walkthrough
Skill2 starsChanged 39 days agoMITView on GitHub
Show what a finished change did and prove it works with its evidence: a short markdown note after any change, or a customer-facing HTML walkthrough of a release.
30 other projects use a version of this.
Adopted by
Most recently taken up by
- taecontrol/skillstheir own version, first seen Nov 8, 2026
- marnickvda/pop-quiztheir own version, first seen Nov 8, 2026
- command-center-ai/skillstheir own version, first seen Nov 8, 2026
- marnickvda/know-it-alltheir own version, first seen Nov 8, 2026
- spencerldixon/skillstheir own version, first seen Nov 8, 2026
- udecode/dotaitheir own version, first seen Nov 7, 2026
Used with
Projects with it also use chips, redo and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
udecodeSkillchips
Run the project's suggested-task ledger, the chips. /chips alone brings this session's open tasks up to date: it applies what the ledger helper says, settles stale and split tasks, cards what can start and reports the board, and /chips all also takes the ship-list tasks and the ones no live session owns, and merges duplicate tasks across the board. Use for /chips, /chips offer, start or drop, \"what chips are open\", \"suggest a task for X\", \"update the chips\", \"clean up the chips\", \"which chips are stale\", \"is this chip still worth doing\", \"merge duplicate chips\", at the last hand-back of a run that offered a task, and whenever pstack would file a tracked issue, a follow-up, a reflect Backlog item or an out-of-band PR.
Also in 3 projects that use this
Scylla23Skillredo
Re-run the last routed task one model tier up (haiku → sonnet → opus → fable). User-invoked escalation when the previous result missed. Pass --dry-run to preview without running.
Also in 3 projects that use this
Aetheria-Labs1Skillplan-page
Plan one storefront page from strategy through final copy, section order, proof, offers, template direction, responsive layout, and production asset requirements. Stops at an approved execution-ready specification.
Also in 3 projects that use this
SocialKit-APISkillvideo-transcripts
Extract transcripts from social videos through SocialKit across YouTube, TikTok, Instagram, Facebook, X/Twitter, LinkedIn, and direct video file URLs. Use when the user asks to transcribe, get captions or subtitles, pull the spoken text, or read what a video says, including one video or many at once. Returns full text plus timestamped segments. Not for posting, private accounts, or unrelated website scraping.
Also in 3 projects that use this
MicrosoftSkillcleanup
Also in 3 projects that use this
BlockSkilltest-audit
Invoke whenever writing, changing, reviewing, or sweeping tests in buzz-app. Authoring gate for new tests plus audit workflow for low-value, implementation-coupled, or duplicative tests and the test-only production seams they demand.
Also in 3 projects that use this
Other versions30 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 30
- udecode/plate★ 17ksame name
- udecode/dotai★ 1.2ksame name
- udecode/better-convex★ 452same name
- Mathews-Tom/armory★ 340same name
- smallnest/goal-workflow★ 283same name
- andrew-yangy/gru-ai★ 153same name
- oborchers/proqi★ 45same name
- huytd/skills★ 19same name
- wycats/ai-plugin★ 19same name
- wycats/vscode-ai-plugin★ 19same name
20 more not listed
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 repository16 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- codebase-researchSkill
- context-transferSkill
- feature-pr-descriptionSkill
- how-it-worksSkill
- human-voice-writingSkill
- i18n-translateSkill
- impact-checkSkill
- memory-reviewSkill
- orchestrateSkill
- promotion-pr-descriptionSkill
- pr-review-triageSkill
- qa-testerSkill
- release-notesSkill
- thorough-diff-reviewSkill
- writeups-cleanupSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 30 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
When it is unclear, take you, the smaller deliverable, and say so in one line.
- Very specific: line 42
If a qa-tester run for this change exists, reuse its folder, writeups/changes/<date>-<run>/qa/: its results.md and the evidence files it …
- Distinctive: line 47
Never invent a screenshot, a transcript or a step nobody ran.
- Distinctive: line 53
After a change has been QA'd, at the end of an orchestrated run, or when the user asks for one.
- Very specific: line 135
Read references/customer-release.md before building anything: what to show, capturing, the evidence each sentence needs, the page, the …
- Very specific: line 139
If local.md exists next to this file, read it.
“Common” is measured: kept by at least 6 in 10 of the 30 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Walkthrough
One skill, two audiences. Pick by what was asked:
- You: "walk me through the change", "show me what changed", or the end of an orchestrated run. A short markdown note for the developer who made or reviews the change, for any kind of change (section 2).
- Customers: "release walkthrough", "what's new for customers", or a release version. A self-contained HTML page showing what a release changed for the people who use the product, for releases they see in the UI (section 3).
When it is unclear, take you, the smaller deliverable, and say so in one line.
Neither audience gets a PR description or a test report: a walkthrough does not argue for merging and does not list every check that ran.
Repo facts
This skill needs the facts below. Look each one up in the repo's CLAUDE.md (or AGENTS.md) and
the files it imports or points to (of a repo-facts file, only this skill's section:
sed -n '/^## walkthrough$/,/^## /p' <file>), then in local.md next to this file. If neither has it, ask
the user once, then suggest the one line to add to CLAUDE.md so the next run finds it.
| Fact | Example |
| Where drafts and walkthroughs go | A gitignored writeups/ folder in the main checkout, resolved from the main checkout |
| How to reach the product, and how to sign in | The qa-tester Repo facts: its surfaces, test account and sign-in steps |
| Where release notes live, and who reads them | releases/release-<version>.md, written for the people who use the product (see release-notes) |
| The brand look to copy | The site's own fonts as a system stack, its accent colours, both light and dark |
| Words banned from customer-facing text | The "Jargon banned from reader-facing text" row of release-notes, plus the internal terms in section 3 |
1. Evidence first
Both audiences need evidence from the running product, never invented.
- If a
qa-testerrun for this change exists, reuse its folder,writeups/changes/<date>-<run>/qa/: itsresults.mdand the evidence files it lists. - Otherwise invoke
qa-testeron the change first; it saves the evidence this skill reads. Never start, stop or restart a server: the product is reached the wayqa-tester's Repo facts say, and reported as down if it is not there. - Never invent a screenshot, a transcript or a step nobody ran.
2. For you
When
After a change has been QA'd, at the end of an orchestrated run, or when the user asks for one.
The input is the diff plus the qa-tester folder: the cases that ran, their verdicts and their
evidence. If no evidence exists, get it first (section 1).
Format
Pick the file by surface. A change that touched a UI surface (web, native or mobile app, game)
gets walkthrough.html, one self-contained page with the screenshots inlined, and no .md. Every
other change (API, CLI, library, docs) gets walkthrough.md.
Both carry the same four sections and about 40 lines of content, in plain words (see
human-voice-writing):
## What changed: 3 to 5 lines, described by behaviour, not by file.## Try it: numbered steps, written as the surface's user performs them: a click, a request, a command, a call. Each step has one "look for" line saying what should be visible, and its evidence.## Files worth reading: at most 5 paths, one line each on why that file matters.## Not covered: what was not exercised, so the reader knows where to look themselves.
To build the HTML page, copy references/for-you.src.html to the session scratchpad as
page.src.html and fill it. Each shot gets a full-width row in a .shots block; only light and
dark shots of one screen share a row, in a .shots pair block. Every shot opens full size on click.
Reference each screenshot as {{SHOT:<path relative to the run folder, without .png>|<alt text of
20+ characters saying what to notice>}}. Take the shots sharp and cropped, as qa-tester's web
reference says. Then, from the top of the product repo (Playwright comes from
there), with the sandbox off as section 3 says:
node "${CLAUDE_SKILL_DIR}/scripts/inline.mjs" "$SCRATCH/page.src.html" "<run folder>" "<run folder>/walkthrough.html"
node "${CLAUDE_SKILL_DIR}/scripts/verify.mjs" "<run folder>/walkthrough.html" "$SCRATCH/verify.png"Fix every broken image, missing alt or sideways scroll the check reports, and run both again.
Evidence per surface
Each step shows the evidence of the surface it exercises, from qa/:
| Surface | What a step shows |
| Web, native or mobile app, game | the screenshot, inlined in the HTML page as {{SHOT:qa/02-settings-saved|alt text}} |
| HTTP API | the request line and the response's decisive lines in an http block, then a link to the full qa/NN-<case>.http |
| CLI and scripts | the command and its decisive output lines in a console block, then a link to the full transcript |
| Library | the few lines of the call and their output, before and after when the behaviour changed, from qa/NN-<case>-before.txt and -after.txt |
Inline only the lines a reader needs to see the point; the full file is one link away. Link into
qa/ relatively: qa-tester keeps that folder. A screenshot you take yourself for the note goes in
shots/NN-<step>.png, next to it.
Where it goes
The note goes in the run's folder, next to qa/, resolved from the main checkout, never from a
worktree, so a worktree cleanup cannot delete it. For a writeups/ folder in the main checkout,
this resolves to the right folder from the main checkout and from any worktree alike:
WRITEUPS="$(dirname "$(git rev-parse --path-format=absolute --git-common-dir)")/writeups"
NOTE="$WRITEUPS/changes/<date>-<run>/walkthrough.md"
PAGE="$WRITEUPS/changes/<date>-<run>/walkthrough.html"The note is never committed and never pasted into a PR, unless the user asks. The HTML page is a
local file, never an artifact, never committed. Report its path when done, with the exact command
for the user to open it, open '<path>', because the sandbox blocks open.
3. For customers
The goal is one HTML file the user can send on, showing only what a customer will notice, with every
claim verified and no internal or personal data in it. It goes in the same drafts folder as above,
as releases/<version>/walkthrough.html with its screenshots in releases/<version>/shots/, and is
never committed or published unless the user asks.
It is for releases a customer sees in the product's UI: web, native or mobile. A release with no visible change (an API, a CLI, a library) gets no page: say so, and point to its release notes.
Confirm the release and its scope with the user before building anything. A release note is a starting point, not the scope. If a state needs staging you cannot do (a second account, a plan change, a backend lever), name it and ask once, with the exact commands. When staging is refused, say what the page will not cover; do not quietly reshape it around the states you could reach.
Read references/customer-release.md before building anything: what to show, capturing, the
evidence each sentence needs, the page, the scripts that capture, inline and verify it, and the
hand-off.
If local.md exists next to this file, read it; where it disagrees with this file, it wins.
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
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.

