export declare const HASHLINE_EDIT_DESCRIPTION = "Edit files using exact hash-anchored line references from the latest Read output.\n\n<must>\n1. SNAPSHOT: All edits in one call reference the ORIGINAL file state. Do NOT adjust line numbers for prior edits in the same batch \u2014 the system applies them bottom-up automatically.\n2. replace replaces ONE line at pos. It does NOT accept end. For multi-line replacement, use replace_range.\n3. replace_range with pos+end replaces ALL lines FROM pos THROUGH end (BOTH INCLUSIVE). The end line WILL BE replaced. If you set end to a line that belongs to the next function/method/statement, that line is DELETED.\n   CORRECT: pos on the first line to replace, end on the LAST line to replace \u2014 not the line after.\n4. lines must contain ONLY the content that belongs inside the replaced range. Lines AFTER end survive unchanged \u2014 do NOT include them in lines. If you do, they will appear twice.\n5. Tags MUST be copied exactly from read output or >>> mismatch output. NEVER guess or reconstruct tags.\n6. Batch = multiple operations in edits[], NOT one big replace covering everything. Each operation targets the smallest possible change.\n7. lines must contain plain replacement text only (no LINE#HASH#ANCHOR| prefixes, no diff + markers).\n</must>\n\n<operations>\nANCHOR FORMAT:\n  Each anchor is `{line_number}#{hash_id}#{anchor_hash}` from read output like `42#VK#AB|content`.\n  Backward-compatible `{line}#{hash}` also accepted.\n\nOPERATION CHOICE:\n  replace with pos -> replace ONE line at pos (end is rejected)\n  replace_range with pos+end -> replace range pos..end INCLUSIVE (both lines replaced)\n  append with pos -> insert lines AFTER the anchored line (use when you need to ADD lines, not replace)\n  prepend with pos -> insert lines BEFORE the anchored line\n  append/prepend without pos -> EOF/BOF insertion (also creates missing files)\n\nCONTENT FORMAT:\n  lines: string (single line) or string[] (multi-line, preferred)\n  lines: null or lines: [] with replace -> DELETE those lines\n\nFILE MODES:\n  delete=true deletes file and requires edits=[] with no rename\n  rename moves final content to a new path and removes old path\n<operations>\n\n<examples>\nGiven this file after read:\n  10#VK#AB|function hello() {\n  11#XJ#CD|  console.log(\"hi\");\n  12#MB#EF|  console.log(\"bye\");\n  13#QR#GH|} // end of hello()\n\nSingle-line replace (change line 11):\n  { op: \"replace\", pos: \"11#XJ#CD\", lines: [\"  console.log(\"hello\");\"] }\n  Result: line 11 replaced. Lines 10, 12-13 unchanged.\n\nRange replace (replace lines 11-12, function body):\n  { op: \"replace_range\", pos: \"11#XJ#CD\", end: \"12#MB#EF\", lines: [\"  return \"hello world\";\"] }\n  Result: lines 11-12 removed, replaced by 1 new line. Lines 10, 13 unchanged.\n\nBAD - end is one line too far (DELETES closing brace):\n  { op: \"replace_range\", pos: \"11#XJ#CD\", end: \"13#QR#GH\", lines: [\"  return \"hello world\";\"] }\n  Result: line 13 (closing brace) is REPLACED too \u2014 function is broken!\n  CORRECT: use end: \"12#MB#EF\" \u2014 only replace lines 11-12, keep line 13 unchanged.\n\nBAD - lines extend past end (DUPLICATES line 13):\n  { op: \"replace_range\", pos: \"11#XJ#CD\", end: \"12#MB#EF\", lines: [\"  return \"hi\";\", \"}\"] }\n  Line 13 is \"}\" which already exists after end. Including it in lines duplicates it.\n  CORRECT: { op: \"replace_range\", pos: \"11#XJ#CD\", end: \"12#MB#EF\", lines: [\"  return \"hi\";\"] }\n\nAppend after a line (insert between functions):\n  { op: \"append\", pos: \"13#QR#GH\", lines: [\"\", \"function added() {\", \"  return true;\", \"}\"] }\n  Result: 4 lines inserted after line 13. All existing lines unchanged.\n</examples>\n\n<auto>\nBuilt-in autocorrect (you do NOT need to handle these):\n  Merged lines are auto-expanded back to original line count.\n  Indentation is auto-restored from original lines.\n  BOM and CRLF line endings are preserved automatically.\n  Hashline prefixes and diff markers in text are auto-stripped.\n  Boundary echo lines (duplicating adjacent surviving lines) are auto-stripped.\n</auto>\n\nRecovery:\n- If you get a hash mismatch error, copy the updated anchors shown in that error or re-read the file before retrying.";