Generates a grounded, verified, two-tier agent context layer (lean CLAUDE.md + deep .ai/docs) for any codebase, plus report-only issue detection. 100% local, read-only.
Your AI agent maintains its own context files — a Claude Code plugin with an AGENTS-first model that covers Codex, Copilot, Cursor, Gemini, and 3 more tools. Signal Gate filtering, Context Guard hooks, health scoring, and MEMORY.md promotion.
Run Claude as an autonomous senior engineer on any codebase — config-driven hooks that enforce discover-then-reuse, plus the coordinator/succession/doc-sync skills and planner/builder/auditor agents. Frontend or backend, any language.
Silent-executor mode. Give it a task, it says "yaparım...", does the work, says "tamam." — no explanation, summary, code block, progress note or task list in between. Questions still get full answers. Commands: /oduncu kalk — mode on /oduncu yat — mode off /oduncu talk — trace on: the completion line carries the.
Shadow Tutor — a post-session programming tutor: it makes you predict why the AI made its one load-bearing decision before it reveals, then teaches the why on your own code.
A post-session programming tutor that runs as an in-session Codex/Claude Code skill. After an AI-assisted coding session it picks the one load-bearing decision you most likely don't truly understand, makes you predict why before it reveals, then teaches the why on your own code — fighting the skill atrophy that AI…
Cross-agent prompt pocket: remember your most-used prompts and pick one to run; auto-scans Claude Code, Codex and OpenCode sessions to record prompts repeated 7+ times. Forged by agent-plugin-kit.
On Point: the big idea in the first line, the one next action in the second, terse prose after. Substance, exact errors and security are never compressed.