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

Codex Session Product Reviewer Maintainer

by Undertone0809 in Undertone0809/rudder

A skill your agent uses when reviewing a local Codex session, task, thread, or commit as a product manager or first-principles reviewer for product correctness, scope, behavior, validation, and…

Apache-2.0Auto-check passedDevelopment

Install Codex Session Product Reviewer Maintainer

skills CLI
$ npx skills add Undertone0809/rudder --skill codex-session-product-reviewer-maintainer -a claude-code

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

GitHub CLI
$ gh skill install Undertone0809/rudder codex-session-product-reviewer-maintainer --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/Undertone0809/rudder.git skills-src && mkdir -p .claude/skills && cp -r skills-src/agent-skills-bak/maintainer/codex-session-product-reviewer-maintainer .claude/skills/codex-session-product-reviewer-maintainer && 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
codex-session-product-reviewer-maintainer
GitHub stars
292
Token cost
~2.3k tokens
SKILL.md length
1,273 words
Files
2 (incl. references)
Skills in repo
30
Repo updated
First seen
Licence
Apache-2.0

At a glance

A skill your agent uses when reviewing a local Codex session, task, thread, or commit as a product manager or first-principles reviewer for product correctness, scope, behavior, validation, and…

  • Works in 5 steps: Locate the Codex session. → Extract the real user request and… → Extract what the agent actually did. → …
  • Reviewing a local Codex session
  • SKILL.md covers Use When, Do Not Use When, Inputs and Evidence Collection, plus 6 more sections
  • Calls git

What it does

Codex Session Product Reviewer Maintainer is an agent skill from Undertone0809/rudder. Use when reviewing a local Codex session, task, thread, or commit as a product manager or first-principles reviewer for product correctness, scope, behavior, validation, and whether the task solved the right user problem.

Its SKILL.md is about 2.3k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files, including reference files (for example `references/runbook.md`).

It sits in Development. The repository describes itself as: Open-source local Agent harness for self-improving agent teams: run agents, review work, and turn feedback into reusable skills. The licence is Apache-2.0.

When your agent uses it

  • Reviewing a local Codex session
  • Commit as a product manager
  • First-principles reviewer for product correctness
  • Whether the task solved the right user problem

Example prompts

  • “/codex-session-product-reviewer-maintainer”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. Locate the Codex session.
  2. Extract the real user request and corrections.
  3. Extract what the agent actually did.
  4. Read the relevant product context.
  5. Check validation evidence.

What it can do on your machine

Read from SKILL.md and the folder at commit 2676a5c. 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

    Shell commands in SKILL.md call:

    • git

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

  • Network

    No URLs in SKILL.md. Its commands use git, which can reach the network depending on how they are called.

    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

Codex Session Product Reviewer Maintainer loads about 2.3k tokens when it runs, and up to ~4.7k if it reads all its reference files. Until then it costs about 66 tokens; SKILL.md has 1,273 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~66
When it runs · the whole SKILL.md, loaded when a task matches
~2.3k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~4.7k

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 Undertone0809/rudder at commit 2676a5c, republished under its Apache-2.0 licence (© Undertone0809). 1,273 words, ~2,348 tokens.

Download SKILL.mdSave it as .claude/skills/codex-session-product-reviewer-maintainer/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
codex-session-product-reviewer-maintainer
description
Use when reviewing a local Codex session, task, thread, or commit as a product manager or first-principles reviewer for product correctness, scope, behavior, validation, and whether the task solved the right user problem.

Codex Session Product Reviewer Maintainer

This skill reviews a completed or in-progress Codex task from a product manager's point of view. It is a reviewer workflow, not an implementation workflow.

The core question is:

Did this session solve the right product problem, with the right scope, behavior, and validation?

Use it to turn a vague request like "review this Codex session as a PM" into a grounded product review based on the session transcript, actual code changes, tests, plan docs, and repository product standards.

Use When

Use this skill when the user asks to:

  • review a Codex session id, task, thread, or run as a product manager
  • evaluate whether another agent's implementation was product-correct
  • apply first-principles thinking to a shipped task
  • judge whether a proposal or implementation solved the right user problem
  • produce a reviewer-style accept / conditional accept / reject
  • translate fuzzy dissatisfaction into explicit product critique

Common trigger phrases:

  • "review 一下 codex session id ..."
  • "as a 专业的产品经理 review"
  • "第一性原理思考一下"
  • "作为 reviewer"
  • "PM review this task"
  • "这个 session 做得怎么样"
  • "这个实现是不是产品上对"

Do Not Use When

Do not use this skill for:

  • generic code review where the user mainly wants bugs and line comments
  • debugging a failed Rudder agent run transcript
  • creating a new product idea or brainstorming from scratch
  • implementing the fixes found during review, unless the user explicitly asks
  • summarizing a session without judgment
  • judging only from the final assistant message when local evidence is available

If the user asks for code correctness review, use a code-review workflow. If the user asks why a Rudder agent run failed, use the run transcript debugging workflow first.

Inputs

Required:

  • A Codex session id, commit hash, PR, branch, or clearly identified task to review.

Optional:

  • The review lens, such as PM, first principles, design, workflow, release, or founder mode.
  • A desired output style, such as short verdict, detailed memo, or findings only.

Evidence Collection

Never start with opinion. Build the evidence packet first.

  1. Locate the Codex session.

    • Search ~/.codex/session_index.jsonl, ~/.codex/sessions/**/*.jsonl, and ~/.codex/archived_sessions/*.jsonl for the session id or prefix.
    • Confirm the session title, timestamp, cwd, branch, and final assistant message.
    • If the session cannot be found, say that and review only the artifacts the user provided.

  2. Extract the real user request and corrections.

    • Ignore injected AGENTS.md, environment context, skill bodies, and system text.
    • Capture each meaningful user prompt, especially mid-task corrections such as scope changes, product constraints, or "also consider..." feedback.

  3. Extract what the agent actually did.

    • Summarize proposals, plans, implementation steps, changed files, commands, validation results, commits, and pushes.
    • Use git show, git diff, plan docs, tests, and relevant source files when a commit or file path is mentioned.
    • Distinguish committed changes from unrelated dirty worktree changes.
    • Compare the changed-file set with the user's actual request. Flag unrelated files or commits mixed into the task, especially skill renames, dependency/version churn, generated artifacts, release files, or broad runtime changes inside a narrow product/UI fix.

  4. Read the relevant product context.

    • Start with doc/product/GOAL.md, doc/product/PRODUCT.md, doc/product/README.md plus relevant doc/product/domains/**, and doc/engineering/DESIGN.md when the task affects product behavior or UI.
    • Read the task's plan doc if one exists.
    • For release, desktop, plugin, database, or CLI work, follow the relevant AGENTS.md doc route.

  5. Check validation evidence.

    • Record which focused tests, full checks, E2E tests, builds, screenshots, or browser checks actually passed.
    • Treat "attempted but timed out" as unverified, not as passed.
    • For visible UI changes, look for real visual verification of the non-empty and edge-case states, not only empty states.

First-Principles Review Frame

Use this frame before writing the verdict:

  1. User job

    • What real user or operator problem was this task supposed to solve?
    • Was the original request a symptom of a deeper workflow or information architecture issue?

  2. Object model

    • What is the product object really? View, navigation shortcut, workflow state, preference, project artifact, run, issue, or setting?
    • Did the implementation model it as the correct kind of object?

  3. Core loop impact

    • Does the change help Rudder complete real agent-work loops end to end?
    • Does it reduce operator friction, improve control, or clarify agent work?

  4. Scope discipline

    • Did the agent delete or avoid unnecessary surface area?
    • Did it create a new concept when an existing one should have been reused?
    • Did it handle the user's explicit edge cases?

  5. Behavioral completeness

    • Are the important entry points covered?
    • Does the behavior remain correct for empty, normal, long, cross-org, mobile, direct-link, and legacy-link states when relevant?

  6. Trust and validation

    • Did the agent prove the change works in the states users actually care about?
    • Are failed or skipped checks clearly separated from passing checks?
Show full SKILL.md (511 more words)Show less

Review Method

Follow this sequence.

  1. State the evidence basis.

    • Mention the session id, title, date, commit hash if present, and files or plan docs inspected.

  2. Give the verdict early.

    • Use one of: accept, conditional accept, reject, or needs more evidence.
    • Add a numeric score only if useful.

  3. Explain the product essence.

    • In one paragraph, say what the task was really about.
    • Translate UI wording into product language. For example, "recent views in main content feels wrong" becomes "navigation history was mis-modeled as a workspace view."

  4. Name what was right.

    • Focus on product decisions, not implementation busywork.
    • Explain why the chosen model fits or does not fit Rudder's operating layer.

  5. Name the gaps.

    • Prioritize product correctness, behavioral gaps, missed edge cases, validation gaps, and scope mistakes.
    • Be concrete: cite files, lines, tests, or session events when possible.
    • Separate "not done" from "not proven."

  6. Recommend the next move.

    • Give one pragmatic follow-up, not a wishlist.
    • Say whether to keep direction, patch behavior, rewrite spec, gather evidence, or reject and reframe.

Output Shape

Default to Chinese when the user asks in Chinese.

Keep the final review compact:

markdown
结论：...

我会给这次任务：7/10，conditional accept。

做得好的地方：
- ...

关键缺口：
1. ...
2. ...

下一步建议：...

For higher-stakes reviews, add:

  • Evidence inspected
  • First-principles diagnosis
  • Product acceptance criteria
  • Residual risks

Do not bury the verdict after a long transcript summary.

Judgment Rules

  • A session can be directionally correct and still not product-done.
  • Passing typecheck/build does not prove product behavior.
  • A failed E2E launch is not a product failure by itself, but it leaves the affected user flow unverified.
  • A visible UI task is not fully reviewed unless at least the relevant rendered states were inspected or explicitly marked unverified.
  • "Too many items" should be reviewed as an experience question, not only as a rendering limit.
  • Compatibility paths should not preserve the old wrong mental model.
  • The best review says which part to keep, which part to patch, and why.

Common Findings To Look For

  • Mis-modeled product object: history treated as a view, settings treated as content, workflow state treated as global preference.
  • Incomplete entry points: only list clicks record behavior, but direct detail, board cards, sidebar links, deep links, or mobile paths do not.
  • Validation mismatch: tests cover the new API but not the actual operator path.
  • Commit hygiene mismatch: the final commit claims to fix one product problem but includes unrelated skill, dependency, generated, or release changes.
  • Edge-case theater: code handles a number limit but not whether that limit is the right product experience.
  • Over-scoped fix: the implementation adds management UI when a small recovery affordance was enough.
  • Under-scoped fix: the implementation moves UI but leaves the underlying behavioral contract inconsistent.

Boundaries

Do not edit code during this review unless the user explicitly asks to fix the findings. If asked to fix, switch from reviewer mode to normal implementation mode and follow repository validation and commit rules.

Do not claim certainty about user satisfaction from logs alone. Phrase such claims as evidence-based judgment, not truth.

Do not quote long chunks of the transcript. Summarize the evidence and cite the local file paths or commit hashes that matter.

© Undertone0809, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file (references) in agent-skills-bak/maintainer/codex-session-product-reviewer-maintainer of Undertone0809/rudder.

  • SKILL.md
  • references/runbook.md

Open the folder on GitHubat commit 2676a5c

Compare with similar skills

Codex Session Product Reviewer Maintainer 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.

Codex Session Product Reviewer Maintainer compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Codex Session Product Reviewer Maintainer this skillUndertone0809/rudder292—~2.3kAutomated safety check: PassApache-2.0
Vercel Composition Patternssupabase/supabase111k58 repos~726Automated safety check: PassMIT
Finishing a Development Branchobra/superpowers297k5 repos~1.9kAutomated safety check: PassMIT
Typescript Advanced Typesrolling-scopes/rsschool-app10k25 repos~4.2kAutomated safety check: PassMPL-2.0
PR Babysitteropeninterpreter/openinterpreter69k3 repos~4.2kAutomated safety check: PassApache-2.0
Code Review ChecklistshareAI-lab/learn-claude-code78k4 repos~1.1kAutomated safety check: PassMIT

Similar skills

  • Vercel Composition Patterns

    supabase/supabase

    Official

    React composition patterns that scale. An agent skill from supabase/supabase.

    111k GitHub starsUsed in 58 repos~726 tokens
    DevelopmentAuto-check passed

  • Finishing a Development Branch

    obra/superpowers

    Walks the last step of a branch: confirm tests pass, detect the git environment, ask how to integrate, carry out your choice and clean up the worktree.

    297k GitHub starsUsed in 5 repos~1.9k tokens
    DevelopmentAuto-check passed

  • Typescript Advanced Types

    rolling-scopes/rsschool-app

    Master TypeScript's advanced type system including generics, conditional types, mapped types, template literals, and utility types for building type-safe applications.

    10k GitHub starsUsed in 25 repos~4.2k tokens
    DevelopmentAuto-check passed

  • PR Babysitter

    openinterpreter/openinterpreter

    Watches an open GitHub pull request until it merges, handling review comments, diagnosing CI failures and retrying flaky checks along the way.

    69k GitHub starsUsed in 3 repos~4.2k tokens
    DevelopmentAuto-check passed

  • Code Review Checklist

    shareAI-lab/learn-claude-code

    Reviews code against a five-part checklist covering security, correctness, performance, maintainability and testing, and reports findings in a fixed format.

    78k GitHub starsUsed in 4 repos~1.1k tokens
    DevelopmentAuto-check passed

  • Greploop

    onyx-dot-app/onyx

    Iteratively improves a PR (GitHub), MR (GitLab), or shelved changelist (Perforce) until Greptile gives it a 5/5 confidence score with zero unresolved comments.

    32k GitHub starsUsed in 4 repos~3.3k tokens
    DevelopmentAuto-check passed

More from Undertone0809/rudder

All 30 skills in this repo

  • Conversation To Skill

    Undertone0809/rudder

    Turn the current conversation's workflow into a reusable agent skill.

    292 GitHub stars~3.6k tokensUpdated today
    Auto-check passed

  • Rudder Worktree Preview Maintainer

    Undertone0809/rudder

    A skill your agent uses when starting the current Rudder checkout as a temporary managed local preview with a stable URL, readiness check, logs, stop command, and cleanup path for manual inspection…

    292 GitHub stars~2.1k tokensUpdated today
    Auto-check passed

  • Stop Rudder Dev Maintainer

    Undertone0809/rudder

    A skill your agent uses when the user explicitly asks to stop, restart, kill, or clean Rudder repo-local pnpm dev processes or local dev runtime residue, including “把 pnpm dev 停了”, “重启 dev”, or “清掉…

    292 GitHub stars~1.3k tokensUpdated today
    Auto-check passed

  • Deep Research

    Undertone0809/rudder

    Conducts enterprise-grade research with multi-source synthesis, citation tracking, and verification.

    292 GitHub stars~1.2k tokensUpdated today
    Auto-check passed

  • Rudder Workspace Hygiene Maintainer

    Undertone0809/rudder

    A skill your agent uses to audit or clean Rudder worktrees, generated artifacts, logs, caches, and repo-owned processes without deleting active work, user data, or unrelated machine state.

    292 GitHub stars~664 tokensUpdated today
    Auto-check passed

  • Visualize

    Undertone0809/rudder

    Create safe inline visual explanations in Rudder Chat. An agent skill from Undertone0809/rudder.

    292 GitHub stars~1.8k tokensUpdated today
    Auto-check passed

Categories

Questions about Codex Session Product Reviewer Maintainer

What does Codex Session Product Reviewer Maintainer do?

A skill your agent uses when reviewing a local Codex session, task, thread, or commit as a product manager or first-principles reviewer for product correctness, scope, behavior, validation, and…. Codex Session Product Reviewer Maintainer is an agent skill from Undertone0809/rudder. Use when reviewing a local Codex session, task, thread, or commit as a product manager or first-principles reviewer for product correctness, scope, behavior, validation, and whether the task solved the right user problem.

When should I use Codex Session Product Reviewer Maintainer?

Codex Session Product Reviewer Maintainer fits situations like: reviewing a local Codex session; commit as a product manager; first-principles reviewer for product correctness; whether the task solved the right user problem.

How do I install Codex Session Product Reviewer Maintainer in Claude Code?

Run `npx skills add Undertone0809/rudder --skill codex-session-product-reviewer-maintainer -a claude-code`. Or copy the skill folder (agent-skills-bak/maintainer/codex-session-product-reviewer-maintainer in Undertone0809/rudder) into .claude/skills/codex-session-product-reviewer-maintainer in your project. Claude Code loads it when a task matches its description.

How do I install Codex Session Product Reviewer Maintainer in Codex?

Run `npx skills add Undertone0809/rudder --skill codex-session-product-reviewer-maintainer -a codex`. Or copy the skill folder (agent-skills-bak/maintainer/codex-session-product-reviewer-maintainer in Undertone0809/rudder) into .agents/skills/codex-session-product-reviewer-maintainer in your project. Codex loads it when a task matches its description.

Can I use Codex Session Product Reviewer Maintainer 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 Undertone0809/rudder --skill codex-session-product-reviewer-maintainer -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/codex-session-product-reviewer-maintainer, .gemini/skills/codex-session-product-reviewer-maintainer, .github/skills/codex-session-product-reviewer-maintainer and .opencode/skills/codex-session-product-reviewer-maintainer in your project.

What does Codex Session Product Reviewer Maintainer need to run?

Going by SKILL.md and its folder, Codex Session Product Reviewer Maintainer needs the command-line tools its instructions call (git).

Does Codex Session Product Reviewer Maintainer access the network?

SKILL.md contains no URLs. Its commands use git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Codex Session Product Reviewer Maintainer 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 Codex Session Product Reviewer Maintainer use?

Codex Session Product Reviewer Maintainer is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Codex Session Product Reviewer Maintainer use?

About 2.3k tokens (SKILL.md is roughly 9.4k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 2.4k tokens, read only when the agent opens those files.

What are the alternatives to Codex Session Product Reviewer Maintainer?

Skills that share tags, products or a category with Codex Session Product Reviewer Maintainer: Vercel Composition Patterns (supabase/supabase, 111k stars), Finishing a Development Branch (obra/superpowers, 297k stars), Typescript Advanced Types (rolling-scopes/rsschool-app, 10k stars) and PR Babysitter (openinterpreter/openinterpreter, 69k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Codex Session Product Reviewer Maintainer?

Undertone0809 (a GitHub user) maintains it in Undertone0809/rudder, which has 292 GitHub stars. The repository holds 30 skills in this directory. The repository was last updated on October 10, 2026.

Source: Undertone0809/rudder on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.