import z from "zod"import * as fs from "fs"import * as path from "path"import { Tool } from "./tool"import { LSP } from "../lsp"import { FileTime } from "../file/time"import DESCRIPTION from "./read.txt"import { Instance } from "../project/instance"import { Identifier } from "../id/id"import { assertExternalDirectory } from "./external-directory"import { InstructionPrompt } from "../session/instruction"import { readImageDimensions } from "../util/image" const DEFAULT_READ_LIMIT = 2000const MAX_LINE_LENGTH = 2000const MAX_BYTES = 50 * 1024// Anthropic's API caps base64-embedded PDFs/images at ~32 MB. Reject bigger// files up front instead of OOMing while encoding.const MAX_ATTACHMENT_BYTES = 32 * 1024 * 1024// Anthropic rejects images with any dimension > 2000px when a request contains// multiple images. Reject at attach time so a single oversized figure cannot// poison the entire session's history (which would fail every follow-up turn).const MAX_IMAGE_DIMENSION = 2000 export const ReadTool = Tool.define("read", { description: DESCRIPTION, parameters: z.object({ filePath: z.string().describe("The path to the file to read"), offset: z.coerce.number().describe("The line number to start reading from (0-based)").optional(), limit: z.coerce.number().describe("The number of lines to read (defaults to 2000)").optional(), }), async execute(params, ctx) { let filepath = params.filePath if (!path.isAbsolute(filepath)) { filepath = path.resolve(Instance.directory, filepath) } const title = path.relative(Instance.worktree, filepath) await assertExternalDirectory(ctx, filepath, { bypass: Boolean(ctx.extra?.["bypassCwdCheck"]), }) await ctx.ask({ permission: "read", patterns: [filepath], always: ["*"], metadata: {}, }) const file = Bun.file(filepath) if (!(await file.exists())) { const dir = path.dirname(filepath) const base = path.basename(filepath) const dirEntries = fs.readdirSync(dir) const suggestions = dirEntries .filter( (entry) => entry.toLowerCase().includes(base.toLowerCase()) || base.toLowerCase().includes(entry.toLowerCase()), ) .map((entry) => path.join(dir, entry)) .slice(0, 3) if (suggestions.length > 0) { throw new Error(`File not found: ${filepath}\n\nDid you mean one of these?\n${suggestions.join("\n")}`) } throw new Error(`File not found: ${filepath}`) } const instructions = await InstructionPrompt.resolve(ctx.messages, filepath, ctx.messageID) // Exclude SVG (XML-based) and vnd.fastbidsheet (.fbs extension, commonly FlatBuffers schema files) const isImage = file.type.startsWith("image/") && file.type !== "image/svg+xml" && file.type !== "image/vnd.fastbidsheet" const isPdf = file.type === "application/pdf" if (isImage || isPdf) { const kind = isImage ? "Image" : "PDF" const attachStat = await file.stat() if (attachStat.size > MAX_ATTACHMENT_BYTES) { throw new Error( `${kind} too large to attach (${attachStat.size} bytes > ${MAX_ATTACHMENT_BYTES}). ` + `Anthropic's API caps base64 attachments at ~32 MB. ` + (isPdf ? "Use the liteparse skill to extract text via the `lit` CLI instead " + "(install with `npm i -g @llamaindex/liteparse` if missing)." : "Crop or downscale the image."), ) } const mime = file.type const fileBytes = await file.bytes() if (isImage) { const dims = readImageDimensions(fileBytes) if (dims && Math.max(dims.width, dims.height) > MAX_IMAGE_DIMENSION) { throw new Error( `Image too large to attach (${dims.width}x${dims.height}). ` + `Anthropic's API rejects any image dimension > ${MAX_IMAGE_DIMENSION}px in multi-image requests, ` + `which would poison the entire session. Downscale first, e.g.:\n` + ` magick "${filepath}" -resize ${MAX_IMAGE_DIMENSION - 200}x${MAX_IMAGE_DIMENSION - 200}\\> "${filepath}"`, ) } } const msg = `${kind} read successfully` return { title, output: msg, metadata: { preview: msg, truncated: false, ...(instructions.length > 0 && { loaded: instructions.map((i) => i.filepath) }), }, attachments: [ { id: Identifier.ascending("part"), sessionID: ctx.sessionID, messageID: ctx.messageID, type: "file", mime, url: `data:${mime};base64,${Buffer.from(fileBytes).toString("base64")}`, }, ], } } const isBinary = await isBinaryFile(filepath, file) if (isBinary) throw new Error(`Cannot read binary file: ${filepath}`) const limit = params.limit ?? DEFAULT_READ_LIMIT const offset = params.offset || 0 const lines = await file.text().then((text) => text.split("\n")) const raw: string[] = [] let bytes = 0 let truncatedByBytes = false for (let i = offset; i < Math.min(lines.length, offset + limit); i++) { const line = lines[i].length > MAX_LINE_LENGTH ? lines[i].substring(0, MAX_LINE_LENGTH) + "..." : lines[i] const size = Buffer.byteLength(line, "utf-8") + (raw.length > 0 ? 1 : 0) if (bytes + size > MAX_BYTES) { truncatedByBytes = true break } raw.push(line) bytes += size } const content = raw.map((line, index) => { return `${(index + offset + 1).toString().padStart(5, "0")}| ${line}` }) const preview = raw.slice(0, 20).join("\n") let output = "<file>\n" output += content.join("\n") const totalLines = lines.length const lastReadLine = offset + raw.length const hasMoreLines = totalLines > lastReadLine const truncated = hasMoreLines || truncatedByBytes if (truncatedByBytes) { output += `\n\n(Output truncated at ${MAX_BYTES} bytes. Use 'offset' parameter to read beyond line ${lastReadLine})` } else if (hasMoreLines) { output += `\n\n(File has more lines. Use 'offset' parameter to read beyond line ${lastReadLine})` } else { output += `\n\n(End of file - total ${totalLines} lines)` } output += "\n</file>" // just warms the lsp client LSP.touchFile(filepath, false) FileTime.read(ctx.sessionID, filepath) if (instructions.length > 0) { output += `\n\n<system-reminder>\n${instructions.map((i) => i.content).join("\n\n")}\n</system-reminder>` } return { title, output, metadata: { preview, truncated, ...(instructions.length > 0 && { loaded: instructions.map((i) => i.filepath) }), }, } },}) async function isBinaryFile(filepath: string, file: Bun.BunFile): Promise<boolean> { const ext = path.extname(filepath).toLowerCase() // binary check for common non-text extensions switch (ext) { case ".zip": case ".tar": case ".gz": case ".exe": case ".dll": case ".so": case ".class": case ".jar": case ".war": case ".7z": case ".doc": case ".docx": case ".xls": case ".xlsx": case ".ppt": case ".pptx": case ".odt": case ".ods": case ".odp": case ".bin": case ".dat": case ".obj": case ".o": case ".a": case ".lib": case ".wasm": case ".pyc": case ".pyo": return true default: break } const stat = await file.stat() const fileSize = stat.size if (fileSize === 0) return false // Only read the header, not the whole file — a 2 GB binary would OOM otherwise. const bufferSize = Math.min(4096, fileSize) const head = file.slice(0, bufferSize) const buffer = await head.arrayBuffer() if (buffer.byteLength === 0) return false const bytes = new Uint8Array(buffer) let nonPrintableCount = 0 for (let i = 0; i < bytes.length; i++) { if (bytes[i] === 0) return true if (bytes[i] < 9 || (bytes[i] > 13 && bytes[i] < 32)) { nonPrintableCount++ } } // If >30% non-printable characters, consider it binary return nonPrintableCount / bytes.length > 0.3}
import z from "zod"
import * as fs from "fs"
import * as path from "path"
import { Tool } from "./tool"
import { LSP } from "../lsp"
import { FileTime } from "../file/time"
import DESCRIPTION from "./read.txt"
import { Instance } from "../project/instance"
import { Identifier } from "../id/id"
import { assertExternalDirectory } from "./external-directory"
import { InstructionPrompt } from "../session/instruction"
import { readImageDimensions } from "../util/image"
const DEFAULT_READ_LIMIT = 2000
const MAX_LINE_LENGTH = 2000
const MAX_BYTES = 50 * 1024
// Anthropic's API caps base64-embedded PDFs/images at ~32 MB. Reject bigger
// files up front instead of OOMing while encoding.
const MAX_ATTACHMENT_BYTES = 32 * 1024 * 1024
// Anthropic rejects images with any dimension > 2000px when a request contains
// multiple images. Reject at attach time so a single oversized figure cannot
// poison the entire session's history (which would fail every follow-up turn).
const MAX_IMAGE_DIMENSION = 2000
export const ReadTool = Tool.define("read", {
description: DESCRIPTION,
parameters: z.object({
filePath: z.string().describe("The path to the file to read"),
offset: z.coerce.number().describe("The line number to start reading from (0-based)").optional(),
limit: z.coerce.number().describe("The number of lines to read (defaults to 2000)").optional(),
}),
async execute(params, ctx) {
let filepath = params.filePath
if (!path.isAbsolute(filepath)) {
filepath = path.resolve(Instance.directory, filepath)
}
const title = path.relative(Instance.worktree, filepath)
await assertExternalDirectory(ctx, filepath, {
bypass: Boolean(ctx.extra?.["bypassCwdCheck"]),
})
await ctx.ask({
permission: "read",
patterns: [filepath],
always: ["*"],
metadata: {},
})
const file = Bun.file(filepath)
if (!(await file.exists())) {
const dir = path.dirname(filepath)
const base = path.basename(filepath)
const dirEntries = fs.readdirSync(dir)
const suggestions = dirEntries
.filter(
(entry) =>
entry.toLowerCase().includes(base.toLowerCase()) || base.toLowerCase().includes(entry.toLowerCase()),
)
.map((entry) => path.join(dir, entry))
.slice(0, 3)
if (suggestions.length > 0) {
throw new Error(`File not found: ${filepath}\n\nDid you mean one of these?\n${suggestions.join("\n")}`)
}
throw new Error(`File not found: ${filepath}`)
}
const instructions = await InstructionPrompt.resolve(ctx.messages, filepath, ctx.messageID)
// Exclude SVG (XML-based) and vnd.fastbidsheet (.fbs extension, commonly FlatBuffers schema files)
const isImage =
file.type.startsWith("image/") && file.type !== "image/svg+xml" && file.type !== "image/vnd.fastbidsheet"
const isPdf = file.type === "application/pdf"
if (isImage || isPdf) {
const kind = isImage ? "Image" : "PDF"
const attachStat = await file.stat()
if (attachStat.size > MAX_ATTACHMENT_BYTES) {
throw new Error(
`${kind} too large to attach (${attachStat.size} bytes > ${MAX_ATTACHMENT_BYTES}). ` +
`Anthropic's API caps base64 attachments at ~32 MB. ` +
(isPdf
? "Use the liteparse skill to extract text via the `lit` CLI instead " +
"(install with `npm i -g @llamaindex/liteparse` if missing)."
: "Crop or downscale the image."),
)
}
const mime = file.type
const fileBytes = await file.bytes()
if (isImage) {
const dims = readImageDimensions(fileBytes)
if (dims && Math.max(dims.width, dims.height) > MAX_IMAGE_DIMENSION) {
throw new Error(
`Image too large to attach (${dims.width}x${dims.height}). ` +
`Anthropic's API rejects any image dimension > ${MAX_IMAGE_DIMENSION}px in multi-image requests, ` +
`which would poison the entire session. Downscale first, e.g.:\n` +
` magick "${filepath}" -resize ${MAX_IMAGE_DIMENSION - 200}x${MAX_IMAGE_DIMENSION - 200}\\> "${filepath}"`,
)
}
}
const msg = `${kind} read successfully`
return {
title,
output: msg,
metadata: {
preview: msg,
truncated: false,
...(instructions.length > 0 && { loaded: instructions.map((i) => i.filepath) }),
},
attachments: [
{
id: Identifier.ascending("part"),
sessionID: ctx.sessionID,
messageID: ctx.messageID,
type: "file",
mime,
url: `data:${mime};base64,${Buffer.from(fileBytes).toString("base64")}`,
},
],
}
}
const isBinary = await isBinaryFile(filepath, file)
if (isBinary) throw new Error(`Cannot read binary file: ${filepath}`)
const limit = params.limit ?? DEFAULT_READ_LIMIT
const offset = params.offset || 0
const lines = await file.text().then((text) => text.split("\n"))
const raw: string[] = []
let bytes = 0
let truncatedByBytes = false
for (let i = offset; i < Math.min(lines.length, offset + limit); i++) {
const line = lines[i].length > MAX_LINE_LENGTH ? lines[i].substring(0, MAX_LINE_LENGTH) + "..." : lines[i]
const size = Buffer.byteLength(line, "utf-8") + (raw.length > 0 ? 1 : 0)
if (bytes + size > MAX_BYTES) {
truncatedByBytes = true
break
}
raw.push(line)
bytes += size
}
const content = raw.map((line, index) => {
return `${(index + offset + 1).toString().padStart(5, "0")}| ${line}`
})
const preview = raw.slice(0, 20).join("\n")
let output = "<file>\n"
output += content.join("\n")
const totalLines = lines.length
const lastReadLine = offset + raw.length
const hasMoreLines = totalLines > lastReadLine
const truncated = hasMoreLines || truncatedByBytes
if (truncatedByBytes) {
output += `\n\n(Output truncated at ${MAX_BYTES} bytes. Use 'offset' parameter to read beyond line ${lastReadLine})`
} else if (hasMoreLines) {
output += `\n\n(File has more lines. Use 'offset' parameter to read beyond line ${lastReadLine})`
} else {
output += `\n\n(End of file - total ${totalLines} lines)`
}
output += "\n</file>"
// just warms the lsp client
LSP.touchFile(filepath, false)
FileTime.read(ctx.sessionID, filepath)
if (instructions.length > 0) {
output += `\n\n<system-reminder>\n${instructions.map((i) => i.content).join("\n\n")}\n</system-reminder>`
}
return {
title,
output,
metadata: {
preview,
truncated,
...(instructions.length > 0 && { loaded: instructions.map((i) => i.filepath) }),
},
}
},
})
async function isBinaryFile(filepath: string, file: Bun.BunFile): Promise<boolean> {
const ext = path.extname(filepath).toLowerCase()
// binary check for common non-text extensions
switch (ext) {
case ".zip":
case ".tar":
case ".gz":
case ".exe":
case ".dll":
case ".so":
case ".class":
case ".jar":
case ".war":
case ".7z":
case ".doc":
case ".docx":
case ".xls":
case ".xlsx":
case ".ppt":
case ".pptx":
case ".odt":
case ".ods":
case ".odp":
case ".bin":
case ".dat":
case ".obj":
case ".o":
case ".a":
case ".lib":
case ".wasm":
case ".pyc":
case ".pyo":
return true
default:
break
}
const stat = await file.stat()
const fileSize = stat.size
if (fileSize === 0) return false
// Only read the header, not the whole file — a 2 GB binary would OOM otherwise.
const bufferSize = Math.min(4096, fileSize)
const head = file.slice(0, bufferSize)
const buffer = await head.arrayBuffer()
if (buffer.byteLength === 0) return false
const bytes = new Uint8Array(buffer)
let nonPrintableCount = 0
for (let i = 0; i < bytes.length; i++) {
if (bytes[i] === 0) return true
if (bytes[i] < 9 || (bytes[i] > 13 && bytes[i] < 32)) {
nonPrintableCount++
}
}
// If >30% non-printable characters, consider it binary
return nonPrintableCount / bytes.length > 0.3
}