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

Codetrial Contribute

by sysprog21 in sysprog21/codetrial

Help CodeTrial contributors turn observations into clear English GitHub issues, small contribution plans, and pull request descriptions, and see them through.

MITAuto-check passedDevelopment

Install Codetrial Contribute

skills CLI
$ npx skills add sysprog21/codetrial --skill codetrial-contribute -a claude-code

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

GitHub CLI
$ gh skill install sysprog21/codetrial codetrial-contribute --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/sysprog21/codetrial.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/codetrial-contribute .claude/skills/codetrial-contribute && 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
codetrial-contribute
GitHub stars
149
Token cost
~3.8k tokens
SKILL.md length
2,156 words
Files
1
Skills in repo
4
Repo updated
First seen
Licence
MIT

At a glance

Help CodeTrial contributors turn observations into clear English GitHub issues, small contribution plans, and pull request descriptions, and see them through.

  • Works in 4 steps: The path: which file and function handle… → The evidence: the log line, measurement… → One hypothesis, labeled as such, and the… → …
  • File a CodeTrial issue
  • SKILL.md covers Draft or improve an issue, Get the facts a maintainer…, Write the initial analysis a… and Make a first contribution…, plus 4 more sections
  • Calls git, gh and cargo

What it does

Codetrial Contribute is an agent skill from sysprog21/codetrial. Help CodeTrial contributors turn observations into clear English GitHub issues, small contribution plans, and pull request descriptions, and see them through. Use to draft, improve or file a CodeTrial issue, answer what a maintainer asked on it (a revision, a log, an initial analysis, a retest), prepare a PR description, respond to review requests such as rebase, squash or commit message fixes, check which of your own issues and PRs are waiting on you, ask a focused maintainer question, or choose a bounded first…

Its SKILL.md is about 3.8k 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 Development, covering Pull requests and Commit messages. It works with GitHub. The repository describes itself as: A live technical interview simulator. The licence is MIT.

When your agent uses it

  • File a CodeTrial issue
  • Answer what a maintainer asked on it (a revision
  • An initial analysis
  • Prepare a PR description

Example prompts

  • “/codetrial-contribute”

Requirements

  • Python 3

Workflow steps

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

  1. The path: which file and function handle the behavior, found with git grep
  2. The evidence: the log line, measurement or state that shows where the
  3. One hypothesis, labeled as such, and the experiment that would refute it.
  4. Follow-up actions, each small enough to be a commit or a question.

What it can do on your machine

Read from SKILL.md and the folder at commit a77b0db. 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
    • gh
    • cargo
    • make
    • jq

    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

Codetrial Contribute loads about 3.8k tokens when it runs. Until then it costs about 190 tokens; SKILL.md has 2,156 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~190
When it runs · the whole SKILL.md, loaded when a task matches
~3.8k

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 sysprog21/codetrial at commit a77b0db, republished under its MIT licence (© sysprog21). 2,156 words, ~3,786 tokens.

Download SKILL.mdSave it as .claude/skills/codetrial-contribute/SKILL.md (or your agent's skills folder).
name
codetrial-contribute
description
Help CodeTrial contributors turn observations into clear English GitHub issues, small contribution plans, and pull request descriptions, and see them through. Use to draft, improve or file a CodeTrial issue, answer what a maintainer asked on it (a revision, a log, an initial analysis, a retest), prepare a PR description, respond to review requests such as rebase, squash or commit message fixes, check which of your own issues and PRs are waiting on you, ask a focused maintainer question, or choose a bounded first contribution with an AI agent. The deliverable is copy-ready text; judging an existing backlog for duplicates is codetrial-triage. Opening a PR goes through gh-submit when it is installed; this skill covers it otherwise.

Contribute to CodeTrial

CodeTrial is a mock interview platform, and contributing with an AI agent is a supported way to learn. Start from the contributor's observation or intended outcome. Explain unfamiliar terms briefly in their language, then produce an English issue or PR draft they can understand and check. Never make a questionnaire, skills survey or quiz a prerequisite for help.

Titles, body formatting, issue references and redaction follow CONTRIBUTING.md; the rules for writing to GitHub, approval included, are in the GitHub section of codetrial-conventions. Keep technical identifiers and quoted errors exact.

Example requests: "Turn these notes into an issue", "Help me make my first CodeTrial contribution", or "Write a PR description from this diff". Produce the artifact the user asked for.

Draft or improve an issue

Check existing open and closed issues and related PRs first, using the retrieval and comparison workflow in codetrial-triage. If a report already covers the problem, offer a focused addition to that thread with new evidence. When the deliverable is instead a verdict on someone else's thread, that comment is codetrial-triage's. If GitHub access is unavailable, still prepare the draft and explicitly mark the duplicate search as unverified.

Use the information already available in notes, logs and the checkout. Ask only for missing facts that change the report's meaning. Never invent reproduction steps, versions, root causes or successful tests. A plausible explanation is a hypothesis; observed behavior is evidence.

Choose the smallest useful structure:

  • Bug: concrete symptom, expected and actual results, numbered minimal reproduction steps, relevant version/environment, and a short sanitized error or other evidence. Identify the build by revision rather than by date, since published binaries move under a rolling tag (docs/install.md); triage has why that makes a date one-directional. Ask first what the binary reports about itself; when it reports nothing, say the revision is unknown and give the exact download timestamp, which still bounds the build from above. For an interview failure, .github/ISSUE_TEMPLATE/bug.yml is the field list; ask only for the ones that matter. The problem language matters because Python and JavaScript run in the browser while C, C++ and Java run remotely, so one symptom means two different things across that split. Record frequency or regression range when known. Unknown details can stay explicitly unknown; a useful report need not diagnose a fix. Retest on the current latest release or main before filing: the lobby and the interviewer change daily, and #146 and #171 both reported behavior that had already been fixed after the reporter's download.
  • Feature: who is affected (candidate, operator or contributor) and in which workflow, desired outcome, present workaround, scope and observable acceptance criteria. Keep a suggested implementation separate from the need it serves.
  • Documentation: page or section, what is missing or misleading, and the intended correction or reader outcome. Do not demand runtime logs for prose.

Do not split a single problem across multiple issues just to fill categories. Do not combine two problems in one issue either; #151 and #205 each had to be split after the fact.

Get the facts a maintainer asks for

These are the requests that recur on this tracker. The contributor may never have done any of them, so give the exact steps for their platform, run what can be run in the checkout, and check the result before it is posted.

Asked forHow to get it
The revisioncodetrial --version with the binary that ran, or cargo run -- --version in a checkout. It prints codetrial 0.1.0 (<sha>); post that line.
The revision of a binary too old for --versionBuilds before 3ec7cb4 (2026-09-29) print nothing. A saved report's header shows Contract: bundle N. Put that number in place of N in git log -S'INTERVIEW_CONTRACT_BUNDLE_VERSION: u32 = N;' --format='%h %cI' -- src/agent.rs, which then gives the commit that set it and the one that replaced it; the build lies between them, as #188 shows. Without a report, give the exact download time, which bounds the build from above.
The agent terminal outputThe terminal window where codetrial is running prints the agent log. Copy the lines around the failure, starting at starting interview: room=.... Remove API keys and tokens, and every line that quotes what the candidate said: personal interview content stays out of a public thread even when the contributor would not mind.
The browser consoleF12 or Ctrl+Shift+J (Cmd+Option+J on macOS) in Chrome and Edge, Ctrl+Shift+K in Firefox; Safari needs Settings, Advanced, "Show features for web developers" first. Copy the text rather than a screenshot.
A retest of an open PRFrom a checkout: gh pr checkout <N> --repo sysprog21/codetrial, or git fetch https://github.com/sysprog21/codetrial.git pull/<N>/head:pr-<N> && git switch pr-<N>, then cargo run -- web. Report the result on the PR, not on the issue.
A retest after a mergeDownload the latest release again, or git pull on main, and confirm --version names the merge commit or later.
An edit to the issueOn the issue page, the ... menu on the first post, then Edit; or gh issue edit <N> --repo sysprog21/codetrial --body-file <file>. Edit in place rather than filing again.

Write the initial analysis a maintainer asked for

"Provide an initial analysis and a list of follow-up actions" asks the contributor to read the code, not the UI. A useful answer has four parts, and #66, #133 and #211 are worked examples on this tracker:

  1. The path: which file and function handle the behavior, found with git grep on a log line or a UI string the contributor saw. Give path:line at a named revision.
  2. The evidence: the log line, measurement or state that shows where the behavior diverges from the expected one.
  3. One hypothesis, labeled as such, and the experiment that would refute it.
  4. Follow-up actions, each small enough to be a commit or a question.

An agent's summary of the codebase is not an analysis. A maintainer rejected one on #66 because it had no motivation and no evidence. Post the analysis as a comment, or fold it into the body, then say which one you did.

Return a copy-ready title and body, followed by any remaining questions outside the draft. Link related work and explain whether it overlaps or differs. For a maintainer question, give the context, what was inspected or attempted, and one specific decision needed to continue.

Make a first contribution manageable

Offer one bounded task matched to the contributor's stated experience and time, or a small first step when neither is given. Inspect current code and tests before naming likely files; a good first issue label alone is not proof of suitable scope.

Give the contributor:

  • a visible result, such as a minimal reproduction, focused docs correction or regression test and fix;
  • likely entry points and a concrete acceptance check;
  • a stopping condition and fallback, such as drafting the exact maintainer question if the expected behavior is still undecided.

Keep unrelated refactoring out of the task. Do not require filing an issue for every small, well-understood fix. For uncertain product direction, establish the desired behavior before investing in a large implementation. When coaching, help the contributor explain what fails before and succeeds after the change.

Prepare a pull request people can review

Read the full branch diff against its intended base, relevant commits and linked issues. Check that the branch contains the intended work only, and use a topic branch as described in conventions. Preserve unrelated working-tree edits. AI-generated summaries are starting points: validate every claim against the diff and actual test results. Do not paste a conversation transcript.

Use codetrial-verify for the appropriate validation and codetrial-web when browser or wire behavior is involved. Report commands and outcomes, including failures and skipped or unrun checks with reasons. Local success is not evidence that remote CI passed.

The description should cover:

  • the concrete problem and resulting behavior, with a before/after example when it clarifies the change;
  • the essential implementation choice or tradeoff a reviewer needs to assess;
  • validation evidence and material limitations; include screenshots or sanitized logs only when they help verify the result;
  • relevant compatibility or follow-up work, then the issue reference line.

For a simple PR, one or two paragraphs plus validation is enough. Rewrite the title and body when scope changes.

Show full SKILL.md (793 more words)Show less

Publish only after the user sees the text

A request to file an issue or open a PR authorizes preparing it; posting still waits for the approval described in conventions. Show the target repository and the title and body verbatim. Feedback that changes the draft without saying to post it gets the revised draft shown again; "change the title to X and submit" approves the result and needs no second round.

To open a PR, pass the drafted title and body to gh-submit when it is installed, and tell it the base is sysprog21/codetrial, since on a fork its origin is the contributor's copy. Its single confirmation of repository, branches, commits, title and body is the approval; do not ask separately first. Without gh-submit, show the same five things, create the PR with --repo sysprog21/codetrial --head <fork-owner>:<branch> after the yes, and follow with gh pr checks <N> --repo sysprog21/codetrial --watch so the contributor learns whether CI passed rather than just receiving a URL.

For an issue, repeat a narrow search on the title terms just before creating it, in case someone filed it meanwhile:

sh
gh issue list --repo sysprog21/codetrial --state all --search '<title terms>' --limit 5 --json number,title
gh issue create --repo sysprog21/codetrial --title "$(cat "$SCRATCH/title.txt")" \
  --body-file "$SCRATCH/issue.md"

Finish with the verified URL and the next concrete step, including any outstanding evidence the contributor can supply. Tell the contributor that filing starts the work rather than finishing it: a maintainer usually asks something within a day, and an issue whose question goes unanswered for two weeks is closed.

Follow through on what you filed

When a contributor asks what needs their attention, or comes back to an issue or PR, start from the same snapshots triage uses, filtered to them:

sh
me=$(gh api user --jq .login)
.claude/skills/codetrial-triage/issue-sweep.sh |
  jq --arg me "$me" '.waiting[] | select(.author == $me)'
gh pr list --repo sysprog21/codetrial --author "@me" \
  --json number,title,reviewDecision,mergeable,updatedAt

pr-sweep.sh is no help for one's own pull requests: it leaves out the viewer's PRs, since its nudges and rebase requests are a maintainer's to send. A PR whose reviewDecision is CHANGES_REQUESTED or whose mergeable is CONFLICTING is waiting on its author.

For each hit, read the newest maintainer comment and do what it asks, using the table above. Answer it in that thread, quoting the question when the thread has several. A fact that belongs in the report also goes into the body by editing it, so the next reader does not need the comments. When a maintainer links a PR that may fix the issue, retest it and say on the PR whether it does. When a fix merged and the retest passes, say so with the --version line, and close the issue if a maintainer has not. An open issue nobody can confirm stays on everyone's list.

Answer a review

Most review requests here are about git and the commit message, not about the code, and they recur. What each asks:

  • "Rebase latest main and resolve conflicts": find the remote that points at sysprog21/codetrial in git remote -v. On a fork it is usually missing, so add it with git remote add upstream https://github.com/sysprog21/codetrial.git; in a clone of upstream itself it is origin, so use that name wherever upstream appears below. Then git fetch upstream && git rebase upstream/main, resolve each conflict, ./scripts/test.sh, and git push --force-with-lease. A merge commit from main is not a rebase.
  • "Squash" or "fold similar commits": one commit per functional change, with the review fixups folded into the commit they fix. git rebase -i upstream/main is the tool. An agent that cannot drive an interactive editor can run git reset --soft "$(git merge-base HEAD upstream/main)" and commit again, but only when the result is meant to be a single commit. Show the contributor the new git log before pushing.
  • "Read cbea.ms/git-commit" or "refine commit messages": run ./scripts/git-commit-msg.sh --rules, then make hooks so the next commit is checked before it leaves the machine. The body says what and why. The conversation that produced the change stays out of it, and so do Co-Authored-By trailers and "Generated with" lines for an AI tool.
  • "Run make indent": run it, check that git diff shows only formatting, and amend the commit that introduced the code rather than adding a "format" commit.
  • "Append Closes #N" or "avoid Refs": put Closes #N alone on the last line of the commit message as well as the PR body, when the change finishes the issue. Refs #N is for a change that does not, and then the body says what is left; a bare Refs: #60 told the reviewer nothing.
  • A question in a review thread: answer it in that thread. A changed line answers itself, so resolve the thread rather than replying "done".

Replies follow the register in CONTRIBUTING.md, and maintainers here enforce it in public. A reply opens with the fact. It carries no greeting, thanks, apology ("Sorry for the delay"), self-introduction ("my first issue"), or list of what the last push changed. Maintainers called out each of those on #83, #94, #123, #185 and #60.

© sysprog21, 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 .claude/skills/codetrial-contribute of sysprog21/codetrial.

Open the folder on GitHubat commit a77b0db

Compare with similar skills

Codetrial Contribute 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.

Codetrial Contribute compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Codetrial Contribute this skillsysprog21/codetrial149—~3.8kAutomated safety check: PassMIT
PR Finalize Reviewmicrosoft/garnet12k—~3.1kAutomated safety check: PassMIT
React Router Pull Request Creatorremix-run/react-router57k—~2.5kAutomated safety check: PassMIT
Verdaccio Pull Request Workflowverdaccio/verdaccio18k—~1.9kAutomated safety check: PassMIT
Gh PR Metadatamodule-federation/core2.7k—~1kAutomated safety check: PassMIT
Draft Pull Request Creatorwordpress-mobile/WordPress-Android3.2k—~881Automated safety check: NotesGPL-2.0

Similar skills

  • PR Finalize Review

    microsoft/garnet

    Official

    Checks that a pull request's title and description match its implementation and reviews the code for Garnet best practices, reporting findings without posting them.

    12k GitHub stars~3.1k tokensUpdated today
    DevelopmentAuto-check passed

  • React Router Pull Request Creator

    remix-run/react-router

    Packages finished React Router work into a draft pull request: branch, commit, push, a written PR body and the right GitHub labels.

    57k GitHub stars~2.5k tokensUpdated yesterday
    DevelopmentAuto-check passed

  • Verdaccio Pull Request Workflow

    verdaccio/verdaccio

    Takes a change through a verdaccio pull request: branch, local checks, changeset, title and body, labels, CI and review rounds, and ports to other release lines.

    18k GitHub stars~1.9k tokensUpdated yesterday
    DevelopmentAuto-check passed

  • Gh PR Metadata

    module-federation/core

    Update the current GitHub PR title and body for this repository so they match the repo's pull request template and conventional-commit title style.

    2.7k GitHub stars~1k tokensUpdated today
    DevelopmentAuto-check passed

  • Draft Pull Request Creator

    wordpress-mobile/WordPress-Android

    Commits and pushes current changes, writes a pull request title and body from the branch history and template, and opens a draft PR on GitHub after you approve it.

    3.2k GitHub stars~881 tokensUpdated today
    DevelopmentAuto-check: notes

  • Submit PR

    tisfeng/Easydict

    创建或复用当前 checkout 已提交变更的 GitHub PR，必要时推送任务分支。PR 审查使用 review-pr；仅本地提交使用 git-commit。

    15k GitHub stars~595 tokensUpdated 2 days ago
    DevelopmentAuto-check passed

More from sysprog21/codetrial

  • Codetrial Conventions

    sysprog21/codetrial

    The CodeTrial conventions the gate does not settle on its own, several of them enforced by the git hooks instead - the register a comment, a commit message, an issue or PR title and a PR reply are…

    149 GitHub stars~2.3k tokensUpdated today
    Auto-check passed

  • Codetrial Verify

    sysprog21/codetrial

    How a CodeTrial change is validated - scripts/test.sh as the credential-free gate, which generated artifacts have to be regenerated before it passes, the checks that need credentials or a browser…

    149 GitHub stars~2.2k tokensUpdated today
    Auto-check passed

  • Codetrial Web

    sysprog21/codetrial

    The CodeTrial browser half and the boundary it talks across - why web/ has no build step, how a file reaches the browser embedded or from disk, the vendored checksum-pinned assets, the data-channel…

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

Works with

Categories

Questions about Codetrial Contribute

What does Codetrial Contribute do?

Help CodeTrial contributors turn observations into clear English GitHub issues, small contribution plans, and pull request descriptions, and see them through. Codetrial Contribute is an agent skill from sysprog21/codetrial. Help CodeTrial contributors turn observations into clear English GitHub issues, small contribution plans, and pull request descriptions, and see them through.

When should I use Codetrial Contribute?

Codetrial Contribute fits situations like: file a CodeTrial issue; answer what a maintainer asked on it (a revision; an initial analysis; prepare a PR description.

How do I install Codetrial Contribute in Claude Code?

Run `npx skills add sysprog21/codetrial --skill codetrial-contribute -a claude-code`. Or copy the skill folder (.claude/skills/codetrial-contribute in sysprog21/codetrial) into .claude/skills/codetrial-contribute in your project. Claude Code loads it when a task matches its description.

How do I install Codetrial Contribute in Codex?

Run `npx skills add sysprog21/codetrial --skill codetrial-contribute -a codex`. Or copy the skill folder (.claude/skills/codetrial-contribute in sysprog21/codetrial) into .agents/skills/codetrial-contribute in your project. Codex loads it when a task matches its description.

Can I use Codetrial Contribute 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 sysprog21/codetrial --skill codetrial-contribute -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/codetrial-contribute, .gemini/skills/codetrial-contribute, .github/skills/codetrial-contribute and .opencode/skills/codetrial-contribute in your project.

What does Codetrial Contribute need to run?

Going by SKILL.md and its folder, Codetrial Contribute needs the command-line tools its instructions call (git, gh, cargo, make and jq). Our summary lists: Python 3.

Does Codetrial Contribute 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 Codetrial Contribute 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 Codetrial Contribute use?

Codetrial Contribute 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 Codetrial Contribute use?

About 3.8k tokens (SKILL.md is roughly 15k 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 Codetrial Contribute?

Skills that share tags, products or a category with Codetrial Contribute: PR Finalize Review (microsoft/garnet, 12k stars), React Router Pull Request Creator (remix-run/react-router, 57k stars), Verdaccio Pull Request Workflow (verdaccio/verdaccio, 18k stars) and Gh PR Metadata (module-federation/core, 2.7k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Codetrial Contribute?

sysprog21 (a GitHub organization) maintains it in sysprog21/codetrial, which has 149 GitHub stars. The repository holds 4 skills in this directory. The repository was last updated on October 7, 2026.

Source: sysprog21/codetrial on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.