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

Bug Report Triage

by OrchestratorInc in OrchestratorInc/agent-orchestrator

Helps a reporter describe a bug, searches for duplicates and gathers diagnostic evidence kept separate from a short, human-worded issue draft.

Apache-2.0Auto-check passedDevelopment

Install Bug Report Triage

skills CLI
$ npx skills add OrchestratorInc/agent-orchestrator --skill bug-triage -a claude-code

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

GitHub CLI
$ gh skill install OrchestratorInc/agent-orchestrator bug-triage --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/OrchestratorInc/agent-orchestrator.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/bug-triage .claude/skills/bug-triage && 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
bug-triage
GitHub stars
13k
Token cost
~1.8k tokens
SKILL.md length
989 words
Files
2 (incl. scripts)
Skills in repo
5
Repo updated
First seen
Licence
Apache-2.0

At a glance

Helps a reporter describe a bug, searches for duplicates and gathers diagnostic evidence kept separate from a short, human-worded issue draft.

  • Turning a vague chat message into a clear bug report
  • SKILL.md covers Clarify the observation, Keep the report and…, Gather relevant evidence and Check for duplicates, plus 1 more section
  • Runs Python scripts from its folder; calls gh
  • Checking whether a reported bug already exists as an issue

What it does

A short human observation is enough to report a bug, and reproduction steps, logs and diagnosis are optional. The agent reads the report, screenshots and follow-ups first, keeps the reporter's own words and attribution, and does not guess a GitHub username from a chat display name. It asks only questions that would clarify the symptom, a few at a time and never repeated: what happened versus what was expected, where and how often, a screenshot for visual problems and the version and operating system. It never invents steps or makes the reporter diagnose the bug.

The draft has a specific symptom-based title and a short body built only from details the human supplied, with no agent-inferred impact, root cause, confidence scores or proposed fixes. Anything the agent finds goes in a separate section labeled Agent-collected evidence, with the command or read-only query used, the collection time, the version or commit and the observed output, distinguishing the reporter's environment from a test build. A helper script, `scripts/push_fix_to_github.py`, is bundled.

When your agent uses it

  • Turning a vague chat message into a clear bug report
  • Checking whether a reported bug already exists as an issue
  • Collecting logs or version details without mixing them into the reporter's account

Example prompts

  • “A user says the terminal went blank after switching projects. Help me turn that into a bug report.”
  • “Search for duplicates of this bug report before we file it.”
  • “Gather diagnostic evidence for this issue and keep it separate from the reporter's description.”

What it can do on your machine

Read from SKILL.md and the folder at commit 42d304d. 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 1 file in scripts/ (Python), which the agent can run.

    Shell commands in SKILL.md call:

    • gh

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

  • Network

    No URLs in SKILL.md. Its commands use 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

Bug Report Triage loads about 1.8k tokens when it runs. Until then it costs about 34 tokens; SKILL.md has 989 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~34
When it runs · the whole SKILL.md, loaded when a task matches
~1.8k

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); the scripts in this folder are not scanned.

SKILL.md

The full file from OrchestratorInc/agent-orchestrator at commit 42d304d, republished under its Apache-2.0 licence (© OrchestratorInc). 989 words, ~1,752 tokens.

Download SKILL.mdSave it as .claude/skills/bug-triage/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
bug-triage
description
Help clarify human bug reports, search for duplicates, and gather diagnostic evidence separately from a concise issue draft.

Bug triage

Help the reporter describe what they observed and add useful evidence. A short human observation is enough to report a bug; reproduction, logs, and a diagnosis are optional. Use this skill for reports from chat, an existing issue, or a local user.

Clarify the observation

Read the available report, screenshots, and follow-ups first. Preserve the original reporter's words and attribution, including when someone else relays their report. Don't infer a GitHub username from a chat display name.

Ask only questions that would materially clarify the symptom, a few at a time. Don't repeat questions already answered or turn these suggestions into required fields:

  • What did you do, and what happened instead of what you expected?
  • Where did it happen, and does it happen every time or only sometimes?
  • For a visual problem, can you share a screenshot or short recording showing it?
  • If relevant, which AO version and OS were you using, and did it start after a change?

If a screenshot is unavailable, accept a description. If the reporter cannot remember steps or reproduce the problem, preserve that uncertainty and proceed with what is known. Never invent steps or make the reporter diagnose the bug before accepting it.

Keep the report and investigation separate

Draft a specific symptom-based title and a short body, normally one paragraph or a few bullets. Include only details supplied directly by the human: their observation, expected behavior, steps, frequency, environment, and impact when they provided them. Light editing for clarity is fine; preserve meaning and uncertainty. Omit empty sections and unnecessary background. Do not add agent-inferred impact, root cause, confidence scores, or proposed fixes to the report.

Example human report:

After I switched projects, the terminal stayed blank. I expected to see the session prompt. It happened twice today on macOS; I don't know how to reproduce it reliably.

Keep agent-discovered information separate from the human report and label it Agent-collected evidence, even when it confirms the report. Include only relevant excerpts and:

  • The command or read-only query used, collection time, and relevant AO version or checkout commit, so another person can interpret the result.
  • Observed output, screenshots, or reproduction results, with enough context to verify the finding. Distinguish the reporter's environment from a separate test build.
  • A brief explanation of what the evidence supports and its limits. Label hypotheses explicitly; don't present a code-path guess as a confirmed cause.

If no useful evidence is available, deliver the human report alone. Never pad it with speculation or claim a reproduction succeeded when it didn't.

Gather relevant evidence

Use diagnostics that fit the symptom, within the user's requested scope. Prefer read-only checks against the affected installation. Don't restart sessions, upgrade AO, change live data, or alter the user's checkout just to collect evidence.

AO uses a Go daemon and an Electron/React frontend. Read docs/architecture.md for current boundaries and locate actual files with rg before citing code. Do not rely on the old TypeScript implementation or assume which runtime backs a session.

Before trusting diagnostics, identify the AO executable and affected daemon. A PATH lookup may resolve to a legacy install. Check the executable's version and status, and the configured run file/data directory (AO_RUN_FILE / AO_DATA_DIR when set). The primary daemon normally uses loopback port 3001; port alone does not establish which version or installation is affected. Consult the installed CLI help for commands.

Collect only what helps explain this symptom:

SymptomPotential evidence
UI rendering or interactionScreenshot/recording, relevant console errors, affected view
Session or terminal stateAffected session's daemon state, relevant log excerpt, runtime details
Daemon or CLI failureVersion/status, exact command and error, relevant daemon logs
Persisted state mismatchRelevant rows and schema from a read-only SQLite query
Show full SKILL.md (373 more words)Show less

Use the daemon API/CLI for state first. If database evidence is needed, discover the actual database location and schema, open it read-only, and select only relevant rows and columns. Never attach an entire database or broad log dump. Redact tokens, credentials, private prompts, and unrelated personal or repository data from every artifact, including screenshots. Keep diagnostic artifacts under ~/.ao and out of Git commits. Preserve evidentiary content when redacting, and verify that unrelated content did not change. Share a useful excerpt rather than a wall of output.

Try a safe reproduction when practical, recording the build actually tested. Failure to reproduce does not invalidate the human report. Code tracing or dependency research is optional when it can add concrete evidence. Keep findings in the separate evidence.

Check for duplicates

Search open and closed issues and related PRs using the symptom or exact error:

bash
gh issue list --repo Untrivial-ai/agent-orchestrator --state all --search '<symptom or error>'
gh pr list --repo Untrivial-ai/agent-orchestrator --state all --search '<symptom or error>'

Read likely matches before calling a report a duplicate. If one matches, give the reporter its link and draft only their new observation for a comment, with evidence attached separately. If a closed issue appears to have regressed, explain that to the reporter rather than silently treating it as resolved. If search is unavailable, state that briefly; it does not block preparing the report.

Preserve attribution and submission scope

A local coding agent may help draft and gather attachments. Existing authorization still applies; triage alone is not permission to publish.

For an existing issue, preserve its human body and attribution. Add evidence separately when authorized. Don't rewrite the report into an investigation narrative.

When authorized to publish, include useful collected evidence before finishing. Use a clearly labeled comment or attachment, whichever fits the evidence. If evidence cannot be published, state the exact blocker and provide its local path. Never invent attachment URLs or create asset branches to host diagnostics. Keep the report body limited to human details. Apply only existing, relevant labels; don't add priority/confidence prose just to fill a template.

Finish with the concise draft (or issue link if submitted), any duplicate link, and attachment paths/links. Do not automatically file an issue, implement a fix, push a PR, or spawn a worker. Follow the user's requested scope and the repository's normal contribution workflow if they separately ask for implementation.

© OrchestratorInc, 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 (scripts) in .agents/skills/bug-triage of OrchestratorInc/agent-orchestrator.

  • SKILL.md
  • scripts/push_fix_to_github.py

Open the folder on GitHubat commit 42d304d

Compare with similar skills

Bug Report Triage 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.

Bug Report Triage compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Bug Report Triage this skillOrchestratorInc/agent-orchestrator13k—~1.8kAutomated safety check: PassApache-2.0
OpenROAD Issue TriageThe-OpenROAD-Project/OpenROAD3.2k—~842Automated safety check: PassBSD-3-Clause
Triage Issuesoftspark/ai-toolkit179—~1.3kAutomated safety check: NotesApache-2.0
Triage IssuesClickHouse/clickhouse-java1.6k—~904Automated safety check: PassApache-2.0
Issue Triagepnp/powershell905—~1.8kAutomated safety check: PassMIT
GitHub Issue Creatoraiskillstore/marketplace4303 repos~1kAutomated safety check: PassNone

Similar skills

  • OpenROAD Issue Triage

    The-OpenROAD-Project/OpenROAD

    Reproduces an OpenROAD GitHub bug from an attached tarball and shrinks the failing design with whittle.py so maintainers get a minimal test case.

    3.2k GitHub stars~842 tokensUpdated yesterday
    DevelopmentAuto-check passed

  • Triage Issue

    softspark/ai-toolkit

    Bug triage: explores codebase for root cause, files GitHub issue with TDD fix plan.

    179 GitHub stars~1.3k tokensUpdated 2 days ago
    DevelopmentAuto-check: notes

  • Triage Issues

    ClickHouse/clickhouse-java

    Analyzes a single GitHub issue at a time. An agent skill from ClickHouse/clickhouse-java.

    1.6k GitHub stars~904 tokensUpdated yesterday
    Testing & QAAuto-check passed

  • Issue Triage

    pnp/powershell

    Take a PnP PowerShell GitHub issue, find the cmdlet that owns it, trace the code path, and decide whether the cause is in this repo or in PnP Framework, PnP Core SDK or the service.

    905 GitHub stars~1.8k tokensUpdated yesterday
    Testing & QAAuto-check passed

  • GitHub Issue Creator

    aiskillstore/marketplace

    Turn error logs, screenshots, voice notes, and rough bug reports into crisp, developer-ready GitHub issues with repro steps, impact, and evidence.

    430 GitHub starsUsed in 3 repos~1k tokens
    Testing & QAAuto-check passed

  • Issue Triage Workflow

    Hack23/cia

    Triage GitHub issues: labeling, prioritization, assignment, and resolution tracking

    239 GitHub stars~1.1k tokensUpdated yesterday
    Product & Project ManagementAuto-check passed

More from OrchestratorInc/agent-orchestrator

  • AO Desktop App Launcher

    OrchestratorInc/agent-orchestrator

    Launches, restarts and troubleshoots the real AO Electron desktop app from a checkout, with isolated or real local data and checks for stale processes.

    13k GitHub stars~2.4k tokensUpdated yesterday
    Auto-check passed

  • AO CLI Catalog

    OrchestratorInc/agent-orchestrator

    Catalog of the ao command-line tool for Agent Orchestrator: spawning workers, managing sessions and projects, Cues, messages, the shared browser, previews and the daemon.

    13k GitHub stars~1.3k tokensUpdated yesterday
    Auto-check passed

  • AO Cloud CLI Catalog

    OrchestratorInc/agent-orchestrator

    Documents the ao CLI for spawning, listing, messaging, and killing worker sessions, and for reporting back to an orchestrator from inside a sandbox.

    13k GitHub stars~674 tokensUpdated yesterday
    Auto-check passed

  • PR Change-Count Header

    OrchestratorInc/agent-orchestrator

    Calculates and keeps up to date the required code, test and other change-count header at the top of a pull request description in this repository.

    13k GitHub stars~364 tokensUpdated yesterday
    Auto-check passed

Works with

Categories

Questions about Bug Report Triage

What does Bug Report Triage do?

Helps a reporter describe a bug, searches for duplicates and gathers diagnostic evidence kept separate from a short, human-worded issue draft. A short human observation is enough to report a bug, and reproduction steps, logs and diagnosis are optional. The agent reads the report, screenshots and follow-ups first, keeps the reporter's own words and attribution, and does not guess a GitHub username from a chat display name.

When should I use Bug Report Triage?

Bug Report Triage fits situations like: turning a vague chat message into a clear bug report; checking whether a reported bug already exists as an issue; collecting logs or version details without mixing them into the reporter's account.

How do I install Bug Report Triage in Claude Code?

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

How do I install Bug Report Triage in Codex?

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

Can I use Bug Report Triage 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 OrchestratorInc/agent-orchestrator --skill bug-triage -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/bug-triage, .gemini/skills/bug-triage, .github/skills/bug-triage and .opencode/skills/bug-triage in your project.

What does Bug Report Triage need to run?

Going by SKILL.md and its folder, Bug Report Triage needs Python for the scripts in its folder and the command-line tools its instructions call (gh).

Does Bug Report Triage access the network?

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

Is Bug Report Triage 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. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Bug Report Triage use?

Bug Report Triage 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 Bug Report Triage use?

About 1.8k tokens (SKILL.md is roughly 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 Bug Report Triage?

Skills that share tags, products or a category with Bug Report Triage: OpenROAD Issue Triage (The-OpenROAD-Project/OpenROAD, 3.2k stars), Triage Issue (softspark/ai-toolkit, 179 stars), Triage Issues (ClickHouse/clickhouse-java, 1.6k stars) and Issue Triage (pnp/powershell, 905 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Bug Report Triage?

OrchestratorInc (a GitHub organization) maintains it in OrchestratorInc/agent-orchestrator, which has 12,881 GitHub stars. The repository holds 5 skills in this directory. The repository was last updated on October 8, 2026.

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