tui: single-line file mention (lineStart == lineEnd, or "#N") attaches from that line to the end of the file #58033

Activity

1. opencode-agent commented on Oct 22, 2026

   

   opencode-agentBot

   on Oct 22, 2026 – with OpenCode Agent

   Contributor

   More actions

   Thanks for the detailed report. I can reproduce this on v2.0.32, and it still happens on the latest dev (c93aaae).

   What I did, in a project with a 286-line big.txt:

   1. Typed @big.txt#10-10 in the prompt and pressed Tab to accept the autocomplete entry. The chip shows big.txt#10.
   2. Sent the prompt with some text.
   3. Checked the request sent to the model: it contained lines 10 to 286 (277 lines), not only line 10.

   The recording and screenshot below show it, and the script below reproduces it. I tested the typed #10-10 mention rather than the editor at_mentioned path, but both drop the end line through the same lineEnd > lineStart / start < end checks you pointed out in insertFileMention and parseFileLineRange.

   recording.mp4

   

   Reproduction script (drive.ts)

   // Repro for #58033: a single-line mention (@big.txt#10-10) attaches lines 10..EOF.
   import { Effect, Stream } from "effect"
   import { Llm, OpenCodeDriver } from "opencode-drive"
   
   const lines = Array.from({ length: 286 }, (_, i) => `LINE_${i + 1}_MARK`).join("\n") + "\n"
   
   export default OpenCodeDriver.use(
     {
       ...(process.env.OC_DEV ? { opencode: { dev: process.env.OC_DEV } } : {}),
       project: { git: true, files: { "big.txt": lines } },
       tui: { recording: true, keypressOverlay: true },
     },
     ({ ui, llm }) =>
       Effect.gen(function* () {
         let body = ""
         yield* llm.serve((request: any) => {
           body = JSON.stringify(request.body ?? request)
           return Stream.make(Llm.text("ok"))
         })
         yield* ui.type("@big.txt#10-10")
         yield* Effect.sleep("1500 millis")
         yield* ui.screenshot("autocomplete")
         yield* ui.press("tab")
         yield* Effect.sleep("500 millis")
         yield* ui.screenshot("mention-inserted")
         yield* ui.type(" summarize")
         yield* ui.enter()
         yield* ui.waitFor("ok")
         yield* Effect.sleep("1000 millis")
         yield* ui.screenshot("after-send")
         const sent = [...body.matchAll(/LINE_(\d+)_MARK/g)].map((m) => Number(m[1]))
         const uris = [...new Set(body.match(/file:\/\/[^"\\ ]+/g) ?? [])]
         yield* Effect.log(`file URIs in request: ${JSON.stringify(uris)}`)
         yield* Effect.log(
           `lines sent to model: count=${new Set(sent).size} min=${Math.min(...sent)} max=${Math.max(...sent)}`,
         )
         yield* Effect.log(new Set(sent).size > 1 ? "BUG: more than one line attached" : "OK: one line attached")
       }),
   )

   Run it with opencode-drive: opencode-drive run ./drive.ts. It drives opencode from your PATH; set OC_DEV to the path of an OpenCode checkout to run that from source instead.
2. opencode-agent

   added

   reproduced

   on Oct 22, 2026