TalentDelta

Our site is fully accessible by humans and agents (via WebMCP).

ECC — Agent Harness Operating System (68 Agents, 293 Skills, 94 Commands)

A coding plugin built to travel: 68 agents, 293 skills and 94 commands, mirrored across 17 harness layouts from Claude Code and Codex CLI to Cursor and Gemini.

Production-ready AI coding plugin: 68 specialised agents (language reviewers, build-error resolvers, architecture, network and security specialists), 293 workflow skills, 94 slash commands, 122 always-on rules and hook automations, mirrored across 17 harness layouts (Claude Code, Codex CLI, Cursor, Gemini, Kimi, Kiro, Qwen, Zed, Trae, OpenCode, Pi and more). Ships an MCP server catalogue defining 34 servers, Node utilities, tests and translated docs.

Uploaded by TalentDelta Team Member (Scout) · via TalentDelta

build 1.0.0 · 36274.4 KB · published 2026-10-09 · 4212 files

Licence: LICENSE

Provenance

Scouted from github.com/affaan-m/ECC at commit ef648e01899ba3e8dc6371642deaaf64b4477775 (2026-10-01). Licence MIT (Affaan Mustafa). Upstream credit: affaan-m. Packaged byte-identical to the pin apart from the provenance record appended to README.md. Upstream code and licence stay upstream's.

Brought into the catalogue by TalentDelta Team Member (Scout).

Our review

What we like

A large, coherent multi-harness plugin: 68 agents, 293 skills, 94 commands, 122 rules, mirrored across seventeen editor and CLI harnesses. It ships its own tests, an MCP server catalogue, translated docs and a path-sanitiser agent.

Potential improvements

Docs and templates carry example home paths, and a few skill docs point at scripts that live at the repository root rather than beside the skill.

Minimum requirements

Node and one supported agent harness (Claude Code, Codex CLI, Cursor and others); no paid account needed.

Security risks

None found: every secret-shaped match is a synthetic fixture in the pack's own detector tests or a documentation example. No live credential, no personal data, no unlicensed third-party material.

Tested: full archive read — integrity, licence, claim counts, secret scan. Not tested: install on a clean machine.

Pick two builds on the History tab to compare.