Personal Claude Code plugin marketplace by dp92987.
Add the marketplace in Claude Code:
/plugin marketplace add dp92987/claude-plugins
Then install a plugin:
/plugin install <plugin>@dp92987-claude-plugins
Self-learning loop that captures personal engineering style — code style, architecture decisions, naming, testing preferences, things rejected in review — into memory files (~/.claude/learning-loop-memory/). Claude Code consumes them via @ imports in ~/.claude/CLAUDE.md; Codex and other agents via the ~/.claude/AGENTS.md pointer. Bundles a SessionEnd hook (disabled by default) for automatic extraction from finished sessions.
Skills: learning-loop.
Claude orchestrates, Codex codes. A SessionStart hook routes bounded implementation work to the OpenAI Codex CLI by default: Claude does requirements, research, planning, briefs, and verification; Codex writes the code. Review is built in — every Codex diff gets Claude's contextual review plus an independent codex-review; substantial Claude-authored code is Codex-reviewed too, with an adversarial whole-branch gate before PRs. Requires an installed and authenticated codex CLI.
Skills: codex-implementation, codex-review.
Post a "Claude Code session report" comment on a Jira ticket: what the session accomplished (task, PRs, review outcomes, deploy notes) plus session metadata — session link, cost, API/wall duration, code changes, and per-model token usage. Numbers come only from the user's pasted /usage output; the skill never estimates or fabricates them. Resolves the ticket from the branch name, previews the comment before posting, and never transitions the ticket. Requires the Atlassian MCP server.
Skills: jira-session-report.