Organized along the agent's working flow — intake → analyze → output — plus the infrastructure that runs it and the practices that guide it. A few skills serve more than one stage; those carry an ↺ also… marker.
Export OpenCode, Claude Code, Codex, Antigravity, and Second Mind sessions into one incremental Markdown archive for browsing and search.
Search multi-source session archives with source routing, lexical-first retrieval, fresh semantic file lists, and privacy-safe result links.
Ingest voice recording notes, run local transcription, and maintain organized audio logs.
Download, archive, and send Outlook.com emails, parse calendar invites, and format Markdown.
Employs Docling parser for high-accuracy tabular and layout extraction from PDFs.
Local-text embedding + cosine similarity search CLI with atomic cache. Supports any OpenAI-compatible endpoint.
Read-only Stripe integrations to extract sales metrics, subscription trends, and logs.
Track cross-platform API token consumption, estimate costs, and export stats.
Ingest Apple Health metrics, track caffeine/sleep logs, and run regressions in local SQLite.
Scrape Roest coffee roaster API, detect crack clusters, and chart roasting telemetry.
Gathers info using multiple parallel agents with overlapping topics to cross-validate.
Iteratively extracts predictable cognitive axioms and traits from conversation logs.
Encodes SIT and Think Bigger as executable pipelines with hard validators, turning an agent into a structured innovation engine that produces buildable candidates with derivation chains.
Audit every linked claim for accessibility, support, evidence type, and confidence, then validate JSON and deterministically render an interactive two-column HTML review.
Search, publish, share, and manage Google Docs using Markdown and Tab interfaces.
Build previewable agent slide decks (Reveal.js): image-rendered slides by default, HTML modules on request—not PPTX.
Generate text-to-image prompts, edit images, and upscale resolution via Gemini/GPT.
Embed default P3-D65 or custom ICC color profiles into unmarked TIFF files.
Send custom domain emails, check attachments, and retrieve notifications.
Schedule draft tweets, queue threads, and fetch X (Twitter) statistics.
Xcode simulator builds, XCTest UI click-through testing, and visual screenshot verification flows without private data leakage.
Edit portraits, headshots, and ID photos with identity preservation, photographic coherence, multi-image relighting, and alpha outputs.
Local HTTP process daemon to bridge GUI/TCC permissions, logs, and process cancellations.
Submit and monitor OpenCode tasks, run cron pipelines, and archive SQLite databases.
Docker deployment configurations to quickly containerize OpenCode execution environments.
Architectural rules for designing CLI interfaces and nesting agent-to-agent calls.
Run a headless Gemini agent on Antigravity subscription quota, with first-run setup, file-based prompt/result/log artifacts, sandbox boundaries, timeouts, and fresh conversations.
Orchestrates concurrent background subagents for independent task branches.
Runtime best practices (non-interactive calling, nested agent rules, session recovery, and timeout management).
Use a user-owned ChatGPT Plus/Pro subscription for local browser PKCE, plaintext token lifecycle, refresh, and a minimal Codex request. Owner-only compatibility experiment; unstable and not recommended for production.
Strategy detailing the 70% accuracy boundary and script-writing priority.
Meta-instructions explaining how to write reusable, outcome-focused Skill files.
Maintains a strict "isolate-process-validate" loop to execute dry-runs on code changes.
Guidelines to verify information lying past model knowledge cutoff dates.
Scaffolds loose workspaces into organized directories (docs/, src/, tests/).
arXiv ID query aggregation, HTML/PDF conversion to markdown via Docling/Tavily, and duplicate matching quality checks.
Periodically scrape OpenRouter model activity data (token usage, request counts, rankings, benchmarks) into local SQLite, breaking the 31-day trailing window limit.
CDP step-by-step debugging CLI for AI agents writing Playwright E2E tests; covers SSO flows, modal handling, and known pitfalls.
Create, fork, extend, prune, privacy-review, and release a governed skill registry with explicit success criteria.