研究任务时效过滤：Perplexity web_search 引入 freshness 参数
解决“检索结果太旧”问题。前置：已用 Perplexity 作为 web_search provider。步骤：先跑无过滤基线→加入 freshness（24h/7d）→比较结果时效与覆盖。关键配置：在 cron 查询模板固定 freshness。验证：头部结果时间分布收敛到目标窗口。风险：窗口过窄会漏掉高价值常青资料。
GITHUBDiscovered 2026-02-14Author togotago
Prerequisites
- Perplexity provider and credentials are configured for web_search.
- Your research workflow supports parameterized query payloads.
Steps
- Run one baseline query without freshness and record top-result publication dates.
- Add freshness window (e.g., 24h or 7d) to the same query template.
- Compare relevance and recency between baseline and filtered runs.
- Persist the chosen freshness window in daily cron prompts.
Commands
openclaw gateway statusopenclaw helpVerify
Most top results now fall inside the intended recency window while preserving core topic coverage.
Caveats
- Overly strict freshness can reduce source diversity.
- Accepted freshness values may vary by provider implementation（需验证）.
Source attribution
This tip is aggregated from community/public sources and preserved with attribution.
Open original source ↗