Skip to content

nanocoai/nanoclaw / .claude/skills/add-opencode

add-opencode

Verified

Use OpenCode as an agent provider. OpenRouter, OpenAI, Google, DeepSeek, etc. via OpenCode config — not the Anthropic Agent SDK. Per group via `ncl groups config update --provider opencode`; host passes OPENCODE_* and XDG mount when spawning containers.

FileKindSize
ARCHITECTURE.mdmarkdown8.0 KB
REMOVE.mdmarkdown3.4 KB
SKILL.mdskill md19.8 KB
apply-fixtures.jsonconfig303 B
payload/container/agent-runner/src/provider-contracts/opencode.tsscript866 B
payload/container/agent-runner/src/providers/mcp-to-opencode.test.tsscript2.7 KB
payload/container/agent-runner/src/providers/mcp-to-opencode.tsscript1.6 KB
payload/container/agent-runner/src/providers/opencode-auth.test.tsscript3.1 KB
payload/container/agent-runner/src/providers/opencode-auth.tsscript1.7 KB
payload/container/agent-runner/src/providers/opencode-config.tsscript10.0 KB
payload/container/agent-runner/src/providers/opencode-memory.tsscript4.4 KB
payload/container/agent-runner/src/providers/opencode-registration.test.tsscript625 B
payload/container/agent-runner/src/providers/opencode-turn.tsscript14.1 KB
payload/container/agent-runner/src/providers/opencode.config.test.tsscript16.2 KB
payload/container/agent-runner/src/providers/opencode.conformance.test.tsscript253 B
payload/container/agent-runner/src/providers/opencode.empty-resume.test.tsscript14.7 KB
payload/container/agent-runner/src/providers/opencode.factory.test.tsscript331 B
payload/container/agent-runner/src/providers/opencode.memory.test.tsscript6.9 KB
payload/container/agent-runner/src/providers/opencode.native.test.tsscript17.6 KB
payload/container/agent-runner/src/providers/opencode.question.test.tsscript7.6 KB
payload/container/agent-runner/src/providers/opencode.shared-runtime.test.tsscript30.9 KB
payload/container/agent-runner/src/providers/opencode.sse-cleanup.test.tsscript6.4 KB
payload/container/agent-runner/src/providers/opencode.tsscript24.9 KB
payload/scripts/opencode-auth-config.test.tsscript27.1 KB
payload/scripts/opencode-auth.test.tsscript19.7 KB
payload/scripts/opencode-auth.tsscript21.4 KB
payload/scripts/opencode-gateway.test.tsscript7.1 KB
payload/scripts/opencode-host.test.tsscript13.4 KB
payload/scripts/opencode-host.tsscript8.9 KB
payload/scripts/opencode-model-config.tsscript8.6 KB
payload/scripts/opencode-models.test.tsscript10.6 KB
payload/scripts/opencode-models.tsscript4.0 KB
payload/scripts/opencode-vault.test.tsscript3.3 KB
payload/scripts/opencode-vault.tsscript2.7 KB
payload/scripts/tsconfig.opencode-auth.jsonconfig171 B
payload/setup/providers/opencode.test.tsscript1.2 KB
payload/setup/providers/opencode.tsscript806 B
payload/src/provider-contracts/opencode.tsscript2.9 KB
payload/src/providers/opencode-auth-stub.tsscript909 B
payload/src/providers/opencode-registration.test.tsscript8.5 KB
payload/src/providers/opencode.tsscript2.7 KB

payload/container/agent-runner/src/providers/opencode-memory.ts

View on GitHub
import { spawn } from 'child_process';
import { mkdirSync, renameSync, writeFileSync } from 'fs';
import { homedir } from 'os';
import path from 'path';
import { randomUUID } from 'crypto';

function log(message: string): void {
  console.error(`[opencode-memory] ${message}`);
}

export function openCodeInstructionsPath(): string {
  return path.resolve(process.env.XDG_DATA_HOME || path.join(homedir(), '.local', 'share'), 'nanoclaw-instructions.md');
}

/** Refresh under the turn lock. Native steps and Task children reread this file. */
export async function prepareOpenCodeMemory(
  hook: OpenCodeMemorySessionHook,
  instructions: string | undefined,
  reminder: string,
  file = openCodeInstructionsPath(),
): Promise<void> {
  const content = [await runMemorySessionHook(hook, 'startup'), instructions, reminder].filter(Boolean).join('\n\n');
  mkdirSync(path.dirname(file), { recursive: true, mode: 0o700 });
  const temporary = `${file}.${randomUUID()}.tmp`;
  writeFileSync(temporary, content, { mode: 0o600, flag: 'wx' });
  renameSync(temporary, file);
}

export interface OpenCodeMemorySessionHook {
  readonly command: string;
  readonly legacyCommands: readonly string[];
  readonly sources: readonly string[];
}

/** Sources understood by the shared renderer; turn preparation uses startup. */
export type OpenCodeMemorySource = 'startup' | 'compact';

/** Matches the `timeout: 10` (seconds) the Claude provider registers for the same command. */
const MEMORY_HOOK_TIMEOUT_MS = 10_000;

/**
 * Run one hook command with the payload on stdin; resolve with its exit status
 * and stdout. Async on purpose: under Bun, spawnSync can miss the child's exit
 * and spin forever, past every timeout we own (oven-sh/bun#34069).
 */
export function runHookCommand(
  command: string,
  input: string,
  timeoutMs = MEMORY_HOOK_TIMEOUT_MS,
): Promise<{ status: number | null; stdout: string; error?: Error }> {
  return new Promise((resolve) => {
    const stdout: Buffer[] = [];
    let settled = false;
    const settle = (result: { status: number | null; error?: Error }) => {
      if (settled) return;
      settled = true;
      clearTimeout(deadline);
      resolve({ ...result, stdout: Buffer.concat(stdout).toString('utf-8') });
    };
    const child = spawn(command, { shell: true, stdio: ['pipe', 'pipe', 'ignore'] });
    const deadline = setTimeout(() => {
      // Settle now rather than on `close`: a hook that ignores SIGTERM, or
      // leaves a background child holding stdout, must not hold the turn lock.
      child.kill('SIGTERM');
      child.stdin.destroy();
      child.stdout.destroy();
      child.unref();
      settle({ status: null, error: new Error(`timed out after ${timeoutMs}ms`) });
    }, timeoutMs);
    child.stdout.on('data', (chunk: Buffer) => stdout.push(chunk));
    child.on('error', (error) => settle({ status: null, error }));
    child.on('close', (status, signal) =>
      settle(signal ? { status, error: new Error(`killed by ${signal}`) } : { status }),
    );
    // A hook that never reads stdin (or exits early) closes the pipe under us.
    child.stdin.on('error', () => {});
    child.stdin.end(input);
  });
}

/** Run the registered renderer without duplicating its memory caps. Failures
 * log and return undefined; successful empty output remains distinguishable.
 */
export async function runMemorySessionHook(
  hook: OpenCodeMemorySessionHook | undefined,
  source: OpenCodeMemorySource,
): Promise<string | undefined> {
  if (!hook) {
    log(`No memory session hook registered; skipping ${source} memory injection`);
    return undefined;
  }
  if (!hook.sources.includes(source)) {
    log(`Memory session hook does not declare source ${source}; skipping injection`);
    return undefined;
  }

  try {
    const res = await runHookCommand(hook.command, JSON.stringify({ hook_event_name: 'SessionStart', source }));
    if (res.error || res.status !== 0) {
      const why = res.error ? res.error.message : `exit ${String(res.status)}`;
      log(`Memory session hook (${source}) failed (${why}); continuing without memory`);
      return undefined;
    }
    const out = res.stdout.trim();
    if (!out) {
      log(`Memory session hook (${source}) produced no output; continuing without memory`);
      return '';
    }
    return out;
  } catch (err) {
    log(`Memory session hook (${source}) failed: ${err instanceof Error ? err.message : String(err)}`);
    return undefined;
  }
}
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall