Mark-Life

Mark-Life/telegram-claude-codex

Hook Claude Code

Runs when the agent finishes a response, executing stop-lint-typecheck.sh. From Mark-Life/telegram-claude-codex.

Mark-Life/telegram-claude-codex

Settings file Claude Code

Agent settings declaring 1 hook event (Stop) and 5 allowed tools.

Mark-Life/telegram-claude-codex

Instructions file CodexOpenCode

Instructions for Mark-Life/telegram-claude-codex, a project described as: Telegram bot that runs Claude Code or OpenAI Codex on your VPS and streams responses back to chat. Supports project switching, sessions, voice, plan mode, and queues.

Mark-Life/telegram-claude-codex

Instructions file

Instructions for Mark-Life/telegram-claude-codex, covering claude.md, commands, architecture, module overview (src/) and data flow.

Mark-Life/agent-skills

Skill Claude CodeCodex

Use for public file uploading: when a screenshot, GIF, or video has to be embedded in a PR, issue, doc, or message.

Mark-Life/agent-skills

Skill Claude CodeCodex

Use when starting a brand-new project or repository from scratch. Triggers on "create a new project", "spin up a new app", "scaffold a new repo", "start a new Next.js project". Scaffolds from the personal Next.js monorepo template (Mark-Life/netxjs-monorepo) and runs the standard bootstrap.

Mark-Life/agent-skills

Skill Claude CodeCodex

Observability for TypeScript services: one wide event per request or job, high-cardinality fields, span and metric design, OTel export, tail sampling. Use when adding or reviewing logging, tracing, telemetry, metrics or OTel wiring, deciding what a request or job should record, or debugging production from what a…

Mark-Life/agent-skills

Skill Claude CodeCodex

Use when writing a pull request, issue, or ticket — title or body — or before running gh pr create or gh issue create.

Mark-Life/agent-skills

Skill Claude CodeCodex

Primitive-first product design. Use when shaping a new product, app, SDK, API, or tool surface, deciding what to build and what to leave to the caller, writing a spec or PRD, or judging whether an idea is worth building.

Mark-Life/agent-skills

Skill Claude CodeCodex

Use when producing the final output a person will read — a chat reply, summary, status update, handoff, or a PR, issue, or commit message — not your private reasoning. Triggers on wrapping up a task or reporting back to the user.

Mark-Life/agent-skills

Skill Claude CodeCodex

Use when writing or editing instructions a model will later execute — anything future context windows load as guidance. Triggers on 'edit this skill', 'update CLAUDE.md / AGENTS.md', 'write agent rules', 'improve this prompt'.

Mark-Life/agent-skills

Skill Claude CodeCodex

Writing prose a reader chose to read. Use when drafting or editing a blog post, article, docs page, landing page or newsletter, or when a draft reads like AI. For output handed back to the user, use agent-to-human.

Mark-Life/agent-skills

Skill Claude CodeCodex

Audit and shrink the fixed context a coding agent auto-loads every session — tool/MCP definitions, plugins, skills, subagents, memory/rules files, built-in tools & feature subsystems — by pruning, gating, denying, or routing what's loaded but unused. User-invoked: type it when the context feels bloated.

Mark-Life/agent-skills

Skill Claude CodeCodex

Audit the last N days of agent transcripts across every project at once and find where wall-clock, tokens, and repeated human corrections are going. A zero-dependency script mines every transcript into five fact tables and prints a summary; the model reads the numbers, digs into the tables, and returns a ranked…

Mark-Life/agent-skills

Skill Claude CodeCodex

A viewer for Claude Code's per-project auto-memory. A zero-dep TypeScript script scans /.claude/projects/ /memory/, renders a self-contained, secret-redacted HTML explorer, and opens it in the browser. No model curation, no edits, no findings — the model only runs the script and opens the report. Local, human-facing…

Mark-Life/agent-skills

Skill Claude CodeCodex

Generate a self-contained HTML report that debugs what is in a Claude Code or Codex session's context window and how every token is spent — context budget, the hidden cost of retained thinking, the dumb-zone cutoff, loaded CLAUDE.md/skills, and the full history.