菜单
This skill should be used when the user asks to "investigate", "continue the investigation", "run the detective loop", "next investigation step", or wants to execute the Scan-Evolve-Focus-Act-File cycle on an active case. Runs the core investigation loop with periodic checkpoints.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | investigate |
| description | This skill should be used when the user asks to "investigate", "continue the investigation", "run the detective loop", "next investigation step", or wants to execute the Scan-Evolve-Focus-Act-File cycle on an active case. Runs the core investigation loop with periodic checkpoints. |
Execute the investigation cycle on an active MCP-backed case. Prefer Detective MCP tools over direct JSON edits or legacy scripts.
An active case must exist under .detective/cases/. If the MCP server is unavailable, explain that v2.1 investigation requires the detective MCP server and ask the user to check /mcp.
detective_graph_overview.detective_list_nodes.detective_score_candidate_actions.full_auto, dispatch specialist agents for independent actions; in checkpoint, run up to the checkpoint interval; in manual, ask the user before acting.detective_convergence_status and detective_deadlock_status.detective_export_markdown and detective_export_mermaid after meaningful graph changes.This section is deprecated. Use it only for legacy v1 flat JSON case files when the Detective MCP server is unavailable. It is not the workflow for Detective v2/v2.1 cases.
For v2/v2.1 cases:
/mcp.For legacy v1 flat case files, the old scripts can inspect status when MCP is unavailable:
python $PLUGIN_ROOT/scripts/board.py status .detective/cases/<case-id>.json
python $PLUGIN_ROOT/scripts/scoring.py suggest-phase .detective/cases/<case-id>.json
python $PLUGIN_ROOT/scripts/convergence.py .detective/cases/<case-id>.json
Use the output to assess:
When reviewing a legacy v1 flat case without MCP, treat the old Scan → Evolve → Focus → Act → File loop as a reasoning model only:
Legacy mutation commands such as fragment creation, thread creation, maturity changes, and propagation are deprecated. Do not use them for v2/v2.1 cases.
For legacy v1 fallback, stop when convergence is reported, the user asks to pause, the action budget is exhausted, or discussion is needed. For v2/v2.1, use detective_convergence_status, detective_deadlock_status, and MCP exports instead of script-based checkpointing.
These scripts are legacy v1 fallback references only:
scripts/board.py — Legacy board inspection and deprecated mutation helpersscripts/scoring.py — Legacy phase suggestion and scoring helpersscripts/convergence.py — Legacy convergence inspectionUse when debugging interactive CLI, REPL, TUI, ssh, telnet, nc, watch-mode, dev server, long-running command, terminal prompt flow, or any command that may not exit. Provides a PTY-backed MCP workflow for spawning terminal sessions, sending text and keys, reading output, managing multiple sessions, and exporting full transcripts.
This skill should be used when a user asks to debug an interactive CLI, REPL, TUI, prompt loop, shell wizard, watch mode, long-running terminal process, or any command that hangs, waits for input, redraws incorrectly, needs multiple inputs, behaves differently in a real terminal, or requires comparing tmux panes. Trigger on phrases like “it hangs after Continue?”, “the prompt never appears,” “my curses UI breaks in a small terminal,” “the watcher only fails after I answer the prompt,” or “inspect pane 2.”
macOS offensive security assistant — helps engineers audit applications for security vulnerabilities, identify bypass vectors in macOS security controls, and learn macOS internals through real-world case studies (CVEs). Covers: app vulnerability assessment (entitlement/injection/sandbox/TCC analysis), system internals, binary analysis, shellcode crafting (x64/ARM64), dylib injection, Mach IPC exploitation, function hooking, XPC attacks, sandbox escapes, TCC bypasses, symlink/hardlink attacks, kernel code execution, persistence mechanisms, Gatekeeper/XProtect bypass, AMFI/MACF internals, launch constraints, application-runtime injection (Electron/Chromium/NIB/.NET/Java/Python), IOKit/DriverKit driver attacks, MDM/DEP exploitation, keychain attacks, dangerous entitlements, and full penetration testing workflows. Use this skill whenever the user asks about: checking macOS apps for security issues, auditing entitlements or sandbox profiles, learning macOS security internals, macOS security research, macOS privile
微步在线威胁情报查询工具。支持IP、域名、文件哈希威胁情报查询，漏洞情报查询，资产测绘等功能。支持微信登录自动化和X语言高级搜索。当用户需要查询威胁情报、进行资产测绘或使用微步在线平台时，应使用此技能。
This skill should be used when the user asks to "close the case", "conclude the investigation", "wrap up", "present findings", or when the convergence check indicates the case has been solved. Produces the final resolution with complete evidence chain.
This skill should be used when the user asks to "discuss the case", "brainstorm about the investigation", "I'm stuck", "what do you think", or when the investigation loop detects a deadlock, hypothesis ambiguity, or needs user input to break a tie. Facilitates structured case discussion.