agentleFS
kkrafft1999kkrafft1999/snotra/.claude/skills/user-feedback/SKILL.md

user-feedback

Skill6 starsChanged 10 days agoApache-2.0View on GitHub

Looks for user feedback on GitHub in both of the shapes it arrives in — issues filed by the in-app feedback form (label `feedback`) and contributions by people other than the owner (issues, comments, pull requests, discussions) — and reports only what is new or still in progress: closed items and advertising are left out. For each remaining item it says whether it is answered or waiting, on the board or not, real or a test. Read-only. Triggers on sentences like "gibt es Feedback?", "gibt es Feedback in GitHub Issues?", "gibt es welche mit Tag feedback?", "hat jemand was geschrieben?", "Feedback-Status", "neues Nutzer-Feedback?", "is there any feedback?", "any new user feedback?", "feedback status", "did anyone comment?". Only makes sense in this repository (snotra).

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. jtenniswood/espcontrol★ 1.2ksame name
    2. savannahfeder/agentbox★ 419same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository3 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run these together; OWNER=kkrafft1999, REPO=kkrafft1999/snotra.

    line 63
  2. Very specific:

    If the user asks for a time window ("seit letzter Woche", "since the release"), add since=<ISO date> to the comments call and filter the …

    line 105
  3. Very specific:

    Answered or waiting: Look at the last comment (gh issue view <n> --json comments, or the discussion's comment list): if it is from someone …

    line 113
  4. Very specific:

    Never drop the --query (see .claude/rules/task-management.md).

    line 121
  5. Distinctive:

    Every item is a link with its title or a one-line gist — never a bare number.

    line 138

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Find user feedback and report its state

Feedback reaches this repository in two shapes, and a search for only one of them misses the other:

  1. Through the in-app form (Help › Send Feedback…, #1045). The endpoint on the website files a public issue under the owner's account (kkrafft1999) with the label feedback. The body ends in a table (Kind — Bug / Idea / Praise, App version, Operating system, Architecture, Provider, Language) and the line Sent through the in-app feedback form · feedback id …. The author therefore says nothing about who wrote it.
  2. Directly on GitHub, by someone else. Anyone with a GitHub account: an issue (also the form's fallback when the endpoint is down — it opens GitHub's own issue form prefilled, see buildFeedbackIssueUrl in src/shared/contracts/feedback.js; those issues carry no feedback label), a comment on an existing issue, a pull request, a discussion post.

Everything here is read-only. Nothing is commented, labelled, closed or moved on the board unless the user asks for it afterwards — and a reply on GitHub is public, so it is shown to the user word for word before it goes out.

What is shown: only what is new or still in progress

The user wants to see what still needs attention, not the history (decided 2026-11-07, #1374). Two kinds of item are therefore left out of the report entirely — no row, no footnote:

  • Closed items. A closed form issue, a closed issue or pull request by someone else, a closed discussion, and every comment whose parent issue or pull request is closed. The commands below query open items only, so these never show up.
  • Advertising. A comment or issue that mainly promotes a product or service (a link to the author's own offering, a sign-up, a price list, "message us on Telegram") is dropped, even when it sits on an open issue and even when it carries some useful technical text. Only clear cases: a contributor offering to work on an issue is not advertising.

Closed items come back only when the user asks for them in words ("auch die geschlossenen", "alles", "Historie", "including closed"). Then switch every open below to all (--state all, drop states:OPEN, skip the $OPEN filter) and mark closed items as such.

Step 1 — collect both shapes (without asking)

Run these together; OWNER=kkrafft1999, REPO=kkrafft1999/snotra. Bots (dependabot, github-actions, anything ending in [bot], app/…) are not feedback and are filtered out.

REPO=kkrafft1999/snotra
BOTS='dependabot|github-actions|bot]$|^app/'   # no backslashes: jq rejects "\["
# Open issues and pull requests (the issues endpoint returns both) — comments
# on anything else are on closed items and are skipped.
OPEN=$(gh api --paginate "repos/$REPO/issues?state=open&per_page=100" --jq '.[].number' | jq -s -c .)

echo "== 1. In-app form (label feedback) =="
gh issue list --repo $REPO --state open --label feedback --limit 500 \
  --json number,title,state,createdAt,body,comments \
  --jq '.[] | "#\(.number) \(.createdAt[:10]) \(.title) | kind=\((.body|capture("\\| Kind \\| (?<k>[^|]+)")?.k // "?")|gsub(" ";"")) version=\((.body|capture("\\| App version \\| (?<v>[^|]+)")?.v // "?")|gsub(" ";"")) | comments=\(.comments|length)"'

echo "== 2a. Issues by others =="
gh issue list --repo $REPO --state open --limit 1000 \
  --json number,title,author,createdAt \
  --jq ".[] | select(.author.login != \"kkrafft1999\" and (.author.login|test(\"$BOTS\")|not)) | \"#\(.number) \(.createdAt[:10]) @\(.author.login): \(.title)\""

echo "== 2b. Comments by others (on open issues and PRs) =="
gh api --paginate "repos/$REPO/issues/comments?per_page=100" \
  | jq -r --argjson open "$OPEN" --arg bots "$BOTS" '.[]
      | select(.user.login != "kkrafft1999" and (.user.login|test($bots)|not))
      | (.issue_url|split("/")|last|tonumber) as $n | select($open|index($n))
      | "\(.created_at[:10]) @\(.user.login) #\($n): \(.body[:200]|gsub("\n";" "))"'

echo "== 2c. Pull requests by others =="
gh pr list --repo $REPO --state open --limit 500 --json number,title,author,createdAt \
  --jq ".[] | select(.author.login != \"kkrafft1999\" and (.author.login|test(\"$BOTS\")|not)) | \"#\(.number) \(.createdAt[:10]) @\(.author.login): \(.title)\""

echo "== 2d. Discussions (started by others, or with a comment by others) =="
gh api graphql -f query='{repository(owner:"kkrafft1999",name:"snotra"){discussions(first:50,states:OPEN,orderBy:{field:UPDATED_AT,direction:DESC}){nodes{number title author{login} createdAt comments(first:50){nodes{author{login} createdAt body}}}}}}' \
  --jq '.data.repository.discussions.nodes[] | select(.author.login != "kkrafft1999" or ([.comments.nodes[].author.login] | any(. != "kkrafft1999"))) | "D#\(.number) @\(.author.login) \(.createdAt[:10]) \(.title) | " + ([.comments.nodes[] | "@\(.author.login) \(.createdAt[:10])"] | join(", "))'

Comment hits shown as a 200-character excerpt can hide the advertising at the end — before keeping a comment from an unknown author, read its full body (gh issue view <n> --json comments) and drop it if it turns out to be advertising.

If the user asks for a time window ("seit letzter Woche", "since the release"), add since=<ISO date> to the comments call and filter the rest by createdAt. Without one, report every open item, newest first.

Step 2 — the state of each item

For every hit that is left after the filter, establish:

  • Answered or waiting. Look at the last comment (gh issue view <n> --json comments, or the discussion's comment list): if it is from someone other than kkrafft1999 and nothing from the owner follows, the item is waiting for an answer. A form issue without any comment is not answered — whether it needs one depends on its kind (praise rarely does, a bug report does).
  • On the board: column from gh project item-list 2 --owner kkrafft1999 --query "is:issue is:open" --limit 200 --format json. Never drop the --query (see .claude/rules/task-management.md). An open feedback issue with no board entry is worth saying so.
  • Real or a test. Titles like Test of the feedback endpoint … / Test of the duplicate check … are the owner's own tests. A form issue whose diagnostics match the owner's machine (current macOS on arm64, provider openai, language de, filed on a release day) is probably a self-test too — say "probably", never more: the author field cannot tell. An open test issue stays in the report: it is a leftover to close.

Step 3 — report (in the conversation language)

Two sections, one per shape, each as a short table, newest first:

ItemDateWhoKind / topicState
#1204 Linux sandbox does not start on Ubuntu 26.042026-10-27@sunny-kavixoinvestigation resultsanswered 2026-10-28, ball with the contributor, Backlog
  • Every item is a link with its title or a one-line gist — never a bare number.
  • State combines answered/waiting, board column, test/real. Closed items and advertising are not in the table (see above).
  • A section with no open item is one line ("nothing open"), not an empty table.
  • Close with one sentence of verdict: is there real outside feedback that is still open, and is anything waiting for an answer? If both sections are empty, say so plainly instead of padding the answer.
  • Offer follow-ups only where they exist: drafting a reply to a waiting item, filing an open form issue on the board, closing a leftover test issue. Each of them is a write to GitHub and happens only after the user says yes.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.