Agent Skills by parcadei
Showing 288 vetted skills indexed across 2 GitHub repositories.
parcadeitrace-claude-code
Captures Claude Code sessions, turns, and tool calls as hierarchical Braintrust traces.
parcadeiskill-developer
Creates new Claude Code skills including MCP pipeline wrappers with SKILL.md frontmatter.
parcadeiasync-repl-protocol
Enforces await usage and single-block execution rules for async REPL testing.
parcadeirouter-first-architecture
Route tool calls through domain routers before invoking individual scripts.
parcadeisearch-tools
Select the appropriate code search tool based on query type and intent.
parcadeisearch-hierarchy
Selects the most token-efficient search tool for each code query type.
parcadeiresearch
Documents codebase structure by spawning parallel sub-agents and synthesizing findings into research reports.
parcadeicommit
Creates git commits with user approval and no Claude attribution.
parcadeiindex-at-creation
Indexes artifacts immediately at creation time instead of batch boundaries.
parcadeiworkflow-router
Routes user goals to specialist agents with configurable resource allocation.
parcadeidescribe-pr
Generate comprehensive pull request descriptions following repository templates.
parcadeiskill-development
Defines authoring rules and structural conventions for SKILL.md files.
parcadeisearch-router
Routes code search queries to the most token-efficient search tool.
parcadeimodular-code
Organizes Python code into modular files sized for maintainability and AI-assisted editing.
parcadeiopc-architecture
Explains how OPC orchestrates parallel Claude Code instances via hooks and subprocesses.
parcadeiagentic-workflow
Orchestrates a five-stage multi-agent pipeline for research, planning, validation, implementation, and review.
parcadeicomplete-skill
Tests the end-to-end skill persistence pipeline with a minimal workflow.
parcadeiagentica-claude-proxy
Integrates Agentica SDK agents with Claude Code CLI through a local SSE proxy.
parcadeiexplicit-identity
Enforces passing explicit session and trace identifiers across process and async boundaries.
parcadeisecurity
Audits code for vulnerabilities and verifies security fixes through a two-agent workflow.
parcadeiagentica-sdk
Build Python agents with the Agentica SDK using decorators, spawn, persistence, and MCP integration.
parcadeiresearch-external
Researches external documentation, web sources, and APIs using nia-docs, Perplexity, and Firecrawl.
parcadeiagent-context-isolation
Isolates background agent output from the main context window using file-based coordination.
parcadeino-polling-agents
Prevents sleep-loop polling when waiting for background agents to complete.