AGENT SKILL

review-threads

Use when reading, answering, resolving, dismissing, or submitting review threads on a pull or merge request. Triggers: what is unresolved on this PR, reply to that review comment, resolve the thread, resolve all threads, dismiss that review, submit my review, check PR status, what is blocking the merge, odpovez na komentare v PR, co je nevyresene. Drives qsm-forge gh for GitHub and qsm-forge gl for GitLab instead of raw platform calls. The review itself is the review skills; this carries a finished decision to the forge.

License 许可
GPL-3.0
最近更新
2026-11-13
Stars
★ 0
Forks
0

01 前言 Front Matter

name
review-threads
description
Use when reading, answering, resolving, dismissing, or submitting review threads on a pull or merge request. Triggers: what is unresolved on this PR, reply to that review comment, resolve the thread, resolve all threads, dismiss that review, submit my review, check PR status, what is blocking the merge, odpovez na komentare v PR, co je nevyresene. Drives qsm-forge gh for GitHub and qsm-forge gl for GitLab instead of raw platform calls. The review itself is the review skills; this carries a finished decision to the forge.

03 SKILL.md 正文

Review threads

A review thread is a conversation someone is waiting on, and every call here posts under the user's own account. So read freely and write only what the user asked for: a wrong reply costs the user standing with their reviewer, a drafted one costs a turn. This skill carries a decision already made to the forge; it is not the review methodology.

Each verb is one call where the raw way is several: listing threads on GitHub is a GraphQL query with paging, then a lookup of the thread id behind a comment id, then the mutation. The verbs do that inside, and a thread row has the same fields on both forges (id, file, line, first comment, reply flag), so the plan for a round of review is fewer calls and one shape. When you plan the calls, count them: a plan that lists, looks up and mutates by hand has missed the verb.

qsm-forge gh serves GitHub pull requests, qsm-forge gl GitLab merge requests; the checkout's origin remote picks one. Both resolve the repository from the working tree; pass --repo owner/name or --project namespace/project only for a target elsewhere.

Read

  • status first: state, review decision, merge readiness, unresolved count, checks. Red checks mean the threads can wait.
  • threads <number>: every unresolved thread with comment id, file, line, text, and whether someone replied. The comment id is the handle for every write, so read this before replying.
  • qsm-forge gh comments <pr> --unresolved for the whole conversation. GitLab has no such verb; its threads carry every note.
  • --json when a decision consumes the answer, columns when a person reads it.

Write

Two roles, because a finding is closed by a reader or by a test, never by the claim alone:

  • As the implementer, push the fix first, to the thread's own branch or to a stacked pull request above it, because a reply that names a commit the forge does not have points the reviewer at nothing. Then reply <number> <comment-id> --message "<text>" to every comment, on both forges. Say where the fix landed (Fixed in abc1234, in pull request #<number>) or why not; "done" proves nothing.
  • Resolve your own thread only when a test that fails without the fix proves it, and name the test in the message (resolve <number> <comment-id> --message "Fixed in abc1234. Test: ..."). Without such a test, reply and leave the thread open for the reviewer.
  • As the reviewer, check each fix. When it is right, resolve <number> <comment-id> --message "Good". When it is not, reply with what is still wrong and leave the thread open.
  • qsm-forge gh resolve-all <pr> is the reviewer's sweep after checking every fix: it closes only threads whose latest reply starts with "Fixed in". With --message "<text>" it posts that text on every open thread and closes them all; use it only when the user asked for exactly that.
  • resolve without a message only when nothing needs saying, because a silent resolve on a question reads as avoidance.
  • qsm-forge gh dismiss-reviews <pr> clears reviews still asking for changes, --author <login> for one. It is for a stale bot review after the changes landed; ask before dismissing a person.

Submit a whole review

submit-review <number> --file <path>, or the same payload on standard input. One payload, one call, because on GitLab a review is several requests and a malformed payload posts nothing, which beats a half-posted review.

{
  "decision": "request-changes",
  "body": "Two things before this merges.",
  "inline": [{ "path": "src/a.ts", "line": 42, "body": "This leaks the handle on the error path." }]
}

decision is comment, approve or request-changes; body is required unless approving. request-changes withdraws an approval on a merge request. Approve only when the user said so.

Raw calls and exit codes

A hook refuses raw gh api, glab api and curl calls that touch review state and names the verb that covers it; read the refusal, it is the answer. Cloning, creating, listing and releases stay open. When no verb fits, say so and ask instead of routing around the hook.

0 did what it said. 2 written wrong, nothing posted. 1 the forge or its CLI refused, and the message is theirs: an authentication failure does not fix itself on retry.

Red flags

  • When you are about to reply with a comment id you did not read from threads, read it, because a guessed id answers the wrong thread.
  • When you reach for resolve-all --message to save answering each thread, answer them, because the reviewer reads the same text everywhere.
  • When a reply says "fixed" with no commit, add the hash, because nobody can verify the claim.
  • When the fix is committed but not pushed, push before the first reply, because the hash in it does not exist on the forge yet.
  • When you are about to resolve your own thread with no test that fails without the fix, reply instead and leave it open, because only a test or the reviewer closes a finding.
  • When the user asked for comments and the payload says approve, change it, because approval is the user's call.

Say it in the reply

Name the forge and the identifier, what you posted, and what is still open: "Replied on owner/name#12 src/a.ts:42 with the fix hash; the thread stays open for the reviewer." When something was not posted, say so first and why.

04 同仓库更多 Skills

qsmark-labs
qsmark-labs/qsm-marketplace

worktrunk

Use when working with git worktrees or wt (worktrunk): creating, switching, merging or removing a branch workspace, configuring wt (project or personal config, workflow hooks around switch, commit and merge, commit-message generation, hook approvals), or a wt command that failed. Triggers: make a worktree for this branch, wt merge, wt remove, configure wt, .config/wt.toml, run tests before merge, install deps when a worktree is created, wt hook, wt switch does not cd, nastav wt, wt nefunguje.

★ 0
qsmark-labs
qsmark-labs/qsm-marketplace

web-ladder

Use when a task needs something the internet holds: look something up, search the web, find which page answers a question, read a page, check a library's current docs, a page that needs a browser or shows a captcha. Runs one research loop: searxng search, keywords, cases, search again up to three times, Context7 for a library, Tavily when the keywords will not come, one page in a browser, the person on a wall. Triggers: look this up, search the web, find out online, read this page, check the docs, the page is blocked, captcha, najdi na webu, otevri tu stranku. A multi-round question goes to the qsm-web:web-research agent; one tool's commands are its own skill (searxng, browserbox, opensrc).

★ 0
qsmark-labs
qsmark-labs/qsm-marketplace

searxng

Use when running a web search through the person's own SearXNG instance with qsm-web searxng: choosing a search profile or engines, the instance, time range or language, reading its JSON, a suspended engine. Triggers: search with searxng, qsm-web searxng, search profile, niche search, academic search, IT search, vyhledavaci profil, searxng is down, which engines. The research loop around it is the web-ladder skill.

★ 0
qsmark-labs
qsmark-labs/qsm-marketplace

opensrc

Use when the source code of an open-source library or repository is needed on disk to read: the docs are silent, the behaviour must be checked in code, the version the project pins. Runs qsm-web opensrc get with a package spec or a repository address. Triggers: get the source of this package, clone this library, read the library's code, qsm-web opensrc, stahni zdrojak knihovny. Library docs are the Context7 skill; the research loop is the web-ladder skill.

★ 0
qsmark-labs
qsmark-labs/qsm-marketplace

browserbox

Use when one page needs a real browser through qsm-web browserbox: content built by script, a page that blocks plain fetching, a captcha to solve or hand to the person, a scripted sequence of clicks, a login the person is present for. Triggers: open it in a browser, the page is empty without JavaScript, captcha, take over the browser, run these steps on the page, log in to this site, browserbox. When to reach for a browser at all is the web-ladder skill.

★ 0
qsmark-labs
qsmark-labs/qsm-marketplace

usage

Use when the user wants a review of how they use Claude Code, or setup hands over after a first install: review my Claude Code usage, audit my setup, what does my setup cost, where did my tokens go, clean up my Claude Code, usage report, zkontroluj moje pouzivani. One walkthrough of four stages (machine, work, habits, extensions), questions after each, then one plan. A question about one topic (which hooks are slow) goes to that topic's skill; this session's tokens alone is `/qsm-usage`.

★ 0