/re-pitch
Stop and re-pitch the last message — it didn't land. Re-explains with the context the reader was missing, in plain words, cutting every line that isn't value while keeping every fact that is. Use when the user says wait what, I'm lost, that didn't land, what do you mean, or asks
$ npx -y skills add doodledood/manifest-dev --skill re-pitch --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
- You can call itInvoke it directly when you want it.
- Slash command
/re-pitch
Context preview
The summary Claude sees to decide when to auto-load this skill.
Stop and re-pitch the last message — it didn't land. Re-explains with the context the reader was missing, in plain words, cutting every line that isn't value while keeping every fact that is. Use when the user says wait what, I'm lost, that didn't land, what do you mean, or asks
SKILL.md
re-pitch.SKILL.mdname: re-pitch
description: "Stop and re-pitch the last message — it didn't land. Re-explains with the context the reader was missing, in plain words, cutting every line that isn't value while keeping every fact that is. Use when the user says wait what, I'm lost, that didn't land, what do you mean, or asks for a re-explanation in plain English."
argument-hint: '[what lost you]'
user-invocable: true
The last message didn't land. Re-pitch it — don't defend it, don't shrink it into a summary, don't repeat it louder. Explain it again as if for the first time, better.
- Start from the context the reader was missing: what question this answers and why it matters now, before the answer itself.
- Plain words, short sentences. Keep the project's own established names — check the project glossary (`CONTEXT.md` or equivalent) when one exists; that's the shared language. Every other term of art gets dropped or grounded in one line.
- Attention is sparse. Assume a reader who skims and bails the moment a sentence pays nothing: lead with the point, and cut every line that doesn't carry value — preamble, restated context they already hold, hedges, transitions that only fill space. Losing them a second time is the failure mode.
- Structure carries the skim. Short bullets for parallel points, numbered steps for anything ordered, and paragraphs of two sentences at most — never a wall of text. One idea per line, its load-bearing words bolded at the front, so a three-second pass over the bolds alone tells the whole story and the rest is there only for readers who keep going.
- Cut noise, not value. A re-pitch that drops load-bearing facts to get shorter has also failed: if the original carried ten facts that matter, the re-pitch carries ten — shorter comes from stripping the dead prose around them, never from stripping them.
- When the argument names what lost the reader, aim the re-pitch there first; otherwise re-pitch the whole message.
Read more
name: re-pitch description: "Stop and re-pitch the last message — it didn't land. Re-explains with the context the reader was missing, in plain words, cutting every line that isn't value while keeping every fact that is. Use when the user says wait what, I'm lost, that didn't land, what do you mean, or asks for a re-explanation in plain English." argument-hint: '[what lost you]' user-invocable: true
The last message didn't land. Re-pitch it — don't defend it, don't shrink it into a summary, don't repeat it louder. Explain it again as if for the first time, better.
- Start from the context the reader was missing: what question this answers and why it matters now, before the answer itself.
- Plain words, short sentences. Keep the project's own established names — check the project glossary (`CONTEXT.md` or equivalent) when one exists; that's the shared language. Every other term of art gets dropped or grounded in one line.
- Attention is sparse. Assume a reader who skims and bails the moment a sentence pays nothing: lead with the point, and cut every line that doesn't carry value — preamble, restated context they already hold, hedges, transitions that only fill space. Losing them a second time is the failure mode.
- Structure carries the skim. Short bullets for parallel points, numbered steps for anything ordered, and paragraphs of two sentences at most — never a wall of text. One idea per line, its load-bearing words bolded at the front, so a three-second pass over the bolds alone tells the whole story and the rest is there only for readers who keep going.
- Cut noise, not value. A re-pitch that drops load-bearing facts to get shorter has also failed: if the original carried ten facts that matter, the re-pitch carries ten — shorter comes from stripping the dead prose around them, never from stripping them.
- When the argument names what lost the reader, aim the re-pitch there first; otherwise re-pitch the whole message.
Loop engineering for Claude Code and other agent CLIs: figure the problem out first, define what done means, then execute and verify every criterion independently.
Other skills on manifest-dev.
- /auto-optimize-prompt
Iteratively auto-optimize a prompt until no issues remain. Uses review-prompt in a loop, asks user for ambiguities, applies fixes via prompt-engineering skill. Runs until converged.
Open skill - /compress-prompt
Compresses prompts/skills into minimal goal-focused instructions. Trusts the model, drops what it already knows, maximizes action space. Use when asked to compress, condense, or minimize a prompt.
Open skill - /harden-task-file
Harden a manifest-dev task guidance file for one-shot quality — either /define''s quality-gate/Default set or figure-out''s probe set. Iterates: orthogonality gap analysis, user-approved additions, prompt review, fix, converge. Use when a task file needs comprehensive coverage
Open skill - /learn-from-session
Analyze Claude Code sessions to learn what went right/wrong and suggest high-confidence improvements to skills. Use when asked to analyze a session, learn from a session, or review workflow effectiveness.
Open skill - /optimize-prompt-token-efficiency
Iteratively optimizes prompts for token efficiency by maximizing information density - reduces verbosity, removes redundancy, tightens phrasing while preserving semantic content. Use when asked to compress, shorten, reduce tokens, tighten, maximize density, increase information
Open skill - /sync-claude-code-plugins
Sync the prompt-engineering plugin from a local clone of claude-code-plugins into .claude/ so the repo is self-contained for isolated/web environments. Copies agents/skills, removes only previously-synced items that disappeared upstream. Other content in .claude/ is left alone.
Open skill

