New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Retrospect

by meain in meain/dotfiles

Review the current conversation to find agent mistakes, inefficiencies, and corrections, then propose new guidelines to add to CLAUDE.md to prevent recurrence.

MITAuto-check passedAgent Workflows

Install Retrospect

skills CLI
$ npx skills add meain/dotfiles --skill retrospect -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install meain/dotfiles retrospect --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/meain/dotfiles.git skills-src && mkdir -p .claude/skills && cp -r skills-src/agents/.agents/skills/retrospect .claude/skills/retrospect && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
retrospect
GitHub stars
285
Token cost
~1.4k tokens
SKILL.md length
656 words
Files
1
Skills in repo
35
Repo updated
First seen
Licence
MIT

At a glance

Review the current conversation to find agent mistakes, inefficiencies, and corrections, then propose new guidelines to add to CLAUDE.md to prevent recurrence.

  • Works in 7 steps: Scan the conversation → Read CLAUDE.md first → Derive lessons → …
  • Asks to capture anything from the current session into instructions
  • SKILL.md covers Step 1: Scan the conversation, Step 2: Read CLAUDE.md first, Step 3: Derive lessons and Step 4: Present all proposals, plus 4 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Retrospect is an agent skill from meain/dotfiles. Review the current conversation to find agent mistakes, inefficiencies, and corrections, then propose new guidelines to add to CLAUDE.md to prevent recurrence. Triggers: /retrospect, "what went wrong", "capture lessons", "update CLAUDE.md", "add this to instructions", "remember not to do this", "update the guidelines", "add to agents.md", "end of session review", or when the user asks to capture anything from the current session into instructions. Always use this skill when the user invokes it at the end of a…

Its SKILL.md is about 1.4k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering Agent instruction files. The repository describes itself as: If there is a shell, there is a way! The licence is MIT.

When your agent uses it

  • Asks to capture anything from the current session into instructions
  • The user invokes it at the end of a session — even if the session seemed smooth
  • There may still be patterns worth capturing

Example prompts

  • “what went wrong”
  • “capture lessons”
  • “update CLAUDE.md”
  • “/retrospect”

Workflow steps

7 steps, taken from the step headings in SKILL.md.

  1. Scan the conversation
  2. Read CLAUDE.md first
  3. Derive lessons
  4. Present all proposals
  5. Apply confirmed rules
  6. Apply confirmed skill updates
  7. Handle skill/script suggestions

What it can do on your machine

Read from SKILL.md and the folder at commit 3e336e2. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Retrospect loads about 1.4k tokens when it runs. Until then it costs about 154 tokens; SKILL.md has 656 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~154
When it runs · the whole SKILL.md, loaded when a task matches
~1.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from meain/dotfiles at commit 3e336e2, republished under its MIT licence (© meain). 656 words, ~1,433 tokens.

Download SKILL.mdSave it as .claude/skills/retrospect/SKILL.md (or your agent's skills folder).
name
retrospect
description
Review the current conversation to find agent mistakes, inefficiencies, and corrections, then propose new guidelines to add to CLAUDE.md to prevent recurrence. Triggers: /retrospect, "what went wrong", "capture lessons", "update CLAUDE.md", "add this to instructions", "remember not to do this", "update the guidelines", "add to agents.md", "end of session review", or when the user asks to capture anything from the current session into instructions. Always use this skill when the user invokes it at the end of a session — even if the session seemed smooth, there may still be patterns worth capturing.

Retrospect — Session Review & CLAUDE.md Updater

At the end of a session, scan the conversation for lessons and update CLAUDE.md with new guidelines to prevent the same mistakes or inefficiencies from recurring.

Step 1: Scan the conversation

Read through the entire current conversation and identify moments where:

  • The user corrected an error ("no, don't...", "that's wrong", "actually...", "wait...", "you shouldn't have...")
  • The user redirected approach ("instead of X, just do Y", "don't ask, just...", "skip the...", "you don't need to...")
  • The agent retried something multiple times before getting it right
  • The user had to repeat themselves about something they'd already said
  • The agent asked unnecessary questions or stalled when it could have acted
  • The agent missed a convention or pattern that was already in the codebase
  • A skill was invoked and the agent had to work around a gap, use an undocumented pattern, or the skill instructions were incomplete/misleading

For each moment, note:

  • What the agent did wrong or inefficiently
  • What the user wanted instead
  • The underlying principle being violated or missing

Step 2: Read CLAUDE.md first

Before deriving any lessons, find and read the project's CLAUDE.md (look in the current working directory, then parent directories up to home). The global user AGENTS.md lives at /Users/meain/.agents/AGENTS.md — always check there. This is essential — you need to know what's already there so you don't propose duplicates or near-duplicates.

Step 3: Derive lessons

Group related corrections into concise, actionable lessons. A good lesson is:

  • Specific: About a concrete behavior, not a vague observation
  • General: Applicable beyond just this one case — think about what rule would have prevented it
  • Brief: One to two sentences max
  • New: Not already covered (even partially) by an existing rule in CLAUDE.md

For each lesson, decide the best way to capture it:

  1. CLAUDE.md rule — behavioral guideline, convention, or preference
  2. Skill update — if an existing skill was invoked and its instructions were incomplete, misleading, or missing a pattern that came up during the session
  3. Skill suggestion — if a multi-step workflow recurred that would benefit from a dedicated skill
  4. Script suggestion — if a specific repetitive task could be automated into a helper script

Step 4: Present all proposals

Show the user everything at once before touching any files. Format like this:

I found N things worth capturing from this session:

**1. [Short lesson title]**
   What happened: [one line description of the mistake/inefficiency]
   Proposed rule: "[exact text to add to CLAUDE.md]"
   Section: [which CLAUDE.md section this belongs in]

**2. [Short lesson title]**
   What happened: ...
   Proposed rule: "..."
   Section: ...

📝 **Skill update: [skill name]**
   What happened: [gap or undocumented pattern encountered]
   Proposed change: "[exact text to add/modify in the skill's SKILL.md]"

🔧 **Skill suggestion: [name]**
   What happened: [recurring workflow that came up]
   Suggestion: Create a skill to handle [description]

Which should I apply? (e.g. "1, 3", "all", "none", "1 and the skill")

Wait for the user's response before making any changes.

Show full SKILL.md (274 more words)Show less

Step 5: Apply confirmed rules

You already read CLAUDE.md in Step 2, so you know its structure and tone. For each confirmed rule:

  1. Find the most relevant existing section — insert there rather than appending to the end. For example, a rule about error handling goes under error handling, a testing convention goes under testing patterns
  2. If no section fits, add a new section at the end with an appropriate heading
  3. Match the existing tone and style — look at how other rules are phrased and follow the same pattern
  4. Keep the addition minimal — no preamble or explanation, just the rule

After all edits, show a brief summary of what was added and where.

Step 6: Apply confirmed skill updates

For each confirmed skill update, read the skill's SKILL.md, find the appropriate section, and apply the change — same principles as CLAUDE.md edits (match tone, minimal addition, no duplicates).

Step 7: Handle skill/script suggestions

If the user confirmed a skill suggestion, say:

"I'll kick off the skill creator for [skill name]. Here's a starting description: [brief description of what it should do]. Ready to start?"

Then invoke the skill-creator skill to begin the workflow.

Notes

  • Only analyze the current conversation — do not search past sessions
  • If there are no lessons to capture (session was clean), say so honestly rather than inventing guidelines
  • Prefer editing existing rules over adding redundant new ones if a lesson is already partially covered
  • Don't add rules that are already in CLAUDE.md — check carefully before proposing
  • Only update the project-level CLAUDE.md found via the working directory; don't mention or touch other CLAUDE.md files unless the user brings them up

© meain, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in agents/.agents/skills/retrospect of meain/dotfiles.

Open the folder on GitHubat commit 3e336e2

Compare with similar skills

Retrospect next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Retrospect compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Retrospect this skillmeain/dotfiles285—~1.4kAutomated safety check: PassMIT
Using Agent Skillsaddyosmani/agent-skills103k4 repos~2.4kAutomated safety check: PassMIT
Claude ReflectBayramAnnakov/claude-reflect1.7k2 repos~627Automated safety check: PassMIT
Neat-Freak Knowledge CloseoutKKKKhazix/khazix-skills21k—~1.9kAutomated safety check: PassMIT
Writing For Agentsbestofjs/bestofjs3.1k18 repos~2.7kAutomated safety check: PassMIT
Task Observerrebelytics/one-skill-to-rule-them-all3.2k1 repos~12kAutomated safety check: PassCC-BY-4.0

Similar skills

  • Using Agent Skills

    addyosmani/agent-skills

    Meta-skill for choosing which workflow skill fits the task at hand, plus always-on habits: surface assumptions, stop on confusion, push back, keep it simple and stay in scope.

    103k GitHub starsUsed in 4 repos~2.4k tokens
    Agent WorkflowsAuto-check passed

  • Claude Reflect

    BayramAnnakov/claude-reflect

    Self-learning system that captures corrections during sessions and reminds users to run /reflect to update CLAUDE.md.

    1.7k GitHub starsUsed in 2 repos~627 tokens
    Agent WorkflowsAuto-check passed

  • Neat-Freak Knowledge Closeout

    KKKKhazix/khazix-skills

    Brings project docs, agent rule files, authorized memory and leftover workspace files back in line with what the code and runtime actually do at the end of a work session.

    21k GitHub stars~1.9k tokensUpdated 7 days ago
    Agent WorkflowsAuto-check passed

  • Writing For Agents

    bestofjs/bestofjs

    Writing documents for agents. An agent skill from bestofjs/bestofjs.

    3.1k GitHub starsUsed in 18 repos~2.7k tokens
    Agent WorkflowsAuto-check passed

  • Task Observer

    rebelytics/one-skill-to-rule-them-all

    Monitors task execution for skill improvement opportunities.

    3.2k GitHub starsUsed in 1 repo~12k tokens
    Agent WorkflowsAuto-check passed

  • SkillOpt Sleep Cycle

    microsoft/SkillOpt

    Official

    Runs an on-demand or nightly sleep cycle that reviews past Claude Code sessions and proposes validated updates to CLAUDE.md and skills.

    18k GitHub stars~2.3k tokensUpdated 2 days ago
    Agent WorkflowsAuto-check passed

More from meain/dotfiles

All 35 skills in this repo

  • Recall

    meain/dotfiles

    Search past Claude Code and Codex sessions. An agent skill from meain/dotfiles.

    285 GitHub starsUsed in 1 repo~684 tokens
    Auto-check passed

  • Grill With Docs

    meain/dotfiles

    Grilling session that challenges your plan against the existing domain model, sharpens terminology, and updates documentation (CONTEXT.md, ADRs) inline as decisions crystallise.

    285 GitHub starsUsed in 20 repos~875 tokens
    Auto-check passed

  • Backlog

    meain/dotfiles

    Daily backlog management — full planning review OR add a single entry from a URL.

    285 GitHub stars~3k tokensUpdated yesterday
    Auto-check passed

  • Concern Review

    meain/dotfiles

    Generate an interactive local HTML review page for a large PR or diff, grouping the changed files by logical concern (not just by file) so a reviewer can go through one theme at a time instead of a…

    285 GitHub stars~2.2k tokensUpdated yesterday
    Auto-check passed

  • My Weekly Report

    meain/dotfiles

    Generate a concise weekly status update in team format. An agent skill from meain/dotfiles.

    285 GitHub stars~2.5k tokensUpdated yesterday
    Auto-check passed

  • Web Search

    meain/dotfiles

    Search the web using lynx and DuckDuckGo. An agent skill from meain/dotfiles.

    285 GitHub stars~830 tokensUpdated yesterday
    Auto-check passed

Categories

Questions about Retrospect

What does Retrospect do?

Review the current conversation to find agent mistakes, inefficiencies, and corrections, then propose new guidelines to add to CLAUDE.md to prevent recurrence. Retrospect is an agent skill from meain/dotfiles.md to prevent recurrence.

When should I use Retrospect?

Retrospect fits situations like: asks to capture anything from the current session into instructions; the user invokes it at the end of a session — even if the session seemed smooth; there may still be patterns worth capturing.

How do I install Retrospect in Claude Code?

Run `npx skills add meain/dotfiles --skill retrospect -a claude-code`. Or copy the skill folder (agents/.agents/skills/retrospect in meain/dotfiles) into .claude/skills/retrospect in your project. Claude Code loads it when a task matches its description.

How do I install Retrospect in Codex?

Run `npx skills add meain/dotfiles --skill retrospect -a codex`. Or copy the skill folder (agents/.agents/skills/retrospect in meain/dotfiles) into .agents/skills/retrospect in your project. Codex loads it when a task matches its description.

Can I use Retrospect in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add meain/dotfiles --skill retrospect -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/retrospect, .gemini/skills/retrospect, .github/skills/retrospect and .opencode/skills/retrospect in your project.

What does Retrospect need to run?

SKILL.md names no scripts, command-line tools or credentials: Retrospect is instructions for the agent only.

Does Retrospect access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Retrospect safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Retrospect use?

Retrospect is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Retrospect use?

About 1.4k tokens (SKILL.md is roughly 5.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Retrospect?

Skills that share tags, products or a category with Retrospect: Using Agent Skills (addyosmani/agent-skills, 103k stars), Claude Reflect (BayramAnnakov/claude-reflect, 1.7k stars), Neat-Freak Knowledge Closeout (KKKKhazix/khazix-skills, 21k stars) and Writing For Agents (bestofjs/bestofjs, 3.1k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Retrospect?

meain (a GitHub user) maintains it in meain/dotfiles, which has 285 GitHub stars. The repository holds 35 skills in this directory. The repository was last updated on October 8, 2026.

Source: meain/dotfiles on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.