shreyas-lyzr

Guide users through agent.yaml configuration, SOUL.md writing, and directory structure

Help users install gitagent and create their first agent from scratch

Help users register their agent on the gitagent public registry at registry.gitagent.sh

Help users export agents to other frameworks and import from existing tools

Help users search, install, create, and manage agent skills

Help users run agents with different adapters — Claude, OpenAI, Lyzr, GitHub Models, and more

Fast web browsing for Claude Code via persistent headless Chromium daemon. Navigate to any URL, read page content, click elements, fill forms, run JavaScript, take screenshots, inspect CSS/DOM, capture console/network logs, and more. ~100ms per command after first call. Use when you need to check a website, verify a deployment, read docs, or interact with any web page. No MCP, no Chrome extension — just fast CLI.

CEO/founder-mode plan review. Rethink the problem, find the 10-star product, challenge premises, expand scope when it creates a better product. Three modes: SCOPE EXPANSION (dream big), HOLD SCOPE (maximum rigor), SCOPE REDUCTION (strip to essentials).

Eng manager-mode plan review. Lock in the execution plan — architecture, data flow, diagrams, edge cases, test coverage, performance. Walks through issues interactively with opinionated recommendations.

Weekly engineering retrospective. Analyzes commit history, work patterns, and code quality metrics with persistent history and trend tracking.

Pre-landing PR review. Analyzes diff against main for SQL safety, LLM trust boundary violations, conditional side effects, and other structural issues.

Ship workflow: merge main, run tests, review diff, bump VERSION, update CHANGELOG, commit, push, create PR.

Agent-based prediction market simulation with heterogeneous traders. Informed traders, noise traders, market makers, Kyle lambda price impact, order book dynamics, and emergent price discovery. Based on Gode & Sunder (1993), Farmer et al. (2005), and Kyle (1985).

Dependency modeling beyond correlation matrices. Gaussian, Student-t, Clayton, and Gumbel copulas for correlated prediction market outcomes. Tail dependence quantification and vine copulas for high-dimensional portfolios. The reason Gaussian copulas failed in 2008 — and what to use instead.

Foundation Monte Carlo simulation for binary contracts and prediction markets. GBM-based path simulation, binary payoff estimation, confidence intervals, and Brier score calibration. The base layer everything else builds on.

Sequential Monte Carlo (particle filters) for real-time probability updating. Bootstrap filter with systematic resampling, ESS monitoring, logit-space state evolution, and credible intervals. For live event tracking like election night or real-time market monitoring.

Importance sampling for rare events, antithetic variates, control variates, stratified sampling, and stacking all three. Achieves 100-10,000x variance reduction over crude Monte Carlo. Table stakes for production simulation.

Design effective question-asking mechanisms for AI agents. Covers structured elicitation tools, multi-option UIs, plan-mode integration, and reducing friction in human-agent communication.

Design layered context strategies for AI agents. Avoid prompt stuffing by letting agents discover information incrementally through skill files, recursive references, subagents, and search tools.

Design search tools and context-building strategies for AI agents. Covers the evolution from RAG to agent-driven search, grep/glob patterns, and how to let agents build their own context.

Audit and refine an agent's tool set as model capabilities evolve. Identify tools that are now constraining the model, tools that should be merged or split, and opportunities to replace tools with progressive disclosure.

Design the right action space for an AI agent. Helps choose between bash-only, specialized tools, and hybrid approaches. Uses the "math problem" framework — match tools to the model's abilities.