indusagi-coding-agent 0.2.8 → 0.2.10

package/src/channels/session-ops.ts

@@ -0,0 +1,81 @@
+/**
+ * Session ops — the concrete operation registry that drives the link.
+ */
+import type { ConductorState, LinkSnapshot, Op, ThinkingLevel } from "./contract.js";
+import { defineOps } from "./contract.js";
+
+const DEFAULT_THINKING: ThinkingLevel = "off";
+
+export function projectSnapshot(
+  state: ConductorState,
+  extra: {
+    readonly thinking?: ThinkingLevel;
+    readonly sessionFile?: string;
+    readonly autoCondense?: boolean;
+    readonly messageCount?: number;
+    readonly queuedCount?: number;
+  } = {},
+): LinkSnapshot {
+  const base: LinkSnapshot = {
+    model: state.modelId,
+    thinking: extra.thinking ?? DEFAULT_THINKING,
+    streaming: state.phase === "streaming" || state.phase === "tooling",
+    condensing: state.phase === "condensing",
+    faulted: state.phase === "faulted",
+    sessionId: state.head.sessionId,
+    autoCondense: extra.autoCondense ?? true,
+    messageCount: extra.messageCount ?? 0,
+    queuedCount: extra.queuedCount ?? 0,
+    usage: state.usage,
+  };
+  return extra.sessionFile === undefined ? base : { ...base, sessionFile: extra.sessionFile };
+}
+
+export interface SubmitParams {
+  readonly input: string;
+}
+
+export interface ResumeParams {
+  readonly sessionId: string;
+}
+
+export interface CycleModelParams {
+  readonly modelId: string;
+}
+
+export interface ModelEntry {
+  readonly id: string;
+  readonly active: boolean;
+}
+
+function op<P, R>(method: string, handle: (params: P, ctx: import("./contract.js").ChannelContext) => Promise<R>): Op<P, R> {
+  return { method, handle };
+}
+
+export const SESSION_OPS = defineOps({
+  submit: op("submit", async (params: SubmitParams, ctx) => {
+    const state = await ctx.conductor.submit(params.input);
+    return projectSnapshot(state);
+  }),
+  abort: op("abort", async (_params, ctx) => {
+    ctx.conductor.abort();
+    return projectSnapshot(ctx.conductor.snapshot());
+  }),
+  snapshot: op("snapshot", async (_params, ctx) => {
+    return projectSnapshot(ctx.conductor.snapshot());
+  }),
+  resume: op("resume", async (params: ResumeParams, ctx) => {
+    await ctx.conductor.resume(params.sessionId);
+    return projectSnapshot(ctx.conductor.snapshot());
+  }),
+  listModels: op("listModels", async (_params, ctx) => {
+    const active = ctx.conductor.snapshot().modelId;
+    return [{ id: active, active: true }];
+  }),
+  cycleModel: op("cycleModel", async (params: CycleModelParams, ctx) => {
+    ctx.conductor.cycleModel?.(params.modelId);
+    return projectSnapshot(ctx.conductor.snapshot());
+  }),
+});
+
+export type SessionOps = typeof SESSION_OPS;

package/src/conductor/bash-guard.ts

@@ -0,0 +1,599 @@
+/**
+ * Bash command parser/classifier + catastrophic-command blocklist.
+ *
+ * The permission engine ({@link "./permissions"}) matches a tool call against
+ * induscode-style `Bash(...)` allow/ask/deny rules. For most tools the call is a
+ * single atomic action, so a rule's argument specifier matches against one
+ * subject string. The shell tool is different: `a && b; c | d` is FOUR commands
+ * in one call, and a coarse "match the whole command string" check is trivially
+ * bypassed — a benign-looking prefix can smuggle a destructive tail past an
+ * allow rule (`npm test && rm -rf /`), or hide a blocked binary behind a wrapper
+ * (`env X=1 rm -rf /`, `sudo rm -rf /`) or quoting.
+ *
+ * This module makes bash matching real. It is split in two concerns:
+ *
+ *   1. **A parser/classifier** — {@link parseBashCommand} splits a shell command
+ *      string into its constituent *simple* commands across the control
+ *      operators (`;`, `&&`, `||`, `|`, newlines) and `( … )` subshells /
+ *      `$( … )` / `` ` … ` `` command substitutions, strips benign leading
+ *      wrappers (`env VAR=x`, `nice`, `nohup`, `timeout N`, `xargs`, `command`,
+ *      `builtin`, and `sudo`/`doas` — the last recorded as an *elevation* flag),
+ *      and returns each base command + its raw argument string. Quoting is
+ *      handled well enough that `"rm -rf /"` cannot hide behind quotes or a
+ *      wrapper. The split is quote-aware so an operator *inside* a quoted string
+ *      (`echo "a && b"`) is NOT treated as a command boundary.
+ *
+ *   2. **A catastrophic blocklist** — {@link catastrophicReason} matches a
+ *      single parsed sub-command against a conservative set of always-deny
+ *      patterns (root-targeting `rm -rf`, fork bombs, `curl … | sh` of remote
+ *      scripts, `dd of=/dev/…`, `mkfs`, `chmod -R 777 /`, redirect-to-raw-disk).
+ *      It is deliberately narrow so normal dev commands (`rm -rf node_modules`,
+ *      `rm -rf ./dist`, `curl … -o file`) are NOT blocked. This is
+ *      defense-in-depth, NOT a hard security boundary — the OS sandbox and the
+ *      permission gate are the real boundary.
+ *
+ * Everything here is pure, synchronous, and side-effect free so the classifier
+ * and the blocklist are exhaustively unit-testable, and the permission engine
+ * can fold them into its decision without any I/O.
+ */
+
+/**
+ * One simple command extracted from a (possibly compound) shell string.
+ *
+ * `rm -rf /tmp` parses to `{ name: "rm", args: ["-rf", "/tmp"], elevated: false,
+ * raw: "rm -rf /tmp" }`. The `name` is the base binary AFTER benign wrappers are
+ * stripped; `args` are its tokens (quotes removed); `raw` is the original
+ * sub-command slice (wrappers included) so a caller can echo exactly what the
+ * model asked for; `elevated` is true when a `sudo`/`doas` wrapper was stripped.
+ */
+export interface ParsedCommand {
+  /** The base binary, lower-cased for stable comparison (e.g. `"rm"`). */
+  readonly name: string
+  /** The argument tokens after the binary, with surrounding quotes removed. */
+  readonly args: readonly string[]
+  /** Whether a `sudo`/`doas` elevation wrapper preceded this command. */
+  readonly elevated: boolean
+  /** The original sub-command slice (wrappers included), trimmed. */
+  readonly raw: string
+}
+
+const OPERATORS = ["&&", "||", ";", "|", "&", "\n"]
+
+function splitTopLevel(command: string): string[] {
+  const slices: string[] = []
+  let current = ""
+  let quote: string | undefined
+  let i = 0
+  const flush = () => {
+    const trimmed = current.trim()
+    if (trimmed.length > 0) slices.push(trimmed)
+    current = ""
+  }
+  while (i < command.length) {
+    const ch = command[i]
+    if (ch === "\\" && quote !== "'") {
+      current += ch
+      if (i + 1 < command.length) {
+        current += command[i + 1]
+        i += 2
+        continue
+      }
+      i += 1
+      continue
+    }
+    if (quote !== void 0) {
+      current += ch
+      if (ch === quote) quote = void 0
+      i += 1
+      continue
+    }
+    if (ch === '"' || ch === "'") {
+      quote = ch
+      current += ch
+      i += 1
+      continue
+    }
+    let matched: string | undefined
+    for (const op of OPERATORS) {
+      if (command.startsWith(op, i)) {
+        matched = op
+        break
+      }
+    }
+    if (matched !== void 0) {
+      flush()
+      i += matched.length
+      continue
+    }
+    current += ch
+    i += 1
+  }
+  flush()
+  return slices
+}
+
+function findParenClose(s: string, start: number): number {
+  let depth = 1
+  let quote: string | undefined
+  let i = start
+  while (i < s.length) {
+    const ch = s[i]
+    if (ch === "\\" && quote !== "'") {
+      i += 2
+      continue
+    }
+    if (quote !== void 0) {
+      if (ch === quote) quote = void 0
+      i += 1
+      continue
+    }
+    if (ch === '"' || ch === "'") {
+      quote = ch
+      i += 1
+      continue
+    }
+    if (ch === "(" || (ch === "$" && s[i + 1] === "(")) depth += 1
+    else if (ch === ")") {
+      depth -= 1
+      if (depth === 0) return i
+    }
+    i += 1
+  }
+  return s.length
+}
+
+function findBacktickClose(s: string, start: number): number {
+  let i = start
+  while (i < s.length) {
+    if (s[i] === "\\") {
+      i += 2
+      continue
+    }
+    if (s[i] === "`") return i
+    i += 1
+  }
+  return s.length
+}
+
+function extractSubshells(slice: string): { outer: string; inner: string[] } {
+  const inner: string[] = []
+  let outer = ""
+  let quote: string | undefined
+  let i = 0
+  while (i < slice.length) {
+    const ch = slice[i]
+    if (ch === "\\" && quote !== "'") {
+      outer += ch
+      if (i + 1 < slice.length) {
+        outer += slice[i + 1]
+        i += 2
+        continue
+      }
+      i += 1
+      continue
+    }
+    if (quote !== void 0) {
+      if (quote === '"' && ch === "`") {
+        const close = findBacktickClose(slice, i + 1)
+        inner.push(slice.slice(i + 1, close))
+        i = close + 1
+        continue
+      }
+      outer += ch
+      if (ch === quote) quote = void 0
+      i += 1
+      continue
+    }
+    if (ch === '"' || ch === "'") {
+      quote = ch
+      outer += ch
+      i += 1
+      continue
+    }
+    if (ch === "(" || (ch === "$" && slice[i + 1] === "(")) {
+      const start = ch === "$" ? i + 2 : i + 1
+      const close = findParenClose(slice, start)
+      inner.push(slice.slice(start, close))
+      i = close + 1
+      continue
+    }
+    if (ch === "`") {
+      const close = findBacktickClose(slice, i + 1)
+      inner.push(slice.slice(i + 1, close))
+      i = close + 1
+      continue
+    }
+    outer += ch
+    i += 1
+  }
+  return { outer: outer.trim(), inner }
+}
+
+function tokenizeArgv(slice: string): string[] {
+  const tokens: string[] = []
+  let current = ""
+  let started = false
+  let quote: string | undefined
+  let i = 0
+  const flush = () => {
+    if (started) tokens.push(current)
+    current = ""
+    started = false
+  }
+  while (i < slice.length) {
+    const ch = slice[i]
+    if (ch === "\\" && quote !== "'") {
+      if (i + 1 < slice.length) {
+        current += slice[i + 1]
+        started = true
+        i += 2
+        continue
+      }
+      i += 1
+      continue
+    }
+    if (quote !== void 0) {
+      if (ch === quote) {
+        quote = void 0
+      } else {
+        current += ch
+      }
+      started = true
+      i += 1
+      continue
+    }
+    if (ch === '"' || ch === "'") {
+      quote = ch
+      started = true
+      i += 1
+      continue
+    }
+    if (ch === " " || ch === "\t") {
+      flush()
+      i += 1
+      continue
+    }
+    current += ch
+    started = true
+    i += 1
+  }
+  flush()
+  return tokens
+}
+
+const WRAPPERS = new Set([
+  "env",
+  "nice",
+  "nohup",
+  "timeout",
+  "xargs",
+  "command",
+  "builtin",
+  "time",
+  "stdbuf",
+  "setsid",
+  "ionice",
+])
+
+const ELEVATION = new Set(["sudo", "doas"])
+
+const FLAG_TAKES_VALUE = new Set([
+  "-n",
+  "-u",
+  "-g",
+  "-s",
+  "--kill-after",
+  "--signal",
+  "--user",
+])
+
+function consumeWrapperOptions(rest: string[], wrapper: string): void {
+  for (;;) {
+    const head = rest[0]
+    if (head === void 0) return
+    if (head === "--") {
+      rest.shift()
+      return
+    }
+    if (head.startsWith("-")) {
+      rest.shift()
+      if (!head.includes("=") && FLAG_TAKES_VALUE.has(head)) {
+        const value = rest[0]
+        if (value !== void 0 && !value.startsWith("-")) rest.shift()
+      }
+      continue
+    }
+    if ((wrapper === "timeout" || wrapper === "nice") && /^\d+[a-z]?$/i.test(head)) {
+      rest.shift()
+      continue
+    }
+    return
+  }
+}
+
+function stripWrappers(argv: string[]): { rest: string[]; elevated: boolean } {
+  const rest = [...argv]
+  let elevated = false
+  for (;;) {
+    const head = rest[0]
+    if (head === void 0) break
+    const token = head.toLowerCase()
+    if (ELEVATION.has(token)) {
+      rest.shift()
+      elevated = true
+      consumeWrapperOptions(rest, token)
+      continue
+    }
+    if (WRAPPERS.has(token)) {
+      rest.shift()
+      consumeWrapperOptions(rest, token)
+      continue
+    }
+    if (/^[A-Za-z_][A-Za-z0-9_]*=/.test(head)) {
+      rest.shift()
+      continue
+    }
+    break
+  }
+  return { rest, elevated }
+}
+
+/** Strip wrappers/elevation prefixes from a tokenized argv (read-only variant). */
+export function stripSafeWrappers(argv: readonly string[]): { rest: string[]; elevated: boolean } {
+  return stripWrappers([...argv])
+}
+
+/**
+ * Strip every leading `NAME=value` environment assignment from a token list,
+ * leaving the actual command name (and any non-env-prefix tokens) untouched.
+ */
+export function stripAllLeadingEnvVars(argv: readonly string[]): string[] {
+  const rest = [...argv]
+  while (rest.length > 0 && /^[A-Za-z_][A-Za-z0-9_]*=/.test(rest[0])) {
+    rest.shift()
+  }
+  return rest
+}
+
+function parseInto(command: string, out: ParsedCommand[]): void {
+  for (const slice of splitTopLevel(command)) {
+    const { outer, inner } = extractSubshells(slice)
+    if (outer.length > 0) {
+      const argv = tokenizeArgv(outer)
+      if (argv.length > 0) {
+        const { rest, elevated } = stripWrappers(argv)
+        const name = rest[0]
+        if (name !== void 0 && name.length > 0) {
+          out.push({
+            name: name.toLowerCase(),
+            args: rest.slice(1),
+            elevated,
+            raw: slice.trim(),
+          })
+        }
+      }
+    }
+    for (const body of inner) {
+      if (body.trim().length > 0) parseInto(body, out)
+    }
+  }
+}
+
+/**
+ * Parse a (possibly compound) shell command string into the flat list of every
+ * simple command it runs — across operators, subshells, and command
+ * substitutions, with benign wrappers stripped.
+ *
+ * The result is order-preserving and exhaustive enough that the permission
+ * engine can evaluate a `Bash(...)` rule (and the catastrophic blocklist)
+ * against EACH sub-command independently: a deny on any sub-command denies the
+ * whole call, and an allow must cover every sub-command to auto-allow.
+ *
+ * @returns one {@link ParsedCommand} per simple command; an empty array for an
+ * empty/whitespace command string.
+ */
+export function parseBashCommand(command: string): ParsedCommand[] {
+  if (typeof command !== "string" || command.trim().length === 0) return []
+  const out: ParsedCommand[] = []
+  parseInto(command, out)
+  return out
+}
+
+function catastrophicMessage(reason: string): string {
+  return `This command is blocked: ${reason}. It is on the built-in catastrophic-command blocklist and cannot be run regardless of permission rules.`
+}
+
+function normPath(token: string): string {
+  const collapsed = token.replace(/\/+/g, "/")
+  if (collapsed.length > 1 && collapsed.endsWith("/")) return collapsed.slice(0, -1)
+  return collapsed
+}
+
+const ROOT_TARGETS = new Set([
+  "/",
+  "/*",
+  "~",
+  "~/",
+  "$home",
+  "$home/",
+  "/bin",
+  "/boot",
+  "/dev",
+  "/etc",
+  "/home",
+  "/lib",
+  "/opt",
+  "/proc",
+  "/root",
+  "/sbin",
+  "/srv",
+  "/sys",
+  "/usr",
+  "/var",
+])
+
+function isRootRm(cmd: ParsedCommand): boolean {
+  if (cmd.name !== "rm") return false
+  let recursive = false
+  const targets: string[] = []
+  for (const arg of cmd.args) {
+    if (arg.startsWith("--")) {
+      if (arg === "--recursive") recursive = true
+      continue
+    }
+    if (arg.startsWith("-") && arg.length > 1) {
+      if (arg.includes("r") || arg.includes("R")) recursive = true
+      continue
+    }
+    targets.push(normPath(arg.toLowerCase()))
+  }
+  if (!recursive) return false
+  return targets.some(
+    (t) => ROOT_TARGETS.has(t) || t === "/.." || /^\/[a-z]+\/\*$/.test(t),
+  )
+}
+
+const SHELL_SINKS = new Set(["sh", "bash", "zsh", "dash", "ksh", "fish", "csh", "tcsh"])
+const FETCHERS = new Set(["curl", "wget", "fetch"])
+
+function isRemoteScriptPipedToShell(cmd: ParsedCommand): boolean {
+  if (!SHELL_SINKS.has(cmd.name)) return false
+  return /\b(curl|wget|fetch)\b/i.test(cmd.raw) && /https?:\/\//i.test(cmd.raw)
+}
+
+function isRootChmod(cmd: ParsedCommand): boolean {
+  if (cmd.name !== "chmod") return false
+  let recursive = false
+  let world = false
+  const targets: string[] = []
+  for (const arg of cmd.args) {
+    if (arg === "-R" || arg === "--recursive" || /^-[a-z]*R/i.test(arg)) recursive = true
+    else if (/^[0-7]?777$/.test(arg) || arg.includes("o+w") || arg.includes("a+w")) world = true
+    else if (!arg.startsWith("-")) targets.push(normPath(arg.toLowerCase()))
+  }
+  return recursive && world && targets.some((t) => ROOT_TARGETS.has(t))
+}
+
+interface CatastrophicRule {
+  reason: string
+  test: (cmd: ParsedCommand) => boolean
+}
+
+const FORK_BOMB = /:\s*\(\s*\)\s*\{\s*:\s*\|\s*:\s*&\s*\}\s*;\s*:/;
+
+const CATASTROPHIC_RULES: CatastrophicRule[] = [
+  {
+    reason: "recursive force-delete of a root or system directory (rm -rf /)",
+    test: isRootRm,
+  },
+  {
+    reason: "piping a downloaded remote script straight into a shell (curl | sh)",
+    test: isRemoteScriptPipedToShell,
+  },
+  {
+    reason: "writing raw bytes to a block device (dd of=/dev/...)",
+    test: (cmd) =>
+      cmd.name === "dd" &&
+      cmd.args.some((a) => /^of=\/dev\/(sd|nvme|disk|hd|mmcblk|vd)/i.test(a)),
+  },
+  {
+    reason: "formatting a filesystem (mkfs)",
+    test: (cmd) => /^mkfs(\.|$)/.test(cmd.name),
+  },
+  {
+    reason: "redirecting output over a raw disk device (> /dev/sd...)",
+    test: (cmd) => /[>]\s*\/dev\/(sd|nvme|disk|hd|mmcblk|vd)/i.test(cmd.raw),
+  },
+  {
+    reason: "world-writable recursive chmod of a root or system directory (chmod -R 777 /)",
+    test: isRootChmod,
+  },
+]
+
+function scanForRemotePipeToShell(cmds: ParsedCommand[]): string | undefined {
+  const hasRemoteFetch = cmds.some(
+    (c) => FETCHERS.has(c.name) && c.args.some((a) => /^https?:\/\//i.test(a)),
+  )
+  if (!hasRemoteFetch) return void 0
+  const hasShellSink = cmds.some(
+    (c) =>
+      SHELL_SINKS.has(c.name) &&
+      !c.args.some((a) => !a.startsWith("-") && /\.(sh|bash)$/i.test(a)),
+  )
+  if (hasShellSink) {
+    return "piping a downloaded remote script straight into a shell (curl | sh)"
+  }
+  return void 0
+}
+
+/**
+ * Classify a SINGLE parsed sub-command against the catastrophic blocklist.
+ *
+ * @returns the byte-stable deny reason when the command is catastrophic, else
+ * `undefined`. Use {@link evaluateCatastrophic} for a whole compound call (it
+ * also runs the cross-sub-command `curl | sh` correlation).
+ */
+export function catastrophicReason(cmd: ParsedCommand): string | undefined {
+  for (const rule of CATASTROPHIC_RULES) {
+    if (rule.test(cmd)) return rule.reason
+  }
+  return void 0
+}
+
+/** Convenience boolean form of {@link catastrophicReason}. */
+export function isCatastrophicCommand(cmd: ParsedCommand): boolean {
+  return catastrophicReason(cmd) !== void 0
+}
+
+/**
+ * Evaluate the catastrophic blocklist over a WHOLE bash command string.
+ *
+ * Parses the (possibly compound) command, runs the per-sub-command blocklist on
+ * each leaf, AND runs the cross-sub-command `curl | sh` correlation. Returns a
+ * byte-stable deny message when ANY check fires, else `undefined`.
+ *
+ * This is what the permission engine calls for the bash tool: a catastrophic
+ * command is denied REGARDLESS of allow/ask/deny rules or mode (it is not even
+ * `bypass`-able, mirroring the original always-block guard).
+ */
+export function evaluateCatastrophic(command: string): string | undefined {
+  if (FORK_BOMB.test(command)) {
+    return catastrophicMessage("fork bomb")
+  }
+  const cmds = parseBashCommand(command)
+  if (cmds.length === 0) return void 0
+  for (const cmd of cmds) {
+    const reason = catastrophicReason(cmd)
+    if (reason !== void 0) return catastrophicMessage(reason)
+  }
+  const piped = scanForRemotePipeToShell(cmds)
+  if (piped !== void 0) return catastrophicMessage(piped)
+  return void 0
+}
+
+/**
+ * For the permission engine: emit the per-sub-command subject strings (base
+ * command + args, joined) used to test a `Bash(...)` rule specifier against an
+ * individual call.
+ */
+export function commandSubcommandSubjects(command: string): string[] {
+  return bashSubcommandSubjects(command)
+}
+
+/**
+ * Render the rule-matching SUBJECT for each sub-command of a bash call: the base
+ * command joined with its argument tokens (wrappers stripped). The permission
+ * engine matches an induscode `Bash(...)` specifier against EACH of these — an
+ * allow rule must cover them all to auto-allow, a deny on any one denies the
+ * whole call.
+ *
+ * For `npm test && rm important.txt` this yields `["npm test", "rm
+ * important.txt"]`, so `Bash(npm test)` allows the first but not the second.
+ * Returns the original command (single element) when nothing parses, so a tool
+ * call always has at least one subject to match against.
+ */
+export function bashSubcommandSubjects(command: string): string[] {
+  const cmds = parseBashCommand(command)
+  if (cmds.length === 0) return [command]
+  return cmds.map((c) => [c.name, ...c.args].join(" ").trim())
+}