors are not half-committed
Check r.Context().Err() before expensive work to skip aborted requests
Keep ToApi() outputs JSON-safe: no channels, funcs, or cycles

When it happens

Trigger: Client disconnects or the connection is reset mid-response while json.Encoder writes the GET custom model body; or writing to w returns an error for any transport reason.

Common situations: Client cancels the request or times out before the body is written; reverse proxy/load balancer closes an upstream connection; mobile client drops network during response.

Related errors

AI-assisted analysis of plandex-ai/plandex@e2d772072e (2026-09-05). Data as JSON: /api/errors/d116f893853f983b. Report an issue: GitHub.