القائمة
Load when the user wants to ingest a URL, document, or text snippet into the knowledge base for future retrieval. Trigger phrases: "save this", "ingest", "add to KB".
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Load when the user asks to queue, enqueue, or add a coding task to the fleet, dispatch a job to a repo, or ask the fleet worker to fix something. Trigger phrases: "add to fleet", "queue a job for", "send to fleet", "dispatch to".
Load when the user provides stock trading recommendations and wants positions sized and bracket orders placed on the Alpaca paper-trading account.
Load when the user wants to navigate a website, click through a flow, fill a form, extract data from a page, or record an interaction as a reusable playbook.
Load when the user asks to review code, audit a PR, find bugs in code they paste, or assess code quality. Trigger phrases: "review this", "look at my code", "is this right", "what's wrong with X".
Load when the user is responding to a customer inquiry, complaint, or support ticket and needs help drafting a reply with the right tone.
Load when the user wants analysis on a dataset — trends, distributions, summary statistics, or insights from a CSV/JSON/Excel file.
| name | Knowledge Ingestion |
| slug | knowledge-ingestion |
| description | Load when the user wants to ingest a URL, document, or text snippet into the knowledge base for future retrieval. Trigger phrases: "save this", "ingest", "add to KB". |
| icon | Upload |
| color | #059669 |
| version | 1.1.0 |
| category | research |
| tools | ["ingest_url_to_kb","scrape_webpage","search_knowledge_base"] |
| config_schema | {"type":"object","properties":{"target_kb_id":{"type":"string","default":""},"chunk_strategy":{"type":"string","enum":["page","section","paragraph"],"default":"section"}}} |
Target KB: {target_kb_id}. Chunk strategy: {chunk_strategy}.
Before ingesting, search the KB for the same source — duplicate ingestion creates noisy retrieval.
For URLs: ingest_url_to_kb directly.
For pages requiring navigation (login, JS): scrape_webpage first, then ingest the extracted text.
After ingest, verify by searching for a distinctive phrase from the document. If it doesn't surface, the ingestion silently failed.
Flag stale-prone sources (pricing pages, news, changelogs) in the title so they can be re-ingested later.
