pure-skill-suite

Generates 3D interior models from photos and injects them into Mapbox GL JS map instances.

Generates dynamic video montages and edits video content for social distribution.

The cold-start protocol for an AI agent PICKING UP a handoff bundle (the receiving half of `handoff`). Triggers when a fresh session is pointed at a handoff directory or pasted a handoff block — e.g. "read this handoff", "take over from this bundle", "you're picking up <path>", "continue this work", or when the first message is a handoff block. Boots the agent loaded: reads the bundle README + manifest, runs Prime from it, confirms understanding and locked decisions BEFORE writing code, and refuses to start on a bundle with no manifest/mission. Works in any tool (Claude, Cursor, Copilot, Codex, Gemini, Aider, Windsurf).

The mandatory completion-report skill. When an agent finishes (or is blocked on) a handoff bundle or delegated task, it MUST write the outcome to files — never only into chat. Triggers on "I'm done", "this is shipped", "wrap up", "close out this task", "I'm blocked", or automatically at the end of any `handoff-receiver` / `execute` run. Writes a `<bundle>-shipped/handoff.md` report with a fixed schema (commit SHA, what shipped, files touched, test/smoke results, deferred items, new open questions) and sets the source bundle's `manifest.json` status to `shipped · <SHA>` or `blocked · <reason>`. This is the file-based feedback loop that lets a control plane track every agent across tools.

Lightweight mid-flight progress logging for long agent runs. Where `report-back` writes the final outcome, status-beacon drops short, timestamped progress notes to a file DURING the work so a human or orchestrator can check in without interrupting. Triggers on long/multi-phase tasks, or when the user says "keep a log", "leave progress notes", "beacon this", "let me track this". Appends to a `progress.md` inside the active bundle/work dir. Honest about uncertainty — notes are signals, not guarantees. Works across Claude, Cursor, Copilot, Codex, Gemini, Aider, Windsurf.

Captures insights, rules, and learnings from any AI coding conversation into the user's personal ai-doctrine.md file — a single source of truth that is then auto-synced into Claude Code, Cursor, Copilot, Codex, Gemini CLI, and Windsurf so every tool the user opens reads the same hard-won rules. Use this skill whenever the user says "add to doctrine", "save this rule", "capture this learning", "this should be a rule", "doctrine this", "remember this", or asks to record an insight from a prompt/response. Also triggers when the user shares a hard-won learning, a prompt pattern that worked unusually well, a failure mode to avoid, or any rule they want to remember across tools.

Delegates real work to a coding agent (Claude Code, Cursor agent mode, Codex, Aider) and tracks the run end-to-end. Triggers on "execute", "run agent", "ship this", "delegate this", "hand off to Claude Code". Creates a run reference, watches for completion signals, links output back into the conversation, and offers to save successful execution patterns to ~/ai-doctrine.md so the same delegation playbook works across every agentic tool the user has installed.

Migrates an in-flight AI coding session to a fresh chat (or a larger-context model) without losing state. Triggers when the user says "handoff this", "we're running out of context", "fresh chat for this", "this is getting heavy", "prepare a handoff", "context window is getting full" — OR when the AI detects (heuristically) that the remaining work won't fit the current conversation. Always runs AFTER Prime has completed so the new chat boots loaded with the primer. Produces a copy-pasteable Markdown handoff block that works pasted into any AI tool (Claude, Cursor, Copilot, Codex, ChatGPT). Captures repeated handoff patterns into ~/ai-doctrine.md so the user's handoff discipline carries across every tool via sync_to_agents.sh.

Maintains a registry of agent runs and outputs across every coding agent the user works with (Claude Code, Cursor, Codex, Aider, Gemini CLI, Windsurf). Triggers on "list agents", "show recent runs", "agent status", "what's in flight", "find that run from yesterday". Reads from ./runs/ and ~/.vibe-coding/runs/, provides search, tagging, and outcome tracking. Writes runtime patterns and lessons back to ~/ai-doctrine.md so every tool benefits from the user's accumulated agent operating experience.

Builds a tailored primer for an AI coding session — the stack, style, constraints, threat model, and audience that should be loaded BEFORE any real task. Triggers when the user says "prime me", "set up context", "build a primer", "start a new session", "I'm starting a new feature", or begins a fresh conversation without context. Reads the user's ~/ai-doctrine.md (if present) to seed the primer with rules they already established. Always offers to save high-leverage primers back to doctrine so the same setup carries to Cursor, Copilot, Codex, Gemini, and Windsurf via sync_to_agents.sh.

Runs the full PURE Loop (Prime → Understand → Refine → Execute) on a feature request, calling each phase's skill in sequence with explicit checkpoints for user approval. Triggers on "run PURE on...", "process this feature", "PURE this", "full loop", or any large feature request the user wants handled methodically. Always concludes by offering to capture learnings from all four phases into ~/ai-doctrine.md so the same playbook improves every other AI tool the user works with (Cursor, Copilot, Codex, Gemini, Windsurf).

Runs systematic self-review on AI-generated output before it's accepted. Triggers on "refine this", "review as senior", "improve this", "second pass", or after any AI output that needs a quality gate. Applies role-prompts (senior security engineer, principal architect, hostile reviewer) and produces a v1 → v2 diff with explicit reasoning. Captures effective refine prompts to ~/ai-doctrine.md so the same review patterns travel across Claude, Cursor, Copilot, Codex, Gemini, and Windsurf.

Clarifies scope before code is written. Summarizes existing context, identifies gaps, and asks targeted clarifying questions one at a time so the AI doesn't sprint in the wrong direction. Triggers on "understand this", "what am I missing", "clarify requirements", "make sure you got it", or any vague feature request. Produces a structured comprehension brief that the user approves before any code is generated, and offers to save useful clarification patterns to ~/ai-doctrine.md so every other AI tool (Cursor, Copilot, Codex, Gemini) starts with the same comprehension habit.