mirror of
https://github.com/open-multi-agent/open-multi-agent.git
synced 2026-10-19 01:31:07 -05:00
Jack Chen
ffb2905017
* feat(agent): orchestrate external coding agents over ACP (#204) Add an AgentBackend seam so an Agent can run on either the LLM AgentRunner or an external agent. First backend: an ACP client that drives a coding CLI (e.g. Claude Code via its ACP adapter) as a first-class team member — same task DAG, shared memory, cascade, and token budget as LLM agents. - Extract AgentBackend ({ run, stream }) that AgentRunner already implements; Agent.getBackend() branches on AgentConfig.backend. - AcpBackend spawns the agent over stdio JSON-RPC and maps session updates to a RunResult (text, tool calls, usage); handles permissions and abort. - ACP SDK as a lazy optional peer; new /acp subpath export. - docs/external-agents.md, a hybrid-team example, and tests (in-process fake agent + subprocess integration). Experimental: ACP is young; token accounting maps ACP's single context figure and client-side fs proxying is not implemented yet. * fix(agent): harden ACP backend after review (#204) Follow-up fixes on the ACP external-agent backend (PR #360): - Token accounting: usage_update.used is a cumulative context-size gauge, not a per-turn delta. Record each turn's increment so summing across a reused session telescopes to the latest figure instead of double-counting. - systemPrompt: ACP has no system-prompt field, so prepend the agent's systemPrompt to the session's first turn (once). It was previously dropped and only fed coordinator routing. - Abort: when the signal is already aborted before a turn starts, return empty output without prompting (no in-flight turn to cancel). - Startup: clear the cached start() promise on failure so a later run retries instead of re-throwing the same rejection forever. - Permissions: auto-approve now prefers least-privilege allow_once over a session-wide allow_always. Docs (external-agents.md, CLAUDE.md) updated to match. ACP tests 10 to 15; lint + full suite (1135) green.
89 lines
2.7 KiB
TypeScript
89 lines
2.7 KiB
TypeScript
import { describe, it, expect } from 'vitest'
|
|
import { z } from 'zod'
|
|
import { Agent } from '../src/agent/agent.js'
|
|
import { AgentRunner } from '../src/agent/runner.js'
|
|
import { ToolRegistry } from '../src/tool/framework.js'
|
|
import { ToolExecutor } from '../src/tool/executor.js'
|
|
import type { AgentConfig, LLMAdapter, LLMResponse } from '../src/types.js'
|
|
|
|
const ReviewFindings = z.array(
|
|
z.object({
|
|
priority: z.enum(['critical', 'high', 'medium', 'low']),
|
|
category: z.enum(['security', 'performance', 'style']),
|
|
issue: z.string(),
|
|
fix_hint: z.string(),
|
|
}),
|
|
)
|
|
|
|
function mockAdapter(responses: string[]): LLMAdapter {
|
|
let callIndex = 0
|
|
return {
|
|
name: 'mock',
|
|
async chat() {
|
|
const text = responses[callIndex++] ?? ''
|
|
return {
|
|
id: `mock-${callIndex}`,
|
|
content: [{ type: 'text' as const, text }],
|
|
model: 'mock-model',
|
|
stop_reason: 'end_turn',
|
|
usage: { input_tokens: 10, output_tokens: 20 },
|
|
} satisfies LLMResponse
|
|
},
|
|
async *stream() {
|
|
/* unused in this test */
|
|
},
|
|
}
|
|
}
|
|
|
|
function buildMockAgent(config: AgentConfig, responses: string[]): Agent {
|
|
const adapter = mockAdapter(responses)
|
|
const registry = new ToolRegistry()
|
|
const executor = new ToolExecutor(registry)
|
|
const agent = new Agent(config, registry, executor)
|
|
|
|
const runner = new AgentRunner(adapter, registry, executor, {
|
|
model: config.model,
|
|
systemPrompt: config.systemPrompt,
|
|
maxTurns: config.maxTurns,
|
|
maxTokens: config.maxTokens,
|
|
temperature: config.temperature,
|
|
agentName: config.name,
|
|
})
|
|
;(agent as any).backend = runner
|
|
|
|
return agent
|
|
}
|
|
|
|
describe('multi-perspective code review example', () => {
|
|
it('returns structured findings that match the issue schema', async () => {
|
|
const config: AgentConfig = {
|
|
name: 'synthesizer',
|
|
model: 'mock-model',
|
|
systemPrompt: 'Return structured review findings.',
|
|
outputSchema: ReviewFindings,
|
|
}
|
|
|
|
const findings = [
|
|
{
|
|
priority: 'critical',
|
|
category: 'security',
|
|
issue: 'User input is interpolated directly into the SQL query.',
|
|
fix_hint: 'Use parameterized queries for all database writes.',
|
|
},
|
|
{
|
|
priority: 'medium',
|
|
category: 'style',
|
|
issue: 'Error responses use inconsistent wording across branches.',
|
|
fix_hint: 'Standardize error messages and response payload structure.',
|
|
},
|
|
]
|
|
|
|
const agent = buildMockAgent(config, [JSON.stringify(findings)])
|
|
const result = await agent.run('Synthesize the reviewers into structured findings.')
|
|
|
|
expect(result.success).toBe(true)
|
|
expect(Array.isArray(result.structured)).toBe(true)
|
|
expect(ReviewFindings.parse(result.structured)).toEqual(findings)
|
|
})
|
|
})
|