Agent skill

Tuicr

by agavra in agavra/tuicr

Use tuicr's review CLI to read and add comments in active TUI review sessions, and launch tuicr in cmux, tmux, Zellij, or Herdr when a user needs an interactive review pane.

MITAuto-check passedDevelopment

Install Tuicr

skills CLI
$ npx skills add agavra/tuicr --skill tuicr -a claude-code

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

GitHub CLI
$ gh skill install agavra/tuicr tuicr --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/agavra/tuicr.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/tuicr .claude/skills/tuicr && 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
tuicr
GitHub stars
3.3k
Token cost
~3.2k tokens
SKILL.md length
1,716 words
Files
6
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Use tuicr's review CLI to read and add comments in active TUI review sessions, and launch tuicr in cmux, tmux, Zellij, or Herdr when a user needs an interactive review pane.

  • Works in 2 steps: User-led review of agent-generated changes → Agent review of an AI-generated patch
  • Development work in your project
  • SKILL.md covers Core Rule, Attach To A Session, Start A Session and Reconstruct The Diff, plus 6 more sections
  • Runs Shell scripts from its folder; calls git and gh

What it does

Tuicr is an agent skill from agavra/tuicr. Use tuicr's review CLI to read and add comments in active TUI review sessions, and launch tuicr in cmux, tmux, Zellij, or Herdr when a user needs an interactive review pane.

Its SKILL.md is about 3.2k tokens, which your agent loads only when the skill is triggered. The skill folder holds 5 other files (for example `_tuicr-common.sh`, `tuicr-wrapper-cmux.sh` and `tuicr-wrapper-herdr.sh`).

It sits in Development. It works with tmux and Rust. The repository describes itself as: a code review TUI with vim keybindings. The licence is MIT.

When your agent uses it

  • Development work in your project

Example prompts

  • “/tuicr”

Requirements

  • A Bash shell

Workflow steps

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

  1. User-led review of agent-generated changes
  2. Agent review of an AI-generated patch

What it can do on your machine

Read from SKILL.md and the folder at commit 095c16d. 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

    Ships script files (Shell), which the agent can run.

    Shell commands in SKILL.md call:

    • git
    • gh

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

  • Network

    No URLs in SKILL.md. Its commands use git and gh, 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

Tuicr loads about 3.2k tokens when it runs. Until then it costs about 45 tokens; SKILL.md has 1,716 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~45
When it runs · the whole SKILL.md, loaded when a task matches
~3.2k

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 agavra/tuicr at commit 095c16d, republished under its MIT licence (© agavra). 1,716 words, ~3,219 tokens.

Download SKILL.mdSave it as .claude/skills/tuicr/SKILL.md (or your agent's skills folder). This skill also uses 5 other files; get the full folder from GitHub.
name
tuicr
description
Use tuicr's review CLI to read and add comments in active TUI review sessions, and launch tuicr in cmux, tmux, Zellij, or Herdr when a user needs an interactive review pane.

tuicr Review Workflow

Use tuicr review as the default agent interface. The TUI is where the human reviews code; the CLI is how the agent discovers active sessions, reads user comments, and, only when appropriate, adds agent-authored comments.

Core Rule

First decide which workflow the user is asking for:

  1. User-led review of agent-generated changes

    • The user wants to inspect the patch and write comments in tuicr.
    • Your job is to open or find the session, then retrieve the user's comments with tuicr review comments when they say comments are ready. If you are explicitly waiting while the user reviews, poll the same command periodically and look for new comment IDs.
    • Do not add your own review comments, do not preemptively review your own patch, and do not impersonate the user's comments.
  2. Agent review of an AI-generated patch

    • The user wants you to understand, critique, or summarize a patch.
    • You may inspect the patch and propose findings.
    • If you can confidently identify this workflow and the target session, add findings directly with tuicr review add and an explicit --username identifying the agent. Ask first when the workflow or session is ambiguous.

If the user's intent is ambiguous, ask which workflow they want.

Attach To A Session

  1. Determine the repository directory from the user's request, current working directory, or recent file operations. Ask if it is ambiguous.

  2. List persisted sessions:

    bash
    tuicr review list --repo /path/to/repo   # checkout + its repo's PR sessions
    tuicr review list --repo owner/repo      # all sessions for a forge repo
    tuicr review list --all                  # every session across all repos

    --repo is a selector: a checkout path also surfaces PR sessions for that checkout's origin repo, and a forge coordinate like owner/repo matches local and PR sessions by owner/repo. Each row carries a kind (local or pr) and a usable slug. Use --all when you don't know the repo.

    [] with exit 0 also means "not a repo root" — a subdirectory returns it too. Pass the root, then --all, before concluding nothing is open.

  3. Choose the session:

    • If the CLI clearly reports exactly one relevant active session with "active": true, attach to it.
    • If multiple sessions are active, or the correct session is not clear, ask the user which slug to use. One repo can hold a worktree and a commit-range session at once, and adding to the wrong one exits 0.
    • If the user provided a slug or session JSON path, use it directly.
    • For a PR review, pass the PR slug from the listing (e.g. gh:owner/repo/pr/N) to --session; it is self-contained and needs no --repo.
    • If there is no active session, start or wait for one as described below.
    • Until active-session discovery is formalized as a stable protocol, treat "active": true as a convenience signal. If slug resolution fails, ask the user for the slug or repo path used by the session.

The CLI works even if the agent is not running inside tmux, Zellij, or Herdr, so do not require a multiplexer just to connect to an existing active session.

Start A Session

When the user needs an interactive tuicr pane and no active session exists:

EnvironmentAction
$CMUX_WORKSPACE_ID is setRun tuicr-wrapper-cmux.sh /path/to/repo -- <scope>
$TMUX is setRun tuicr-wrapper.sh /path/to/repo -- <scope>
$ZELLIJ is setRun tuicr-wrapper-zellij.sh /path/to/repo -- <scope>
$HERDR_ENV is 1Run tuicr-wrapper-herdr.sh /path/to/repo -- <scope>
None is setTell the user you are waiting for them to start tuicr in the repo, then attach with tuicr review list after they say it is ready

<scope> is -w for uncommitted working-tree changes or -r <revset> for a commit range — always pass one explicitly so the user is never left to pick staged/unstaged/commit-range manually in the TUI.

If more than one multiplexer marker is set, prefer the innermost multiplexer if that is clear; otherwise ask. cmux hosts a Ghostty terminal, so $TERM_PROGRAM reads ghostty inside cmux — check $CMUX_WORKSPACE_ID, not the terminal name.

tuicr supports both git and Jujutsu (jj) repositories, and jj workspaces may have no .git directory at all. Do not pre-check the directory with git rev-parse or refuse to launch because git does not recognize it; always run the wrapper and let it validate the repository.

Wrapper paths are relative to this skill directory:

bash
<skill-directory>/tuicr-wrapper-cmux.sh /path/to/repo -- -w
<skill-directory>/tuicr-wrapper.sh /path/to/repo -- -w
<skill-directory>/tuicr-wrapper-zellij.sh /path/to/repo -- -w
<skill-directory>/tuicr-wrapper-herdr.sh /path/to/repo -- -w

The Herdr wrapper requires jq to read pane IDs and completion results from Herdr's JSON responses.

Every wrapper accepts pass-through tuicr arguments after --, which is how you scope the review instead of leaving the scope selector for the user to fill in — for example -- -w for uncommitted working-tree changes or -- -r <revset> for a commit range. Always pass one of these explicitly when launching a review pane.

If your tool supports command timeouts, use a long timeout, such as 10 minutes, because the tmux, Zellij, and Herdr wrappers wait for the TUI to exit. The cmux wrapper is the exception: it returns as soon as the pane is running and prints the new surface ref between === TUICR SURFACE === markers. Capture that ref — it is how you close the pane later with cmux close-surface --surface <ref>. Once the TUI creates its active session, use tuicr review list --repo /path/to/repo to capture the slug. If your environment cannot run another command while a blocking wrapper is waiting, read the comments after the user exits tuicr.

Reconstruct The Diff

To review a patch yourself, rebuild the diff the user sees. The slug's source segment says which:

Slug segmentDiff
worktree/<head>, staged-and-unstaged/<head>git diff HEAD
staged/<head>git diff --cached
unstaged/<head>git diff
commits/<base>..<head>git diff <base>~1..<head>
pr/<n>gh pr diff <n>
pristinenone; every tracked file shown in full

Range endpoints are inclusive and printed oldest-first, so <base> without ~1 drops the first commit. Check the file count against the listing row's file_count; a mismatch means every line number you derive will be wrong.

Show full SKILL.md (786 more words)Show less

Read User Comments

This is the main review loop for user-led review.

There is no push stream from tuicr to the agent. Read comments by running the CLI on demand. After the user says comments are ready, or after the TUI exits, run:

bash
tuicr review comments --repo /path/to/repo --session <slug>

The command emits JSON. Each comment includes fields like:

  • id
  • location
  • path
  • start_line
  • end_line
  • side
  • comment_type
  • author
  • lifecycle_state
  • content

Treat these comments as the user's review feedback:

  • issue: blocking problem to fix first
  • suggestion: consider implementing or explain why not
  • note: answer or acknowledge
  • praise: no action required

If you are waiting during an active review, poll this command about every 30 seconds and compare comment IDs with the previous result. Read immediately when the user says comments are ready. Stop polling once the user says the review is done or your tooling would block other work.

An empty result does not by itself mean the review didn't happen. On exit, tuicr always prints a line like tuicr-summary: reviewed 3/3 files, 0 comments added to stderr (visible in the pane's scrollback), and tuicr review list reports the same reviewed_count/file_count for the session. If reviewed_count equals file_count, zero comments is a legitimate "nothing to flag" outcome — treat the review as complete, don't ask the user to confirm. Only ask whether the user saved comments in the intended session, or whether another active session should be selected, when reviewed_count is less than file_count (the user quit before reviewing everything) or you can't find a tuicr-summary: line at all. If the review may have continued while you were working, rerun tuicr review comments before claiming completion.

Add Agent Comments

Only add comments when the workflow allows it and, for agent-authored review, after the user approves writing them into tuicr.

Defaults:

  • Prefer line comments when a specific file and line are known.
  • Use file comments for file-scoped feedback.
  • Use review-level comments only for whole-review summaries.
  • Use --type issue for problems by default.
  • Use suggestion, note, or praise when that better matches the intent.
  • Pass --username so agent comments are visually distinguishable.

Examples:

bash
tuicr review add --repo /path/to/repo --session <slug> \
  --target-file src/main.rs \
  --line 42 \
  --side new \
  --type issue \
  --username "Codex" \
  "Handle the empty case here."
bash
tuicr review add --repo /path/to/repo --session <slug> \
  --target-file src/main.rs \
  --type suggestion \
  --username "Codex" \
  "Consider splitting this file-level concern into a helper."

Omit --target-file for a review-level comment. Add --end-line for a range comment. Use --side old for removed lines and --side new for added or unchanged lines in the new file.

For structured input, use --input with literal JSON, @path/to/file.json, or - for stdin. Supported target types are review, file, line, and line_range. One object per call — an array is a parse error. The file key is file, not path. target.type is inferred from the fields present:

bash
tuicr review add --session <slug> --username "Codex" --input \
  '{"file":"src/main.rs","line":42,"side":"new","comment_type":"issue","content":"Handle the empty case."}'

Then verify. A line outside the diff stores, prints back, and exits 0, but never renders — invisible to the user, successful-looking to you. Re-read tuicr review comments and check each start_line exists on the side you gave (new for added or unchanged, old for removed). Check author to distinguish your comments from the user's, and keep the returned ids to identify the exact comments in later reads.

Legacy Export Output

Older wrapper-driven flows may emit:

text
=== TUICR INSTRUCTIONS ===
...
=== END TUICR INSTRUCTIONS ===

If present, process those instructions. Otherwise prefer tuicr review comments; it is the primary source of review feedback. If the wrapper mentions clipboard export, ask the user to paste it only when the CLI comments are unavailable.

Multiplexer Tips

cmux:

  • Switch panes: click the pane, or cmux focus-pane --pane <ref>
  • Close tuicr: press q; the pane closes itself. Force it with cmux close-surface --surface <ref>
  • List panes: cmux list-panes
  • Read a pane without focusing it: cmux read-screen --surface <ref>

tmux:

  • Switch panes: Ctrl-b then arrow keys
  • Close tuicr: press q
  • Resize panes: Ctrl-b then Ctrl-arrow
  • Zoom pane: Ctrl-b then z

zellij:

  • Switch panes: Alt + arrow keys
  • Close tuicr: press q
  • Resize panes: Ctrl-n, then arrow keys
  • Toggle fullscreen: Alt-f
  • Cycle stacked panes: Alt + [ / ]

Herdr:

  • Select a pane: click it in the Herdr UI
  • Close tuicr: press q; the wrapper then closes the review pane

Error Handling

SituationAction
Multiple plausible active sessionsAsk which session slug to use
No active session, cmux/tmux/Zellij/Herdr availableStart a new tuicr pane with the matching wrapper
No active session, no multiplexerTell the user you are waiting for them to start tuicr
cmux wrapper printed no surface refRun cmux list-panes to find the pane, or ask the user to start tuicr themselves
tuicr not installedTell the user to install tuicr
Not a repositoryAsk for the correct repo directory
Comments are empty, but reviewed_count == file_countTreat as a completed review with nothing to flag — don't ask
Comments are empty and reviewed_count < file_countConfirm the selected session or ask the user to save/add comments

When Not To Use

  • The user only wants raw git diff output.
  • The user explicitly asks for a non-tuicr review workflow.
  • The task is remote PR review and no tuicr PR session is involved.

© agavra, MIT. 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 5 other files in skills/tuicr of agavra/tuicr.

  • SKILL.md
  • _tuicr-common.sh
  • tuicr-wrapper-cmux.sh
  • tuicr-wrapper-herdr.sh
  • tuicr-wrapper-zellij.sh
  • tuicr-wrapper.sh

Open the folder on GitHubat commit 095c16d

Compare with similar skills

Tuicr 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.

Tuicr compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Tuicr this skillagavra/tuicr3.3k—~3.2kAutomated safety check: PassMIT
Luvus ModuleRizRiyz/luvus959—~2.5kAutomated safety check: PassApache-2.0
TUI Change Verification for Warpwarpdotdev/warp65k1 repos~5.5kAutomated safety check: NotesAGPL-3.0
Migrate Core Code to Submodulestinyhumansai/openhuman42k—~2.6kAutomated safety check: PassGPL-3.0
OpenLogi macOS Permissions TriageAprilNEA/OpenLogi23k—~2.5kAutomated safety check: NotesApache-2.0
Rust Best Practicesfarm-fe/farm5.6k3 repos~1.1kAutomated safety check: PassMIT

Similar skills

  • Luvus Module

    RizRiyz/luvus

    Write a luvus module (an extension for luvus, mission control for your AI coding agents).

    959 GitHub stars~2.5k tokensUpdated 3 days ago
    DevelopmentAuto-check passed
  • Runs Warp's terminal UI for real, drives it with keystrokes and reads the rendered screen back to confirm that a change looks and behaves as intended.

    65k GitHub starsUsed in 1 repo~5.5k tokens
    Testing & QAAuto-check: notes
  • Migrate Core Code to Submodules

    tinyhumansai/openhuman

    Plans and carries out moving non-host-specific code and its tests from the OpenHuman core into vendored tiny submodule libraries, then releases the submodule and re-pins the host.

    42k GitHub stars~2.6k tokensUpdated yesterday
    DevelopmentAuto-check passed
  • Decides whether an OpenLogi device problem on macOS is a privacy-permission (TCC) problem, using agent log lines, and says which identity needs which grant.

    23k GitHub stars~2.5k tokensUpdated yesterday
    DevelopmentAuto-check: notes
  • Guide for writing idiomatic Rust code based on Apollo GraphQL's best practices handbook.

    5.6k GitHub starsUsed in 3 repos~1.1k tokens
    DevelopmentAuto-check passed
  • Reproduces a feature from Codex or Claude Code in Qwen Code by running the reference agent under capture, reading the traces, then implementing matching behavior.

    28k GitHub stars~1.5k tokensUpdated today
    DevelopmentAuto-check passed

Works with

Categories

Questions about Tuicr

What does Tuicr do?

Use tuicr's review CLI to read and add comments in active TUI review sessions, and launch tuicr in cmux, tmux, Zellij, or Herdr when a user needs an interactive review pane. Tuicr is an agent skill from agavra/tuicr. Use tuicr's review CLI to read and add comments in active TUI review sessions, and launch tuicr in cmux, tmux, Zellij, or Herdr when a user needs an interactive review pane.

When should I use Tuicr?

Tuicr fits situations like: development work in your project.

How do I install Tuicr in Claude Code?

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

How do I install Tuicr in Codex?

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

Can I use Tuicr 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 agavra/tuicr --skill tuicr -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/tuicr, .gemini/skills/tuicr, .github/skills/tuicr and .opencode/skills/tuicr in your project.

What does Tuicr need to run?

Going by SKILL.md and its folder, Tuicr needs a shell for the scripts in its folder and the command-line tools its instructions call (git and gh). Our summary lists: A Bash shell.

Does Tuicr access the network?

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

Is Tuicr 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 Tuicr use?

Tuicr 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 Tuicr use?

About 3.2k tokens (SKILL.md is roughly 13k 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 Tuicr?

Skills that share tags, products or a category with Tuicr: Luvus Module (RizRiyz/luvus, 959 stars), TUI Change Verification for Warp (warpdotdev/warp, 65k stars), Migrate Core Code to Submodules (tinyhumansai/openhuman, 42k stars) and OpenLogi macOS Permissions Triage (AprilNEA/OpenLogi, 23k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Tuicr?

agavra (a GitHub user) maintains it in agavra/tuicr, which has 3,328 GitHub stars. The repository was last updated on October 9, 2026.

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