Skip to main content
skill CX Codex

tldr · pwguler — Claude Code skill: what it does, how to install

tldr — a skill for Claude Code and Codex that reports completed tasks as concise outcomes instead of tool transcripts. Triggers after builds, verification runs, file changes, or catch-up queries.

MIT 977 installs ★ 10 updated 2026-10-16

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add pwguler/skills --skill tldr -a claude-code
Open on GitHub

What tldr does

The skill forces the agent to report outcomes instead of narrating every tool call. It retains exact file paths, commit hashes, pull request links, and error strings, while stripping reasoning and internal deliberations.

Formatting adapts to task size. Small updates take two or three lines, whereas multi-item outputs switch to bullet points or tables. At most one visual structure, like a file tree or a call flow, is allowed. Any tokens, passwords, or connection strings in commands and logs are replaced with placeholders.

The concise tone applies only to the summary message and leaves project specs or ADRs untouched. It does not restore previous chat sessions, which is handled by a separate recall skill.

What it helps with

  • summarize completed tasks without tool-call transcripts
  • redact API tokens and secrets from command outputs
  • catch up on session changes after background execution
  • generate compact status updates for builds and runs

Example prompts for the agent

  • What happened since my last message?
  • Give me a tldr of the test run and changes
  • Summarize the build outcome

How to install tldr

1Skills CLI (Vercel)

npx skills add pwguler/skills --skill tldr -a claude-code

2Manually: put the files in place

~/.claude/skills/tldr/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/tldr/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about tldr

How to install tldr in Claude Code?

Run npx skills add pwguler/skills --skill tldr -a claude-code or save the rule file to ~/.claude/skills/tldr/SKILL.md.

What does the tldr skill do?

It turns task completions and catch-up requests into concise results with exact paths and error strings, omitting tool logs and model reasoning.

Does tldr handle context from earlier sessions?

No, it only summarizes the active session or task, while reconstructing earlier sessions requires the recall skill.

Source text

SKILL.md · MIT
---
name: tldr
description: Report a finished task as its result, not its transcript. Or tell the user what happened since their last message. Read this skill before your final report on a task that changed files, made an artifact, or ran a check. Examples are a build, a workflow, a background agent, an implement or verify run, or a release. Use it also when the user asks what happened since their last message. Use it when the user asks to catch up on this session.
---

A task just finished, or the user asks what happened. Give the result, not the transcript.

Rules:

- When the user asks what happened since their last message, report as [catch-up.md](references/catch-up.md) says. Its reply order and its picture rules replace the rules below on the first line, the last line, and visuals. The other rules still apply. The `recall` skill rebuilds the work of earlier sessions, not this skill.
- Write in ASD-STE100 Simplified Technical English: one meaning per word, active voice, short sentences, no idioms.
- Fit the length to the task. A small task gets two or three lines. A large task, or one with much output, gets more. The reply is never longer than the result needs.
- Choose the form that the user reads fastest for this result. Use prose for a few outcomes. Use bullets or a table when the user must scan several separate items.
- Start with the result: what the task made or decided.
- Give the outcomes that matter. Keep exact paths, commands, names, and error strings. Link commits, PRs, and files.
- Redact secrets. Replace a token, key, password, or connection string in a command or an error with a placeholder. Never copy it.
- End with what needs the user now, or "nothing" when the task is done.
- Add one small visual when the result is a new structure that prose would only list. Examples are a tree of files or calls, a flow between parts, or a layout. One new or moved file is a path in prose, not a tree. Build the visual as [visual.md](references/visual.md) says. Add one visual at most. A second visual is transcript again.
- Leave out the story of the tool calls. Keep the outcomes. If the task made nothing, say so. Never invent progress.
- Unless asked, never explain what you were about to do, how you reasoned, or why the approach is sound.
- The user sees the diff and the output. Do not repeat them, except the evidence that `verify` requires: what ran, and what it showed.
- This register applies to this reply only. It never becomes the standard register of the session. The reasoning that it leaves out still goes into ADRs and specs on the land path.
- Write no introduction and no closing words. The summary is the whole reply. Stop.

© pwguler · MIT · pwguler/skills

Similar skills

All skills for Agent setup: memory, context, orchestration →

tldr by other authors comes first.

From the same set15

The whole set →
land land is a Claude Code skill that finishes a development branch through merge, PR, keep, or discard. It runs after implementation is complete. CX Codex +1 1,000installs drill A skill for Claude Code interviews the user about a plan or design until its decisions are settled. It runs for new features, unresolved designs, or a request to “drill this.” CX Codex +1 1,000installs implement implement — a skill for Claude Code and Codex that executes settled plans in thin test-driven slices. Runs when prompted to build code from a spec or slice plan. CX Codex +1 1,000installs rubric rubric — a skill for Claude Code finds defects in changes and judges design, interface, or writing quality. It sets evaluation criteria before asking an independent subagent to inspect the work. CX Codex +1 980installs prototype This Claude Code skill builds visual UI prototypes before implementation begins. It activates for mockup requests and helps settle a look or flow that text cannot resolve. CX Codex +1 978installs recall recall — a Claude Code skill that rebuilds recent project context from agent sessions, the live repository, and project history. Run it before starting or resuming work, with a topic or without one. CX Codex +1 922installs reflect reflect — a skill that extracts lessons from a coding agent session and places them into project checks, skills, or instruction files. Runs at the end of a session with corrections, errors, or surprises. CX Codex +1 904installs perf A Claude Code and Cursor skill that measures code performance and diagnoses slow processes using empirical data rather than guesses. CX Codex +1 899installs

Project sponsors

Companies that keep this analytics open