tmustier

Verify pi-for-excel work against a real local Excel desktop host while keeping Excel in the background. Use after changes to workbook tools, recovery, taskpane UI, Office.js host behavior, chart/image paths, or any adversarial verification where browser-only tests are insufficient.

Provision or reuse a local Windows 11 ARM VM as a reusable test harness for validating pi-for-excel inside real China-domestic WPS Spreadsheets. Use when an agent needs to test a specific WPS feature, WPS JSAPI behavior, add-in packaging, China WPS install/login, auth, taskpane boot, workbook operations, or an issue reproduction in real WPS rather than in Excel/Office.js or a browser fallback.

Plan and build Pi for Excel extensions safely: choose skill vs extension plugin vs connection, create flat SKILL.md skills, and handle API keys without asking users to paste secrets in chat.

Local terminal access via the tmux bridge. Use when the user asks about running shell commands, setting up the tmux bridge, or troubleshooting terminal connectivity.

Native Python execution via the local Python bridge. Use when the user asks about running Python locally, setting up the Python bridge, LibreOffice conversion, or troubleshooting Python connectivity.

Discover and call tools from configured MCP servers. Use when external capabilities are needed beyond built-in workbook tools.

Search the public web for up-to-date facts. Works out of the box with Jina (default, no API key needed); optionally Serper, Tavily, or Brave Search. Use when workbook context is insufficient and fresh external references are needed.

Remote control tmux sessions for interactive CLIs (python, gdb, etc.) by sending keystrokes and scraping pane output.

Detect repetitive AI defaults for a domain and generate a reusable avoidance profile. Use when asked to "unslop", "de-slop", "remove AI patterns", "make less generic", generate an anti-slop profile, or analyze AI output patterns for any domain (writing, landing pages, emails, code, etc.).

Audit and reconcile Thomas's todo list against all active data sources — calendar, email, customer deal docs, deal tracker spreadsheet, and any future sources. Use when Thomas asks to "review my todos", "what's outstanding", "todo audit", "anything I'm missing", "check my follow-ups", "what's stale", or as part of a Monday morning / Friday EOD review. Also use when he asks about overdue items, missed follow-ups, or wants a pipeline hygiene check.

Design things that AI agents can operate effectively — CLIs, tools, file formats, codebases, context layers, skills, visual assets, and any interface where agents are the primary or frequent consumer. Use when building a CLI for agents to use, structuring a codebase or repo for AI-assisted development, choosing file formats for agent-readable content, designing MCP tools or skills, creating agent-operable assets (HTML templates, markdown-driven workflows), setting up context infrastructure (AGENTS.md, SKILL.md, progressive disclosure), or making any existing system more agent-friendly. Also use when the user asks about agent DX, agent UX, or how to make something easy for agents to work with.

Extract session cookies from Chrome on macOS by decrypting Chrome's cookie database. Use when you need to authenticate API calls using the user's existing browser session — e.g., refreshing expired session cookies, accessing a service the user is logged into, or bootstrapping API access for a tool that uses cookie auth.

Long-running iterative development loops with pacing control and verifiable progress. Use when tasks require multiple iterations, many discrete steps, or periodic reflection with clear checkpoints; avoid for simple one-shot tasks or quick fixes.

Create or update Agent Skills (SKILL.md plus optional scripts, references, or assets). Use when someone asks to design a new Agent Skill, refine an existing one, or structure skills for Pi discovery, packaging, or other Agent Skills-compatible clients.

Guide for changing or extending Pi's behaviour. Use when someone wants to modify how Pi behaves, add capabilities, decide which Pi extension point or artifact to use, build or package an extension, create an Agent Skill, add prompt templates/themes/context/model providers, configure Pi resources, or asks whether a Pi internal patch is needed.

Coordinate multi-agent teamwork with shared task lists, mailbox messaging, and long-lived teammates. Use when the user asks to spawn workers, delegate tasks, work in parallel with agents, or manage a team of workers.

Inspect, edit, and generate PowerPoint slides. Progressive drill-down inspection, text editing with formatting preservation, table generation from YAML, and PNG rendering for verification.

Read transcripts and summaries from Pocket AI (heypocket.com) recording devices. Use when users want to retrieve, search, or analyze their Pocket recordings, transcripts, summaries, or action items.

Operate pi-symphony — unattended issue orchestration that polls Linear, creates isolated workspaces, and launches Pi workers to implement issues autonomously. Use when the user asks about symphony, setting up a WORKFLOW.md, launching or monitoring symphony runs, preparing Linear issues for orchestration, checking symphony status, or anything related to autonomous/unattended issue execution. Also triggers on "orchestration", "unattended", "autonomous issues", "symphony dashboard", "symphony preflight", or "WORKFLOW.md".

Create, edit, and manipulate PowerPoint presentations from the command line. Add slides, text, images, tables, charts, shapes, and apply professional designs. Use when the user wants to create or modify .pptx files.

Terminal UI + CLI to browse Economist headlines, search articles, and read full content in the terminal. Requires one-time browser login for full articles.

Gmail API skill for email operations via CLI. Supports reading, drafting, sending, archiving, labeling, and batch operations. Use when user needs to: read emails, create/send drafts, reply to messages, archive/trash messages, manage labels, or perform bulk email operations.

Web search and content extraction toolkit. Use for searching documentation, facts, current information, or extracting readable content from URLs. Supports multiple providers (ddgs keyless, brave_api with key), caching, and safe defaults. Prefer this over browser-tools when no interaction is needed.

Maintains continuity across long-running tasks that span multiple agent sessions. Use when starting or resuming a complex project that spans multiple sessions, or for tasks with many discrete features requiring iterative development.