claude-agent-toolkit

Standalone Speedrift-native planning and Workgraph compilation. PlanForge V2 uses a model-mediated planner persona, validated schema, critic pass, and Speedrift module/model assignment to produce an executable Workgraph with tests, evals, drift gates, and fix loops. Use directly when you need discovery, decomposition, or an end-to-end Speedrift execution graph.

Use when agents need Google Workspace access across Drive, Docs, Sheets, Calendar, Slides, Tasks, People, or related GWS services; choose the correct gws-account profile before using gws CLI or domain-specific gws-* skills. Do not use Python gspread/googleapiclient or custom OAuth. Use msgvault for email search/reading.

gws CLI shared reference. Use before any gws-* skill to choose the correct gws-account profile, avoid raw gws config drift, and apply shared Drive flags, auth checks, output formatting, and security rules.

Use when a local Pi agent needs administrator privileges, a password-protected command, or approval for global or ambiguous deletion

Workgraph-first drift-control autopilot (driftdriver + coredrift + optional specdrift/datadrift/depsdrift/uxdrift/therapydrift/yagnidrift/redrift) with optional Amplifier runtime integration. Use for start/resume, dependency-aware execution, and keeping code/spec/intent synced without hard blocks.

Use when managing projects, programs, initiatives, dependencies, schedules, Workboard/PAIA Program tasks, Linear/GitHub/Google/Folio work, cross-system planning, weekly operating reviews, stale work rescue, operational task decomposition, or research-backed planning for Braydon's work. Covers a model-mediated program panel, research/evidence discovery, model-authored records, HTML inspection packets, change plans, approval gates, and execution ledgers.

Use when writing, revising, speaking, or thinking in a named human or agent voice; using golden voice packets, voice packs, voice guides, tone guides, or prompt packs; creating or updating voice packets; or learning voice preferences from user-edited model drafts.

Default personal knowledge skill. Use for note-taking, searching notes, daily notes, knowledge lookup, or whenever the user refers to their notes or knowledge base. Prefer Folio over legacy Obsidian or deprecated paia-vault. Full access to the Folio personal knowledge app via localhost:3520.

Use when porting beads or superpowers workflows into Gemini CLI extensions or designing Gemini CLI command prompts that emulate multi-step agent workflows - covers extension layout, GEMINI.md context, command TOML patterns, and enforceable guardrails (tests/CI/pre-commit)

Use when creating GitHub issues, managing PRs, updating issue states, or searching repositories - provides bash scripts to interact with GitHub REST API without writing API calls directly

Use when users need multi-provider image/video generation workflows in `<workspace-root>/grok-aurora-cli`, including running `grok make`, provider-specific `grok video` and image commands, tracking async tasks, stitching clips, and evaluating outputs with `grok eval`.

Google Calendar: Create a new event.

Use when managing Linear issues, creating tasks from conversations, tracking work items, or querying project status - provides bash scripts to interact with Linear's GraphQL API without writing queries directly (project, gitignored)

Use when designing or building systems with AI/model calls, model-facing APIs, MCPs, CLIs, semantic routing, model memory, agent workflows, tool use, or model-owned decisions where coding agents might add rules, heuristics, regexes, thresholds, fallbacks, or hidden deterministic judgment.

Search and retrieve email from the offline msgvault archive. Full-text, vector/semantic, and hybrid search across all email accounts. Use this as the PRIMARY tool for finding, searching, and reading emails — NOT gws-gmail.

Use only when explicitly working with the legacy Obsidian vault or Obsidian MCP. Folio is the default personal knowledge system; this skill is for legacy Obsidian access, migration, or debugging.

Orchestrate multiple AI instances (clauded/codex CLIs) in tmux panes. Self-discovering coordinator with mandatory verification, synchronous monitoring, and auto-approval. Zero hallucination via imperative commands. (project, gitignored)

Codex as coordinator for tmux multi-agent workflows. MANDATORY verification after EVERY command. Zero hallucination enforced through explicit checkpoints.

Use when drafting, sequencing, reviewing, personalizing, staging, or QAing outbound email, LinkedIn, investor, BD, partner, sales, or follow-up outreach.

Use when working with the user's PAIA workboard, paia-work tasks, Workgraph-backed personal task boards, agent task handoff, task status, task creation, claiming, completion, comments, or planning work from the PAIA board.

Use when researching, qualifying, segmenting, prioritizing, or preparing target lists for prospects, partners, investors, BD contacts, customers, advisors, or outreach campaigns.

Use when asked to recover after a reboot, restart, OS update, lost tmux sessions, dead Codex agents, reboot-backup restore, tmux relaunch, corrupt git index repair, or stranded local commits from pre-reboot agents.

Use when managing Apple Reminders via MCP - covers list organization, reminder creation/updates, search patterns, and integration with other productivity workflows

Backward-compatible alias for speedrift. Use this when users say "rifts"; apply the speedrift Workgraph autopilot workflow and commands.

Request a design review for all commits on the current branch and present the results

Request a design review for a commit and present the results

Iterative review-fix loop for the current branch — reviews via daemon, fixes inline, re-reviews until passing or max iterations reached

Request a code review for all commits on the current branch and present the results

Request a code review for a commit and present the results

Shell methodology scenario planning with worldview elicitation, external scenario construction, actor-relative impact analysis, and rigorous documentation. Use when exploring futures, uncertainty, preparedness, positioning, or other decisions that benefit from multi-scenario reasoning.

This skill should be used when the user asks to "find dead code", "remove unused code", "clean up tech debt", "find duplicates", "identify unused functions", "kill dead code", or mentions tech debt cleanup, code cruft, or unused exports. Provides automated detection and removal of dead code, duplicates, and code quality issues with archive-based safety.

Use when the user needs to automate a real web task through a browser — filling forms, applying filters, extracting data, clicking through multi-step flows, or building reusable web automation scripts. Use when a task requires interacting with a live website where the agent must navigate, select, verify, and capture evidence. Do NOT use for API calls, static scraping, or tasks that don't need a real browser.

Build single-file HTML white papers with brand-ready templates, print-optimized 8.5x11 layout, Gemini media generation, and SVG diagram creation. Includes narrative engine with 10 story arcs and communication frameworks for research papers, strategic reports, and thought leadership content.

Intelligent workspace cleanup using multi-signal detection (similarity, timestamps, references) to identify and archive clutter with two-stage safety review

Use to discover which coding agents are running in tmux, find agents relevant to the current repo, check what they're working on, and coordinate or control them. Run before starting work to avoid conflicts, or to send commands to other agents.

Use when a user asks to address or close a specific roborev code review job after fixing or responding to findings.

Use when the user asks to fix open roborev reviews, invokes $roborev-fix, or provides roborev job IDs to fetch and close.

Use when the user asks to add a comment to an existing roborev code review and close it without changing code.

Use when investigating incidents, failures, regressions, recurring problems, postmortems, RCA requests, surprising metrics, process breakdowns, bad outcomes, near misses, or "why did this happen?" questions.

Build Keynote-style single-file HTML slide decks with brand-ready templates, minimal navigation, and Gemini nano banana media generation. Includes Narrative Engine integration for framework-driven deck creation with 17 proven storytelling structures and 5-agent review panel. Use when creating or editing slide decks, transforming content into presentations, or generating slide visuals.