elephant.ai

When prompt, tool, or agent logic changes need regression testing → run eval benchmarks to detect regressions or improvements.

When a task can be parallelized (bulk code review, multi-module refactor, batch analysis) → split into subtasks running in isolated git worktrees via multiple workers.

When the user wants to research academic papers → search ArXiv, extract PDFs, summarize/review/Q&A, generate experiment code.

When you need to interact with web pages using the user's logged-in browser → control Chrome via @playwright/cli daemon, reusing cookies/session with persistent tabs.

When user needs AnyGenIO content generation → unified CLI for task execution with progressive disclosure.

When you need to persist files (reports, docs, evidence) beyond the session → create/query/delete durable artifacts.

When the user wants text-to-speech audio → generate voice files locally (macOS say + afconvert, outputs m4a).

When a repeated workflow should be codified as a reusable skill → auto-create skill files, optionally using Codex/Claude.

When coding is done and changes need review → multi-dimensional code review (SOLID, security, quality, edge cases), outputs structured report.

When agent config parameters need querying or modification → read/write YAML configuration files.

When a topic needs thorough investigation from multiple sources → multi-source search + evidence assembly + structured report.

When you need to control macOS desktop apps (Atlas, Chrome, Finder) → automate via AppleScript.

When a Mermaid diagram needs to be shared as an image → render to PNG/SVG.

When 飞书多维表格 (Bitable) needs CRUD or management → create/query/edit apps, records, fields, views with advanced filtering and batch ops.

When 飞书日历 events or meetings need managing → CRUD calendar events, manage attendees, query free/busy status.

When 飞书云文档 needs creating, reading, or updating → CRUD docs with Lark-flavored Markdown syntax.

When 飞书 chat history needs reading → fetch group/DM messages, thread replies, and download attached files/images.

When 飞书任务 (tasks/todos) need managing → CRUD tasks, subtasks, and task lists.

When 飞书 plugin errors or auth failures occur → troubleshoot with FAQ and authorization diagnostics.

When the user wants to create or edit images → generate with Seedream (text-to-image + image-to-image).

When structured data needs visual presentation → use json-render templates (flowchart, form, dashboard, cards, gallery, table, kanban, diagram).

When you need to recall past conversation context → search and read conversation memories (stored as Markdown).

When the user wants to interact with Moltbook (AI agent social network) → post, browse, comment, vote, search via shell curl.

When the user needs NotebookLM operations → unified CLI (command/op) returning structured results.

When the user needs a presentation → PPT playbook (goal/audience → storyline → layout → accessibility → delivery).

When the user wants to set reminders → create, query, cancel one-time or recurring reminders with expiry scanning.

When Lark integration tests need running → execute test suite, analyze failures, auto-iterate fixes by severity.

When personality/collaboration strategy needs updating → evolve SOUL config under immutable-section protection with rollback support.

When a subtask should run on a different agent (Codex/Claude/Gemini) → delegate and track cross-agent execution.

When the user wants to create a short video → generate with Seedance, validate and save output files.

When responding in a 飞书 channel → enforce Lark message formatting and writing-style rules (always active).

When the agent needs to call 飞书/Lark APIs locally → unified CLI (auth/tool/api/help) with progressive disclosure.

When multiple skills need coordinated activation → manage skill activation, conflict arbitration, strategy, and audit.