(agents: Agent[])
| 52 | } |
| 53 | |
| 54 | async crawlAllAgents(agents: Agent[]): Promise<CrawlResult> { |
| 55 | if (this.crawling) { |
| 56 | log.debug('Crawl already in progress, skipping'); |
| 57 | return this.lastResult || this.emptyResult(); |
| 58 | } |
| 59 | |
| 60 | this.crawling = true; |
| 61 | |
| 62 | // Filter out agents whose owners have paused monitoring |
| 63 | const pausedUrls = await this.getPausedAgentUrls(); |
| 64 | const activeAgents = agents.filter(a => !pausedUrls.has(a.url)); |
| 65 | if (activeAgents.length < agents.length) { |
| 66 | log.info({ paused: agents.length - activeAgents.length, active: activeAgents.length }, 'Skipping paused agents'); |
| 67 | } |
| 68 | log.info({ agentCount: activeAgents.length }, 'Starting crawl'); |
| 69 | |
| 70 | // Convert our Agent type to AgentInfo for the crawler |
| 71 | const agentInfos: AgentInfo[] = activeAgents.map((agent) => ({ |
| 72 | agent_url: agent.url, |
| 73 | protocol: agent.protocol || "mcp", // Use agent's protocol, default to MCP |
| 74 | publisher_domain: this.extractDomain(agent.url), |
| 75 | })); |
| 76 | |
| 77 | try { |
| 78 | const result = await this.crawler.crawlAgents(agentInfos); |
| 79 | |
| 80 | this.lastCrawl = new Date(); |
| 81 | this.lastResult = result; |
| 82 | this.crawling = false; |
| 83 | |
| 84 | log.info({ |
| 85 | totalProperties: result.totalProperties, |
| 86 | successfulAgents: result.successfulAgents, |
| 87 | totalAgents: agents.length, |
| 88 | publisherDomains: result.totalPublisherDomains, |
| 89 | }, 'Crawl complete'); |
| 90 | |
| 91 | if (result.failedAgents > 0) { |
| 92 | log.warn({ failedAgents: result.failedAgents }, 'Some agents failed (no adagents.json)'); |
| 93 | } |
| 94 | |
| 95 | if (result.errors.length > 0) { |
| 96 | const errorsByDomain: Record<string, string[]> = {}; |
| 97 | for (const err of result.errors) { |
| 98 | (errorsByDomain[err.agent_url] ??= []).push(err.error); |
| 99 | } |
| 100 | log.warn({ errorsByDomain }, 'Crawl errors'); |
| 101 | } |
| 102 | |
| 103 | if (result.warnings && result.warnings.length > 0) { |
| 104 | log.warn({ warnings: result.warnings }, 'Crawl warnings'); |
| 105 | } |
| 106 | |
| 107 | // Snapshot pre-crawl state for diffing |
| 108 | const preCrawlAgents = await this.snapshotAgentState(); |
| 109 | |
| 110 | // Populate federated index from PropertyIndex and adagents.json files |
| 111 | const crawledDomains = await this.populateFederatedIndex(agents); |
no test coverage detected