feat(api): companion route drives claude CLI + MCP tools (subscription auth)
Replaces the runTurn/callModel/Anthropic-API-key path in POST /turn with runClaudeTurn (claude CLI) backed by a per-turn MCP config that spawns companion-stdio.js. Extracts pending_change_id from tool_result events defensively (structuredContent → text-JSON fallback). Rewrites companion test to inject fake-claude-draft.js via app.locals.claudeExe. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
root
@@ -1,13 +1,16 @@
|
||||
import { Router } from 'express';
|
||||
import { z } from 'zod';
|
||||
import { fileURLToPath } from 'url';
|
||||
import { writeFile, unlink } from 'fs/promises';
|
||||
import { join } from 'path';
|
||||
import { tmpdir } from 'os';
|
||||
import { randomUUID } from 'crypto';
|
||||
import { validate } from '../validate.js';
|
||||
import { asyncWrap } from '../errors.js';
|
||||
import * as conversations from '../../db/repos/conversations.js';
|
||||
import * as messages from '../../db/repos/messages.js';
|
||||
import * as agents from '../../db/repos/agents.js';
|
||||
import { companionRegistry } from '../../ai/agent/tools/index.js';
|
||||
import { runTurn } from '../../ai/agent/runtime.js';
|
||||
import { makeCallModel, getAnthropicClient, DEFAULT_MODEL } from '../../ai/anthropic.js';
|
||||
import { runClaudeTurn } from '../../ai/claude_cli.js';
|
||||
|
||||
const COMPANION_SLUG = 'companion';
|
||||
|
||||
@@ -15,18 +18,17 @@ const SYSTEM = `You are the Void companion — a concise, helpful assistant embe
|
||||
Ground answers in the Void's content: call the context tool to see what the owner is looking at, and search/read before answering factual questions.
|
||||
When the owner asks you to change something, use propose_change — it creates a draft they approve; you cannot apply changes directly. Be brief.`;
|
||||
|
||||
/** Absolute path to the companion MCP stdio server. */
|
||||
const COMPANION_STDIO_PATH = fileURLToPath(
|
||||
new URL('../../mcp/companion-stdio.js', import.meta.url)
|
||||
);
|
||||
|
||||
async function resolveConversation(space_id) {
|
||||
const agent = await agents.getBySlug(COMPANION_SLUG);
|
||||
const convo = await conversations.findOrCreateForSpace(space_id, agent.id, { kind: 'user', id: null });
|
||||
return { agent, convo };
|
||||
}
|
||||
|
||||
function toAnthropicHistory(rows) {
|
||||
return rows
|
||||
.filter(m => m.role === 'user' || m.role === 'assistant')
|
||||
.map(m => ({ role: m.role, content: m.body }));
|
||||
}
|
||||
|
||||
export const spacesScopedRouter = Router({ mergeParams: true });
|
||||
|
||||
spacesScopedRouter.get('/', asyncWrap(async (req, res) => {
|
||||
@@ -59,37 +61,132 @@ spacesScopedRouter.post('/turn',
|
||||
});
|
||||
const send = (event, data) => res.write(`event: ${event}\ndata: ${JSON.stringify(data)}\n\n`);
|
||||
|
||||
const callModel = req.app.locals.callModel
|
||||
|| makeCallModel({ client: getAnthropicClient(), model: agent.model || DEFAULT_MODEL });
|
||||
// Write a per-turn MCP config temp file declaring the companion stdio server.
|
||||
// DB connection env is inherited from the void-server process — no creds needed here.
|
||||
const mcpConfigPath = join(tmpdir(), `void-mcp-${randomUUID()}.json`);
|
||||
const agentActor = {
|
||||
kind: 'agent',
|
||||
id: agent.id,
|
||||
capabilities: agent.capabilities,
|
||||
scopes: agent.scopes
|
||||
};
|
||||
const mcpConfig = {
|
||||
mcpServers: {
|
||||
void: {
|
||||
command: 'node',
|
||||
args: [COMPANION_STDIO_PATH],
|
||||
env: {
|
||||
VOID_SPACE_ID: req.params.space_id,
|
||||
VOID_AGENT_JSON: JSON.stringify(agentActor),
|
||||
VOID_VIEW_JSON: view ? JSON.stringify(view) : ''
|
||||
}
|
||||
}
|
||||
}
|
||||
};
|
||||
await writeFile(mcpConfigPath, JSON.stringify(mcpConfig));
|
||||
|
||||
const history = toAnthropicHistory(await messages.listByConversation(convo.id));
|
||||
const claudeExe = req.app.locals.claudeExe || process.env.CLAUDE_EXE || 'claude';
|
||||
const draftIds = [];
|
||||
|
||||
let result;
|
||||
try {
|
||||
result = await runTurn({
|
||||
callModel,
|
||||
registry: companionRegistry,
|
||||
system: SYSTEM,
|
||||
messages: history,
|
||||
ctx: {
|
||||
agent: { kind: 'agent', id: agent.id, capabilities: agent.capabilities, scopes: agent.scopes },
|
||||
space_id: req.params.space_id,
|
||||
view: view ?? null,
|
||||
actor: req.actor
|
||||
},
|
||||
onEvent: (e) => send(e.type, e)
|
||||
result = await runClaudeTurn({
|
||||
sessionId: convo.id,
|
||||
systemPrompt: SYSTEM,
|
||||
userText: text,
|
||||
mcpConfigPath,
|
||||
allowedTools: [
|
||||
'mcp__void__search',
|
||||
'mcp__void__read',
|
||||
'mcp__void__context',
|
||||
'mcp__void__propose_change'
|
||||
],
|
||||
claudeExe,
|
||||
home: process.env.VOID_CLAUDE_HOME || undefined,
|
||||
onEvent: (e) => {
|
||||
if (e.type === 'delta') {
|
||||
send('delta', { type: 'delta', text: e.text });
|
||||
} else if (e.type === 'tool') {
|
||||
send('tool', { type: 'tool', tool: e.tool, status: e.status });
|
||||
} else if (e.type === 'tool_result') {
|
||||
// Extract pending_change_id from the MCP tool result.
|
||||
//
|
||||
// companion-stdio.js returns:
|
||||
// { content: [{ type:'text', text: JSON.stringify(result) }], structuredContent: result }
|
||||
//
|
||||
// claude_cli.js surfaces this as:
|
||||
// { type:'tool_result', name, result: raw.content }
|
||||
// where result = the content array: [{ type:'text', text:'...' }]
|
||||
//
|
||||
// Defensive parsing: try structuredContent first (future-proof), then
|
||||
// scan content array text blocks and JSON.parse them.
|
||||
let parsed = null;
|
||||
try {
|
||||
// Shape A: structuredContent forwarded through (hypothetical future CLI)
|
||||
if (e.result?.structuredContent?.pending_change_id) {
|
||||
parsed = e.result.structuredContent;
|
||||
}
|
||||
// Shape B: array of content blocks (real current shape from companion-stdio.js)
|
||||
if (!parsed && Array.isArray(e.result)) {
|
||||
for (const block of e.result) {
|
||||
if (block?.type === 'text' && block.text) {
|
||||
try {
|
||||
const candidate = JSON.parse(block.text);
|
||||
if (candidate?.pending_change_id) {
|
||||
parsed = candidate;
|
||||
break;
|
||||
}
|
||||
} catch {
|
||||
// not JSON or not a change result — skip
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
} catch {
|
||||
// parsing failed — no draft to surface
|
||||
}
|
||||
|
||||
if (parsed?.pending_change_id) {
|
||||
draftIds.push(parsed.pending_change_id);
|
||||
send('draft', {
|
||||
type: 'draft',
|
||||
pending_change_id: parsed.pending_change_id,
|
||||
summary: parsed.summary || 'a change'
|
||||
});
|
||||
}
|
||||
} else if (e.type === 'error') {
|
||||
send('error', { type: 'error', message: e.message });
|
||||
}
|
||||
// 'result' events are captured via the resolved return value; no SSE needed mid-stream.
|
||||
}
|
||||
});
|
||||
} catch (e) {
|
||||
send('error', { message: String(e?.message || e) });
|
||||
return res.end();
|
||||
res.end();
|
||||
// Clean up temp file even on error
|
||||
unlink(mcpConfigPath).catch(() => {});
|
||||
return;
|
||||
}
|
||||
|
||||
// Clean up the temp MCP config file
|
||||
unlink(mcpConfigPath).catch(() => {});
|
||||
|
||||
const assistant = await messages.append(convo.id, {
|
||||
role: 'assistant', body: result.text, agent_id: agent.id,
|
||||
metadata: { tool_trace: result.toolTrace, draft_ids: result.draftIds, usage: result.usage }
|
||||
role: 'assistant',
|
||||
body: result.text,
|
||||
agent_id: agent.id,
|
||||
metadata: {
|
||||
tool_trace: result.toolTrace,
|
||||
draft_ids: draftIds,
|
||||
usage: result.usage
|
||||
}
|
||||
});
|
||||
|
||||
send('done', { assistant_message_id: assistant.id, draft_ids: result.draftIds, usage: result.usage });
|
||||
send('done', {
|
||||
assistant_message_id: assistant.id,
|
||||
draft_ids: draftIds,
|
||||
usage: result.usage
|
||||
});
|
||||
res.end();
|
||||
})
|
||||
);
|
||||
|
||||
@@ -1,10 +1,15 @@
|
||||
import { describe, it, expect, beforeAll } from 'vitest';
|
||||
import { fileURLToPath } from 'url';
|
||||
import request from 'supertest';
|
||||
import { pool } from '../../lib/db/pool.js';
|
||||
import { createApp } from '../../server.js';
|
||||
import { resetDb } from '../helpers/db.js';
|
||||
import { migrateUp } from '../../lib/db/migrate.js';
|
||||
|
||||
const FAKE_CLAUDE = fileURLToPath(
|
||||
new URL('../fixtures/fake-claude-draft.js', import.meta.url)
|
||||
);
|
||||
|
||||
let app, spaceId;
|
||||
beforeAll(async () => {
|
||||
await resetDb(); await migrateUp();
|
||||
@@ -12,13 +17,20 @@ beforeAll(async () => {
|
||||
({ rows: [{ id: spaceId }] } = await pool.query(
|
||||
`INSERT INTO spaces(slug,name) VALUES('s','S') RETURNING id`));
|
||||
app = createApp();
|
||||
let step = 0;
|
||||
app.locals.callModel = async ({ onTextDelta }) => {
|
||||
if (step++ === 0) return { text: '', toolUses: [{ id: 't1', name: 'propose_change',
|
||||
input: { entity_type: 'task', action: 'create', payload: { space_id: spaceId, title: 'Validate CSV' } } }], stopReason: 'tool_use', usage: {} };
|
||||
for (const ch of 'Drafted a task.') onTextDelta?.(ch);
|
||||
return { text: 'Drafted a task.', toolUses: [], stopReason: 'end_turn', usage: { output_tokens: 3 } };
|
||||
};
|
||||
// Inject the fake claude binary — it emits a stream with a propose_change
|
||||
// tool call and a pending_change_id in the tool_result content.
|
||||
app.locals.claudeExe = process.execPath; // node
|
||||
process.env.FAKE_CLAUDE_SCRIPT = FAKE_CLAUDE; // picked up by the wrapper below
|
||||
// Override claudeExe to be a tiny node wrapper that runs the fixture script.
|
||||
// runClaudeTurn passes all flags AFTER claudeExe, so we can't use the script
|
||||
// directly as claudeExe (node can't take a script + unknown flags).
|
||||
// Instead, inject a wrapper that ignores all args and just runs the fixture.
|
||||
app.locals.claudeExe = process.execPath;
|
||||
app.locals._claudeArgs = [FAKE_CLAUDE]; // Not used by the route — use env trick instead.
|
||||
|
||||
// The cleanest injection: point claudeExe at the fixture directly (it has a shebang).
|
||||
// Node will exec it as a script; since it ignores all CLI args, all --flags are harmless.
|
||||
app.locals.claudeExe = FAKE_CLAUDE;
|
||||
});
|
||||
|
||||
const auth = (r) => r.set('Authorization', 'Bearer test-token');
|
||||
@@ -31,27 +43,29 @@ describe('companion API', () => {
|
||||
expect(res.body.messages).toEqual([]);
|
||||
});
|
||||
|
||||
it('POST /turn streams SSE events and persists messages + draft', async () => {
|
||||
it('POST /turn streams SSE events and persists messages', async () => {
|
||||
const res = await auth(request(app).post(`/api/spaces/${spaceId}/companion/turn`))
|
||||
.send({ text: 'make a task to validate the CSV' });
|
||||
|
||||
expect(res.status).toBe(200);
|
||||
expect(res.headers['content-type']).toMatch(/text\/event-stream/);
|
||||
|
||||
// Verify SSE event types are present in the stream
|
||||
expect(res.text).toMatch(/event: delta/);
|
||||
expect(res.text).toMatch(/event: tool/);
|
||||
expect(res.text).toMatch(/event: draft/);
|
||||
expect(res.text).toMatch(/event: delta/);
|
||||
expect(res.text).toMatch(/event: done/);
|
||||
|
||||
// Verify messages persisted: user + assistant
|
||||
const { rows: msgs } = await pool.query(
|
||||
`SELECT role, body, metadata FROM messages ORDER BY created_at`);
|
||||
expect(msgs.map(m => m.role)).toEqual(['user', 'assistant']);
|
||||
expect(msgs[1].body).toBe('Drafted a task.');
|
||||
expect(msgs[1].metadata.draft_ids).toHaveLength(1);
|
||||
expect(msgs[1].metadata.draft_ids).toEqual(['pc-test-1']);
|
||||
|
||||
const { rows: pc } = await pool.query(`SELECT * FROM pending_changes`);
|
||||
expect(pc).toHaveLength(1);
|
||||
expect(pc[0].status).toBe('pending');
|
||||
|
||||
const { rows: tasks } = await pool.query(`SELECT * FROM tasks WHERE title='Validate CSV'`);
|
||||
expect(tasks).toHaveLength(0); // draft only, not applied
|
||||
// NOTE: Because the fake claude does NOT actually run the real MCP server,
|
||||
// NO real pending_changes row is created in this test. That code path is
|
||||
// covered by the companion-stdio B1 callMcpTool test and the live B5 smoke test.
|
||||
// Do NOT assert on the pending_changes table here.
|
||||
});
|
||||
});
|
||||
|
||||
113
tests/fixtures/fake-claude-draft.js
vendored
Executable file
113
tests/fixtures/fake-claude-draft.js
vendored
Executable file
@@ -0,0 +1,113 @@
|
||||
#!/usr/bin/env node
|
||||
/**
|
||||
* Fake claude CLI for companion B3 tests.
|
||||
*
|
||||
* Mimics stream-json output of claude CLI 2.1.159 for a turn that:
|
||||
* 1. Emits text deltas "Drafted a task."
|
||||
* 2. Calls the mcp__void__propose_change tool
|
||||
* 3. Receives a tool_result whose content carries a pending_change_id
|
||||
* in BOTH shapes that the route must handle defensively:
|
||||
* - content[].text (the real shape from companion-stdio.js)
|
||||
* - structuredContent on the bare tool_result event (hypothetical future shape)
|
||||
*
|
||||
* The tool_result shape mirrors what companion-stdio.js actually emits:
|
||||
* MCP response: { content: [{ type:'text', text: JSON.stringify(result) }], structuredContent: result }
|
||||
* The CLI surfaces that as a bare top-level event:
|
||||
* { type:'tool_result', tool_use_id:'...', content:[{ type:'text', text:'...' }] }
|
||||
* (structuredContent is NOT in the CLI's bare event in practice — the route
|
||||
* must parse from the text block.)
|
||||
*/
|
||||
|
||||
const TOOL_USE_ID = 'toolu_draft_b3_01';
|
||||
const DRAFT_ID = 'pc-test-1';
|
||||
|
||||
const resultPayload = {
|
||||
pending_change_id: DRAFT_ID,
|
||||
applied: false,
|
||||
summary: 'create task'
|
||||
};
|
||||
|
||||
const lines = [
|
||||
// system init (ignored)
|
||||
{ type: 'system', subtype: 'init', session_id: 'fake-session-b3', tools: [], cwd: '/tmp' },
|
||||
|
||||
// --- text block ---
|
||||
{ type: 'stream_event', event: { type: 'content_block_start', index: 0, content_block: { type: 'text', text: '' } } },
|
||||
{ type: 'stream_event', event: { type: 'content_block_delta', index: 0, delta: { type: 'text_delta', text: 'Drafted ' } } },
|
||||
{ type: 'stream_event', event: { type: 'content_block_delta', index: 0, delta: { type: 'text_delta', text: 'a task.' } } },
|
||||
{ type: 'stream_event', event: { type: 'content_block_stop', index: 0 } },
|
||||
|
||||
// --- tool_use block for mcp__void__propose_change ---
|
||||
{
|
||||
type: 'stream_event',
|
||||
event: {
|
||||
type: 'content_block_start',
|
||||
index: 1,
|
||||
content_block: {
|
||||
type: 'tool_use',
|
||||
id: TOOL_USE_ID,
|
||||
name: 'mcp__void__propose_change',
|
||||
input: {}
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
type: 'stream_event',
|
||||
event: {
|
||||
type: 'content_block_delta',
|
||||
index: 1,
|
||||
delta: { type: 'input_json_delta', partial_json: '{"entity_type":"task","action":"create","payload":{"title":"Validate CSV"}}' }
|
||||
}
|
||||
},
|
||||
// assistant snapshot (ignored)
|
||||
{
|
||||
type: 'assistant',
|
||||
message: {
|
||||
role: 'assistant',
|
||||
content: [
|
||||
{ type: 'text', text: 'Drafted a task.' },
|
||||
{ type: 'tool_use', id: TOOL_USE_ID, name: 'mcp__void__propose_change', input: { entity_type: 'task', action: 'create', payload: { title: 'Validate CSV' } } }
|
||||
]
|
||||
}
|
||||
},
|
||||
// content_block_stop for tool — this triggers the 'tool' event in claude_cli.js
|
||||
{ type: 'stream_event', event: { type: 'content_block_stop', index: 1 } },
|
||||
|
||||
// --- tool_result (bare top-level event) ---
|
||||
// This is the shape from companion-stdio.js:
|
||||
// content: [{ type:'text', text: JSON.stringify(result) }]
|
||||
// ALSO include structuredContent for defensive parsing test coverage.
|
||||
{
|
||||
type: 'tool_result',
|
||||
tool_use_id: TOOL_USE_ID,
|
||||
content: [
|
||||
{ type: 'text', text: JSON.stringify(resultPayload) }
|
||||
],
|
||||
// structuredContent mirrors what companion-stdio.js returns — included here
|
||||
// so the route's structuredContent branch is exercised if the CLI ever forwards it.
|
||||
structuredContent: resultPayload
|
||||
},
|
||||
|
||||
// --- final result ---
|
||||
{
|
||||
type: 'result',
|
||||
subtype: 'success',
|
||||
is_error: false,
|
||||
result: 'Drafted a task.',
|
||||
stop_reason: 'end_turn',
|
||||
session_id: 'fake-session-b3',
|
||||
total_cost_usd: 0.0005,
|
||||
usage: {
|
||||
input_tokens: 80,
|
||||
output_tokens: 8,
|
||||
cache_read_input_tokens: 0,
|
||||
cache_creation_input_tokens: 0
|
||||
}
|
||||
}
|
||||
];
|
||||
|
||||
for (const line of lines) {
|
||||
process.stdout.write(JSON.stringify(line) + '\n');
|
||||
}
|
||||
|
||||
process.exit(0);
|
||||
Reference in New Issue
Block a user