$ cat ./welcome.txt
Everything You Need to Run
AI Coding Agents
Compare the leading agents, map IDE + MCP integrations, and follow setup playbooks so your team ships safely with AI copilots.
$ ls -la ./agents/
Agent Resource Catalog
20 curated entries capture agent strengths, supported languages, pricing signals, and integration notes.
>> AI Agents
Autonomous AI assistants and coding tools that help with various tasks
[10] listings available
→ Coding
→ Writing
→ Research
+ 1 more...
// Coming soon: filters + search
>> MCP Servers
Model Context Protocol servers providing contextual data to AI assistants
[2] listings available
→ Database
→ File System
→ API Integration
+ 2 more...
>> Plugins/Extensions
Add-ons and integrations for various platforms and tools
[10] listings available
→ Claude Code Skills
→ n8n Workflows
→ IDE Extensions
+ 1 more...
// Coming soon: filters + search
$ cat integration-recipes.md
Wire Agents Into Your Toolchain
Each recipe lists auth requirements, environment hints, and validation steps so you can go from prototype to production-ready automations quickly.
// IDE & Editor Plugins
Wire agents into VS Code, JetBrains, Cursor, and Neovim to keep reviews and refactors lightning fast.
- Map agent commands to editor actions with readable prompts that include testing expectations.
- Store API keys in environment managers or secret stores—never in `.env.local` checked into git.
- Pin extension versions per environment so evaluations stay reproducible.
// Model Context Protocol
Expose source maps, docsets, and build tools through MCP servers so agents can reason safely.
- Audit every MCP tool for read/write scope before enabling in production sandboxes.
- Ship health checks that confirm context providers respond in under 500 ms.
- Document fallback behavior so agents degrade gracefully if a server goes offline.
// CLI & DevOps
Use agents inside terminals and CI runners for scaffolding, migrations, and release notes.
- Gate destructive commands behind confirmation prompts tied to branch protections.
- Log every agent-issued command so SRE teams can replay and audit activity.
- Run smoke tests after each agent-driven change before merging to main.
$ grep -r "best-practices" ./docs/
Best Practices for AI Coding Agents
Use these playbooks to align security, developer experience, and QA expectations before scaling agent-driven work.
[1] Operational Guardrails
Codify roles, command budgets, and human-in-the-loop checkpoints.
- Define when agents may push commits or require reviewer sign-off.
- Track cost ceilings per workspace and set alerts when usage spikes.
- Mirror prod data? Mask secrets and PII with deterministic fixtures.
[2] Collaboration Loops
Blend agents with engineers, QA, and PMs using structured prompts.
- Capture intent, constraints, and test plans up front in every task prompt.
- Rotate retrospective questions to hone prompting patterns weekly.
- Version control prompts/playbooks like code to share improvements.
[3] Evaluation Playbooks
Measure agent output with regression tests, linters, and reviewers.
- Bundle evaluation suites in `tests/` and run them via `npm test` per change.
- Tag a11y and perf probes so you can isolate `npm run test:a11y|perf` quickly.
- Document remaining coverage gaps in PRs to keep transparency high.