feat(mosaic): WI-2 mutator-class guard for directive-freshness (#837)
Some checks failed
ci/woodpecker/push/ci Pipeline failed
ci/woodpecker/push/publish Pipeline was successful

WI-2: mutator-class guard for the compaction directive-freshness mechanism — command-position parser (prefix x var-indirection unified) + primitive-anchored invariant backstop + all-tools-hook fail-close. Parser-complete on principle: realistic evasion matrix RED-regression-covered, residual exotic evasions proven backstop-caught (B-tests), lens-convergence reached.

closes #829
This commit was merged in pull request #837.
This commit is contained in:
2026-07-18 05:51:58 +00:00
parent 8ec67a1126
commit abd2791f59
27 changed files with 2579 additions and 46 deletions

View File

@@ -1,5 +1,13 @@
import { describe, it, expect, vi } from 'vitest';
import { mkdtempSync, mkdirSync, symlinkSync, rmSync, lstatSync, writeFileSync } from 'node:fs';
import {
lstatSync,
mkdtempSync,
mkdirSync,
readFileSync,
rmSync,
symlinkSync,
writeFileSync,
} from 'node:fs';
import { tmpdir, homedir } from 'node:os';
import { join } from 'node:path';
import {
@@ -14,6 +22,7 @@ import {
buildClaudexEnv,
buildClaudexBanner,
buildClaudexContractNote,
ensureClaudexMutatorGateSettings,
runClaudexProxyGate,
launchClaudex,
type ClaudexHarnessAdapter,
@@ -36,12 +45,20 @@ function makeReport(overrides: Partial<PreflightReport> = {}): PreflightReport {
};
}
function okAdapter(overrides: Partial<ClaudexHarnessAdapter> = {}): ClaudexHarnessAdapter {
type TestAdapterOverrides = Partial<ClaudexHarnessAdapter> & {
exec?: (cmd: string, args: string[], env: NodeJS.ProcessEnv) => void;
};
function okAdapter(overrides: TestAdapterOverrides = {}): ClaudexHarnessAdapter {
const { exec, ...adapterOverrides } = overrides;
return {
harnessPreflight: () => {},
composePrompt: () => '# Composed Claude contract',
exec: () => {},
...overrides,
execLeaseGated:
adapterOverrides.execLeaseGated ??
((args, env, dangerous) =>
exec?.('claude', dangerous ? ['--dangerously-skip-permissions', ...args] : args, env)),
...adapterOverrides,
};
}
@@ -561,11 +578,51 @@ describe('runClaudexProxyGate', () => {
// ─── launch orchestration (fail-closed ordering) ──────────────────────────────
describe('ensureClaudexMutatorGateSettings', () => {
it('does not accept a lookalike hook and preserves isolated settings', () => {
const root = mkdtempSync(join(tmpdir(), 'claudex-gate-settings-'));
try {
writeFileSync(
join(root, 'settings.json'),
JSON.stringify({
preserved: true,
hooks: {
PreToolUse: [
{
matcher: '.*',
hooks: [{ type: 'command', command: 'echo lease-broker/mutator-gate.py' }],
},
],
},
}),
);
ensureClaudexMutatorGateSettings(root);
const settings = JSON.parse(readFileSync(join(root, 'settings.json'), 'utf8')) as {
preserved: boolean;
hooks: { PreToolUse: Array<{ hooks: Array<{ command: string }> }> };
};
const commands = settings.hooks.PreToolUse.flatMap((entry) =>
entry.hooks.map((hook) => hook.command),
);
expect(settings.preserved).toBe(true);
expect(commands).toContain(
'python3 ~/.config/mosaic/tools/lease-broker/mutator-gate.py --runtime claude',
);
expect(lstatSync(join(root, 'settings.json')).mode & 0o777).toBe(0o600);
} finally {
rmSync(root, { recursive: true, force: true });
}
});
});
describe('launchClaudex', () => {
const baseDeps = {
baseEnv: {},
proxyGate: () => Promise.resolve({ ok: true, report: makeReport(), problems: [] }),
resolveConfigDir: () => '/home/agent/.config/mosaic/claudex/home',
prepareConfig: () => {},
log: () => {},
errorLog: () => {},
fail: (() => {

View File

@@ -27,7 +27,14 @@
* unit-testable without spawning Claude Code or touching a real config dir.
*/
import { lstatSync, mkdirSync, realpathSync } from 'node:fs';
import {
chmodSync,
lstatSync,
mkdirSync,
readFileSync,
realpathSync,
writeFileSync,
} from 'node:fs';
import { homedir } from 'node:os';
import { dirname, isAbsolute, join, relative, resolve } from 'node:path';
import {
@@ -278,6 +285,77 @@ export function buildClaudexEnv(
return env;
}
const CLAUDEX_MUTATOR_GATE_COMMAND =
'python3 ~/.config/mosaic/tools/lease-broker/mutator-gate.py --runtime claude';
const CLAUDEX_MUTATOR_GATE_HOOK = {
matcher: '.*',
hooks: [
{
type: 'command',
command: CLAUDEX_MUTATOR_GATE_COMMAND,
timeout: 3,
},
],
};
function isRecord(value: unknown): value is Record<string, unknown> {
return typeof value === 'object' && value !== null && !Array.isArray(value);
}
/**
* Install the mandatory all-tools gate in Claudex's isolated config without
* discarding any existing isolated settings. Malformed settings and symlinked
* settings files fail closed rather than launching with uncertain hook state.
*/
export function ensureClaudexMutatorGateSettings(configDir: string): void {
mkdirSync(configDir, { recursive: true, mode: 0o700 });
const settingsPath = join(configDir, 'settings.json');
let settings: Record<string, unknown> = {};
try {
if (lstatSync(settingsPath).isSymbolicLink()) {
throw new Error('claudex: isolated settings.json must not be a symlink (fail closed).');
}
const parsed: unknown = JSON.parse(readFileSync(settingsPath, 'utf8'));
if (!isRecord(parsed)) {
throw new Error('claudex: isolated settings.json must contain a JSON object (fail closed).');
}
settings = parsed;
} catch (err) {
if ((err as NodeJS.ErrnoException).code !== 'ENOENT') throw err;
}
const hooksValue = settings['hooks'];
if (hooksValue !== undefined && !isRecord(hooksValue)) {
throw new Error('claudex: isolated settings hooks must be an object (fail closed).');
}
const hooks = hooksValue ?? {};
const preToolUseValue = hooks['PreToolUse'];
if (preToolUseValue !== undefined && !Array.isArray(preToolUseValue)) {
throw new Error('claudex: isolated PreToolUse hooks must be an array (fail closed).');
}
const preToolUse = preToolUseValue ?? [];
const gatePresent = preToolUse.some(
(entry) =>
isRecord(entry) &&
entry['matcher'] === '.*' &&
Array.isArray(entry['hooks']) &&
entry['hooks'].some(
(hook) =>
isRecord(hook) &&
hook['type'] === 'command' &&
hook['command'] === CLAUDEX_MUTATOR_GATE_COMMAND,
),
);
if (!gatePresent) preToolUse.unshift(CLAUDEX_MUTATOR_GATE_HOOK);
hooks['PreToolUse'] = preToolUse;
settings['hooks'] = hooks;
writeFileSync(settingsPath, `${JSON.stringify(settings, null, 2)}\n`, { mode: 0o600 });
chmodSync(settingsPath, 0o600);
}
// ─── EXPERIMENTAL classification (P4) ─────────────────────────────────────────
/** Console banner shown at launch. Contains no token material by construction. */
@@ -396,8 +474,8 @@ export interface ClaudexHarnessAdapter {
harnessPreflight: () => void;
/** Compose the full Claude runtime contract (== `composeContract('claude')`). */
composePrompt: () => string;
/** Replace the current process with `claude` using the composed env. */
exec: (cmd: string, args: string[], env: NodeJS.ProcessEnv) => void;
/** Register the Claude parent with the broker, then exec with the same PID. */
execLeaseGated: (args: string[], env: NodeJS.ProcessEnv, dangerous: boolean) => void;
}
export interface LaunchClaudexDeps {
@@ -406,6 +484,7 @@ export interface LaunchClaudexDeps {
resolveConfigDir?: () => string;
models?: () => ClaudexModels;
buildEnv?: (base: NodeJS.ProcessEnv, opts: BuildClaudexEnvOptions) => NodeJS.ProcessEnv;
prepareConfig?: (configDir: string) => void;
log?: (message: string) => void;
errorLog?: (message: string) => void;
fail?: (code: number) => never;
@@ -431,6 +510,7 @@ export async function launchClaudex(
const resolveConfigDir = deps.resolveConfigDir ?? (() => resolveClaudexConfigDir(baseEnv));
const models = deps.models ?? (() => resolveClaudexModels(baseEnv));
const buildEnv = deps.buildEnv ?? buildClaudexEnv;
const prepareConfig = deps.prepareConfig ?? ensureClaudexMutatorGateSettings;
try {
// Harness readiness first (claude on PATH, mosaic home, sequential-thinking).
@@ -447,14 +527,14 @@ export async function launchClaudex(
// Compose the isolated launch env (guard throws → caught below, fail closed).
const resolvedModels = models();
const configDir = resolveConfigDir();
prepareConfig(configDir);
const env = buildEnv(baseEnv, { configDir, models: resolvedModels });
const prompt = `${adapter.composePrompt()}\n\n${buildClaudexContractNote(resolvedModels)}`;
log(buildClaudexBanner(resolvedModels));
const cliArgs = yolo ? ['--dangerously-skip-permissions'] : [];
cliArgs.push('--append-system-prompt', prompt, ...args);
adapter.exec('claude', cliArgs, env);
const cliArgs = ['--append-system-prompt', prompt, ...args];
adapter.execLeaseGated(cliArgs, env, yolo);
} catch (err) {
errorLog(
`[mosaic] claudex launch aborted: ${err instanceof Error ? err.message : String(err)}`,

View File

@@ -115,7 +115,7 @@ function auditClaudeSettings(): SettingsAudit {
// Check required hooks
const hooks = settings['hooks'] as Record<string, unknown[]> | undefined;
const requiredPreToolUse = ['prevent-memory-write.sh'];
const requiredPreToolUse = ['mutator-gate.py', 'prevent-memory-write.sh'];
const requiredPostToolUse = ['qa-hook-stdin.sh', 'typecheck-hook.sh'];
const preHooks = (hooks?.['PreToolUse'] ?? []) as Array<Record<string, unknown>>;
@@ -755,7 +755,7 @@ function launchRuntime(runtime: RuntimeName, args: string[], yolo: boolean): nev
printSettingsWarnings(settingsAudit);
const prompt = buildRuntimePrompt('claude');
const cliArgs = yolo ? ['--dangerously-skip-permissions'] : [];
const cliArgs: string[] = [];
cliArgs.push('--append-system-prompt', prompt);
if (hasMissionNoArgs) {
cliArgs.push(missionPrompt);
@@ -763,7 +763,7 @@ function launchRuntime(runtime: RuntimeName, args: string[], yolo: boolean): nev
cliArgs.push(...args);
}
console.log(`[mosaic] Launching ${label}${modeStr}${missionStr}...`);
execRuntime('claude', cliArgs);
execLeaseGatedRuntime('claude', cliArgs, process.env, yolo);
break;
}
@@ -798,7 +798,7 @@ function launchRuntime(runtime: RuntimeName, args: string[], yolo: boolean): nev
cliArgs.push(...args);
}
console.log(`[mosaic] Launching ${label}${modeStr}${missionStr}...`);
execRuntime('pi', cliArgs);
execLeaseGatedRuntime('pi', cliArgs);
break;
}
}
@@ -806,6 +806,33 @@ function launchRuntime(runtime: RuntimeName, args: string[], yolo: boolean): nev
process.exit(0); // Unreachable but satisfies never
}
function defaultLeaseBrokerSocket(env: NodeJS.ProcessEnv = process.env): string {
if (env['MOSAIC_LEASE_BROKER_SOCKET']) return env['MOSAIC_LEASE_BROKER_SOCKET'];
const runtimeDir = env['XDG_RUNTIME_DIR'];
if (runtimeDir) return join(runtimeDir, 'mosaic-lease', 'broker.sock');
const uid = typeof process.getuid === 'function' ? process.getuid() : 0;
return join('/run/user', String(uid), 'mosaic-lease', 'broker.sock');
}
function execLeaseGatedRuntime(
runtime: 'claude' | 'pi',
args: string[],
baseEnv: NodeJS.ProcessEnv = process.env,
dangerous = false,
): void {
const launcher = resolveTool('lease-broker', 'launch-runtime.py');
const dangerousArgs = dangerous ? ['--dangerous'] : [];
execRuntime(
'python3',
[launcher, ...dangerousArgs, '--runtime', runtime, '--', runtime, ...args],
{
...baseEnv,
MOSAIC_LEASE_BROKER_SOCKET: defaultLeaseBrokerSocket(baseEnv),
MOSAIC_RUNTIME_GENERATION: baseEnv['MOSAIC_RUNTIME_GENERATION'] ?? '1',
},
);
}
/** exec into the runtime, replacing the current process. */
function execRuntime(cmd: string, args: string[], env: NodeJS.ProcessEnv = process.env): void {
try {
@@ -839,7 +866,8 @@ function launchClaudexProduction(args: string[], yolo: boolean): void {
checkSequentialThinking('claude');
},
composePrompt: () => buildRuntimePrompt('claude'),
exec: (cmd, cmdArgs, env) => execRuntime(cmd, cmdArgs, env),
execLeaseGated: (cmdArgs, env, dangerous) =>
execLeaseGatedRuntime('claude', cmdArgs, env, dangerous),
};
void launchClaudex(args, yolo, adapter);
}