cesdperez/llmfiles

cesdperez/llmfiles

Command Claude Code

One day of my work as a short Slack-ready standup post, split into team and no-team work, plus private notes. Defaults to yesterday.

cesdperez/llmfiles

Command Claude Code

Autonomous multi-lens review of a GitLab MR. Fans out parallel read-only lens agents, then posts inline diff comments for findings scoring 7+ without asking. Re-invocation does a follow-up pass like a human reviewer: resolves what got fixed, answers what got questioned, reviews only what is new. Also the deep path…

cesdperez/llmfiles

Command Claude Code

Triage every MR waiting on my review, autopost review comments (never approvals), and brief me in one line per MR. Low-risk MRs get a single-agent fast review; high-risk MRs dispatch to /glablensedreview.

cesdperez/llmfiles

Command Claude Code

Multi-angle code audit of local changes. Fans out parallel read-only lens agents (correctness, tests, standards, reuse, security, performance), then synthesizes one scored report.

cesdperez/llmfiles

Command Claude Code

Build a Claude Code skill for a tech stack, library, or CLI tool.

cesdperez/llmfiles

Command Claude Code

That last message did not land. Re-pitch it in Simplified Technical English using the project's own vocabulary.

cesdperez/llmfiles

Command Claude Code

Recommend how to run the task I am about to start: solo session, subagents, a Workflow, or an agent team. Says why, what it costs, and when to switch.

cesdperez/llmfiles

Command Claude Code

Research the latest Anthropic models (docs, benchmarks, user reports), then propose which model and effort each agent and the session in this repo should run. Adversarially reviewed. Outputs a today-vs-proposed table.

cesdperez/llmfiles

Command Claude Code

Run a bug bounty on a running app. Parallel hunters attack it from every angle (functionality, look, responsiveness on desktop and mobile, accessibility, races, load, abuse) with playwright-cli, curl and the app's own tools, then workers fix what matters test-first, in cycles, until it meets a stated bar. Ends at a…

cesdperez/llmfiles

Command Claude Code

Take a task from a Jira ticket, Slack thread, or plain words all the way to a reviewed, browser-verified draft MR with a green pipeline (work repos) or a local branch (personal repos). Say "open a PR", "ship to main" or "ship to prod" to go further.

cesdperez/llmfiles

Command Claude Code

Curate the agent-facing .md files in a repo (CLAUDE.md, AGENTS.md, rules, skills, commands, agents, and the docs they route to) to the Opus 5.5 bar, so agents load less and hit fewer wrong turns. Quick inline pass by default; --deep adds claims checked against the code, Anthropic's prompt audit, additions, and an…

cesdperez/llmfiles

Command Claude Code

Run several tasks as supervised Orca sessions, one worktree each, in parallel or in order. Tasks come from this conversation, a list of Jira keys, GitHub issues, Slack threads or text, or a Jira or GitHub query. Each worker runs /fullsend (or the command named) to the endpoint asked, and this session answers their…