Prompt of the Day:
🔗 **Source:** [Prompt of the Day:](https://twitter.com/mattpocockuk/status/2106730768789602313/?rw_tt_thread=True)
> [!note]+ Document Notes
> tags: [ai-coding-agents, agent-skills, prompts, claude-code, release-notes]
> author: Matt Pocock
> source: X (Twitter)
> reference: https://twitter.com/mattpocockuk/status/2106730768789602313/
> date created: 2026-10-05 09:46
>
> > [!info] Generated analysis — not the author's words. Sections below are machine-written
> > from the captured source. Only text inside `> [!quote]` blocks is verbatim from the author.
> > Do not quote or attribute any other line to the named author.
>
> # 🏷️ Title: Prompt of the day: Matt Pocock on the mattpocock/skills v1.3 release
>
<!-- deep-read:sections start -->
## 🧭 Table of contents
- [📝 Summary](#-summary)
- [📖 Main content](#-main-content)
- [🗒️ Notes](#️-notes)
- [📚 References](#-references)
- [🗣️ Paraphrase](#️-paraphrase)
- [📰 Newsletter](#-newsletter)
- [✅ Action items](#-action-items)
- [🧪 Hypotheses](#-hypotheses)
- [📊 Methodology](#-methodology)
- [❓ Questions](#-questions)
- [🔮 Future research](#-future-research)
- [💡 Implications](#-implications)
- [🧠 Learning styles assessment](#-learning-styles-assessment)
- [📊 Accessibility assessment](#-accessibility-assessment)
- [🌍 DEIB assessment](#-deib-assessment)
- [✅ Logical validity](#-logical-validity)
- [🛡️ DARVO analysis](#️-darvo-analysis)
- [🎯 Conclusion](#-conclusion)
- [🔄 Refactor appendix](#-refactor-appendix)
Scope note. The captured text is a 68-word tweet by Matt Pocock (@mattpocockuk), published 2026-10-04, six short paragraphs long. It's a prompt meant to be pasted into an AI coding agent. This analysis describes only what the post says. It doesn't describe the contents of the mattpocock/skills repository, what v1.3 contains, or why anything was renamed. Every statement about the release below is the author's claim, as worded in the post. The "Previous:" paragraph points to an earlier tweet that was not captured. Paragraph labels like ¶2 are the pipeline's labels for the captured text.
## 📝 Summary
- **What happened.** Pocock posted a "Prompt of the day:" (¶1) addressed to a coding agent. The prompt tells the agent to check mattpocock/skills for the v1.3 release, rename CONTEXT.md to GLOSSARY.md, check one command for needed changes, and diff the user's skills against the author's (¶2). It then asks the agent to scan the user's last 25 sessions for skill invocations and recommend how the release might improve the user's workflow (¶3). The post closes with "Enjoy v1.3, pals." (¶4).
- **The claim.** The post asserts that a v1.3 release of mattpocock/skills exists (¶2, ¶4). That's the author's statement. The capture doesn't show the release page, a changelog or a version tag.
- **Why it matters.** The post turns a release announcement into a task the reader's own agent performs. The prompt's second step names a rename, and its fourth step reads the reader's own session history, so a reader pasting it should know what each step touches.
- **Key stakeholders.** Named or implied: the author ("Matt's" skills, ¶2); the reader running the prompt ("my repository", "my last 25 sessions", ¶2 and ¶3); and the reader's coding agent, which the text doesn't name.
- **Risks and tradeoffs.** The post gives no detail on what the rename affects, what the setup command changes, or where session history lives. Those are questions for the reader before running it, listed under Questions.
- **Bottom line.** A short, practical prompt with four tasks and a recommendation step. Everything about v1.3 itself, including the rename's purpose, sits outside the capture.
## 📖 Main content
**The prompt's shape (¶1 to ¶3).** The label "Prompt of the day:" (¶1) frames the post as a reusable prompt. ¶2 is a set of imperative commands. ¶3 shifts from changing things to analyzing: first "get a sense" of how the reader uses the author's skills, then recommend how the release "might improve my workflow."
**What the prompt asks an agent to do, step by step.**
1. Re-check mattpocock/skills for the v1.3 release (¶2). The word "Re-check" implies the agent or reader looked before. The capture doesn't say where it checked.
2. Rename CONTEXT.md to GLOSSARY.md (¶2). The text doesn't say in which repository or directory, or why.
3. Check /setup-matt-pocock-skills for any required changes (¶2). The text doesn't describe that command.
4. Perform a diff of the skills in "my repository" versus "Matt's" (¶2). Which repository counts as "my repository" is up to the reader.
5. Scan the reader's last 25 sessions for skill invocations, to get a sense of how the reader uses the author's skills (¶3).
6. Recommend how the release might improve the reader's workflow (¶3).
**The sign-off (¶4).** "Enjoy v1.3, pals." is the author's announcement-in-passing. It's the only sentence that names the version outside the instruction in ¶2.
**The pointer (¶5 and ¶6).** "Previous:" followed by a truncated link to another tweet by the same account. The earlier tweet's content isn't captured, so any context it gives about v1.3 is unknown here.
**Terms.**
- Skill: used by the post without definition. In the post's context it's something an agent can invoke in a session (¶3). The capture doesn't define it further.
- Session: used in "my last 25 sessions" (¶3). The text doesn't say which tool's sessions are meant.
- CONTEXT.md and GLOSSARY.md: two file names. The capture records each as a link, which is an automatic linking of a file name that looks like a domain, not a source (see References).
- /setup-matt-pocock-skills: a slash-style command named in ¶2. Its behavior isn't described.
**Uncertain or contested.**
- What changed in v1.3 is unknown from the capture. The post only names the version.
- Whether the rename is required, optional, or already done in the author's repository isn't stated. ¶2 phrases it as an instruction to the agent.
- What "required changes" from the setup command would be is unknown.
## 🗒️ Notes
#### 📄 ¶1 Label
- "Prompt of the day:" labels the post as a prompt to copy, the first of what may be a series. The capture shows no other posts in it.
#### 📄 ¶2 Release and repository tasks
- Four imperative sentences: re-check, rename, check, diff.
- Uses "my repository" and "Matt's", so the prompt is written from the reader's point of view, with the author referred to by first name.
- Names one file pair (CONTEXT.md, GLOSSARY.md) and one command (/setup-matt-pocock-skills).
#### 📄 ¶3 Usage scan and recommendation
- Asks for "a sense" of how the reader uses the author's skills, which is a qualitative aim.
- Sets a fixed window: the last 25 sessions.
- Ends with a recommendation request hedged by "might."
#### 📄 ¶4 Sign-off
- A single friendly line addressed to "pals." It's the only place the text speaks to readers instead of to the agent.
#### 📄 ¶5 and ¶6 Previous post
- "Previous:" plus a truncated link. The link list resolves it to a tweet on x.com by the same account (see References). Its content isn't in the capture.
## 📚 References
The captured link list has 3 links. Two are artifacts of automatic linking and one is a real pointer.
- [The tweet analyzed here](https://twitter.com/mattpocockuk/status/2106730768789602313/): the post itself, published 2026-10-04.
- [A previous tweet by the same account](https://x.com/mattpocockuk/status/2105951409879849107): the link behind the truncated "x.com/mattpocockuk/s…" in ¶6, introduced by "Previous:". Not captured, so its content is unknown here.
- CONTEXT.md and GLOSSARY.md: the link list maps each file name to an "http://" address of the same name. These look like automatic linking of text that resembles a domain name. They aren't sources, and this note doesn't treat them as such.
- The mattpocock/skills repository: named in ¶2 but not linked in the capture. This note doesn't describe it.
- Claims with no link in the captured markup: the existence and contents of v1.3, and the rename. That wording means only that the capture has no link, not that no source exists.
- Primary source to check next: the mattpocock/skills release page, for what v1.3 says about itself.
## 🗣️ Paraphrase
Pocock offers a prompt for a coding agent. It tells the agent to look in the author's skills repository for version 1.3, rename CONTEXT.md to GLOSSARY.md, look at one setup command for changes it needs, and compare the user's skills with the author's.
It then asks the agent to look back over the user's last 25 sessions to see which of the author's skills they invoked, and to suggest how the new release might help the user's workflow. Pocock ends by telling readers to enjoy the release.
## 📰 Newsletter
**A paste-ready prompt for the v1.3 skills release.** Matt Pocock posted a "Prompt of the day" that sends your coding agent through four tasks: check mattpocock/skills for v1.3, rename CONTEXT.md to GLOSSARY.md, diff your skills against the author's, and scan your last 25 sessions for skill invocations before recommending what the release could change in your workflow.
Before you paste it, decide what you're comfortable letting an agent read. The session scan reads your own history, and the post doesn't say where that history lives or what the agent sees in it. The release details are Pocock's claims, so read the release page too. Link: [the tweet](https://twitter.com/mattpocockuk/status/2106730768789602313/).
## ✅ Action items
Each line ties to the prompt's steps or says plainly that no action follows.
#### DevRel
- Developer advocacy: if you cover the release, link the author's post and the release page, and say which statements are the author's claims.
- Education: a short walkthrough of the six prompt steps, with a note on which steps change files (the rename) and which only read (the diff and the session scan).
- Developer experience: none directly. Nearest relevance: a skills-repo README can say what a major release changes for existing users (inference, not in the source).
- Developer marketing: none. The post is a user-facing prompt, not marketing copy to reuse.
- Community management: when the prompt circulates in a community channel, ask people to read it before pasting it.
- Internal/external/cross-project: none from this source.
#### OSPO
- Policy and governance: none directly. Nearest relevance: a team policy on which agent prompts may read session history (inference, not in the source).
- License compliance: none. The capture says nothing about the repository's license.
- Security management: none directly. Nearest relevance: review any pasted prompt that reads local history (see Questions).
- Software inventory: none.
- Developer support: none from this source.
- Community engagement: none from this source.
- Strategic alignment: none. The post makes no claim about open-source strategy.
#### Community
- Maintainers: no direct action. Nearest relevance: a release post can pair a prompt like this with a plain changelog, so users can check the claims.
- Contributors: none.
- Sponsors: none.
- Ecosystem: none from this source.
#### WordPress
- Core: none. The post doesn't mention WordPress.
- Core Performance: none.
- Core AI: none directly. Nearest relevance: the prompt shows agent skills being managed as files and commands, a pattern WordPress AI work may also face (inference, not in the source).
- **Core Program Team:** none.
- Mobile: none.
- Playground: none.
- Community: none.
- Training: none.
- Marketing: none.
- Support: none.
- Design: none.
- Documentation: none directly. Nearest relevance: docs for any agent skill set can state which files a rename affects.
- Photos: none.
- TV: none.
- Polyglots: none.
- Accessibility: none from this source.
- Meta: none.
- CLI: none.
- Hosting: none.
- Tide: none.
- Openverse: none.
- Themes: none.
- Plugins: none.
#### The WPOCC
- Programs: none.
- Projects: none.
- Funding experiments: none from this source.
### FAIR
- Technical independence: none. The post says nothing about package signing or provenance.
- Policies and covenants: none.
- Productized governance artifacts: none from this source.
## 🧪 Hypotheses
1. Readers who paste a prompt that scans their own session history are less likely to check what the scan reads than readers who run a prompt that only edits files.
2. A recommendation step based on the last 25 sessions produces different suggestions than one based on the last 100, for the same user.
3. A diff of two skill sets surfaces more differences after a rename release than after a release that doesn't rename files.
4. Agents given a rename instruction without a repository path pick the wrong repository in some share of runs.
5. Release announcements written as paste-ready prompts get more adoption within a week than announcements written as changelogs.
## 📊 Methodology
- H1: show two groups a prompt that reads history and a prompt that only edits files, then ask each group what the prompt reads. Compare correct answers.
- H2: for several users with 100 or more sessions, run the recommendation step on the last 25 and the last 100, and have two reviewers rate how much the suggestions overlap.
- H3: pick release pairs with and without renames from a public skills repository and count diff entries per release. This needs the release data, which the capture lacks.
- H4: run the same rename instruction 20 times in a workspace with two similarly named repositories and record which one the agent edits.
- H5: compare adoption signals (stars, installs, replies) for announcements in each format. Adoption data isn't in the capture.
## ❓ Questions
1. What does "scan my last 25 sessions" read: only skill invocation lines, whole transcripts, or tool outputs and file contents too? Where are those sessions stored, and which tool's sessions count?
2. Does the scan send session contents anywhere beyond the agent running it, or stay local?
3. Which repository does "Rename CONTEXT.md to GLOSSARY.md" apply to, and does the rename also update links or references to the old name?
4. What does /setup-matt-pocock-skills do when it finds "required changes": does it only report them, or does it modify files?
5. Is "my repository" the reader's fork, a personal skills directory, or something else?
6. What does the diff compare: file names, file contents, or both, and does it change anything?
7. What did the earlier tweet in ¶6 say about v1.3?
8. What does v1.3 contain, according to its own release notes?
## 🔮 Future research
- Gaps: the capture holds no release notes, no changelog and no description of any skill. Everything about v1.3 depends on sources not captured.
- Direction: read the release page and compare its changes against the six steps, to see which steps follow from the release and which are the author's suggestions.
- Direction: document where common coding agents store session history, so readers can answer Question 1 for their own tool.
- Direction: a record of how prompts-as-announcements are written, to compare with changelog-style posts.
## 💡 Implications
- **Practice.** A prompt that edits files and reads history is two different risk levels in one paste. Reading the steps first, as the numbered list above does, lets a reader run the read-only steps and hold back the edit (inference, not in the source).
- **Policy.** Teams that restrict agent access to local history can ask whether a pasted prompt like ¶3 falls under that rule (inference, not in the source).
- **Security.** The session scan in ¶3 is the step to examine first. The post doesn't say what it reads (Question 1 and 2).
- **Education.** Publishing a release as a prompt teaches readers to see releases as tasks for their own tooling. The post doesn't say whether that's the author's intent.
- **Funding models.** None from this source.
## 🧠 Learning styles assessment
- Reading level estimate: not computed. The sample is 68 words including labels, too short for a stable score. The sentences are short imperatives.
#### 👀 Visual learners
- Present: none. The capture has no image, diagram or code block.
- Improve: a short flow of the six steps, marking which edit files and which only read.
#### 🔊 Aural learners
- Present: none in the capture.
- Improve: a 2-minute screen recording of the prompt running, with narration of each step.
#### 🗣️ Verbal learners
- Present: ¶2 and ¶3 give plain imperative sentences, one per task.
- Improve: a one-line glossary for "skill", "session" and "diff" for readers new to agent tooling.
#### 🧩 Kinesthetic learners
- Present: the prompt itself is a hands-on task the reader runs.
- Improve: a dry-run mode: run steps 1, 4 and 5 first (read-only), then decide on the rename.
#### 🧠 Logical learners
- Present: the order runs from check, to change, to compare, to analyze, to recommend. The sequence is implied, not numbered in the post.
- Improve: number the steps, as this note does, and state what each depends on.
#### 🧑🤝🧑 Social learners
- Present: "pals" in ¶4 addresses a community of readers.
- Improve: a discussion prompt: what did your 25-session scan find, and did the recommendations match your habits?
#### 🧘♀️ Solitary learners
- Present: none. The prompt has no self-check.
- Improve: a reflection prompt: before running it, write down which skills you expect to find in your sessions, then compare.
#### 🌿 Naturalistic learners
- Present: none. The text has no ecosystem or sustainability framing.
- Improve: not relevant for this source.
## 📊 Accessibility assessment
Judged as captured. The capture is plain text from a tweet, so page markup, styling and any media aren't visible.
#### Screen reader semantics
- Present: six short paragraphs in reading order, with bracketed labels from the pipeline. The text shows no headings, lists or images.
- Improve: whether the live tweet has alt text, a lang attribute or a heading structure can't be confirmed here.
#### Keyboard navigation
- Present: not assessable from plain text.
- Improve: the capture shows no interactive elements besides links. Visible focus can't be checked here.
#### Contrast and readability
- Present: unknowable from text. Page styling isn't captured.
- Improve: check text contrast on the live page (WCAG 1.4.3). The short paragraphs help readability.
#### Text resizing and reflow
- Present: the capture reflows as plain text.
- Improve: behavior at 200% zoom and 320 px width can't be verified from the capture (WCAG 1.4.4, 1.4.10).
#### Media and captions
- No media referenced in the captured text.
#### Forms and interactions
- Present: none. No form appears in the capture.
- Improve: for a prompt people copy, the live post's copy behavior and the file names (CONTEXT.md, GLOSSARY.md) being turned into links by the platform could confuse assistive technology users. Whether that happens on the live page isn't confirmed here.
#### Standards reference
- Candidate checks against WCAG 2.2 AA, all unverified from the capture: **1.3.1**, **1.4.3**, **1.4.4**, **1.4.10**, **2.4.4**, **3.1.1**.
## 🌍 DEIB assessment
#### Cultural breadth
- Present: the post is in English and assumes familiarity with coding agents, skills and sessions.
- Improve: define "skill" and "session" on first use for readers outside agent tooling.
#### Inclusive language
- Present: "pals" in ¶4 is informal and friendly. No group labels appear.
- Improve: none needed for the captured text.
#### Neurodiversity support
- Present: short imperative sentences, one task each, in a compact post.
- Improve: numbering the tasks would lower the load of tracking six steps in two paragraphs.
#### Representation
- Present: the text names one person (the author, by first name in "Matt's").
- Improve: not applicable to a single-author prompt post.
#### Psychological safety
- Present: the tone is light and low-pressure ("Enjoy v1.3, pals"). The recommendation step uses "might," not a directive.
- Improve: none needed.
## ✅ Logical validity
Source type: announcement and how-to prompt, a short social post. It states a task list and no argument.
#### Strong reasoning
- **Ordered steps, ¶2 and ¶3.** The tasks go from checking the release, to comparing files, to analyzing usage, to recommending. A recommendation grounded in the reader's own usage is a reasonable design for a personalized release summary.
- **Bounded scan, ¶3.** "Last 25 sessions" gives the agent a fixed window instead of an open-ended history read.
- **How to firm it up.** State what the scan reads and where, and name the repository the rename applies to.
#### Fallacies
- None found. The post makes no argument that a fallacy could attach to. Missing detail (the rename's purpose, the release contents) is an omission in a short post, not a flaw in reasoning.
#### Tone of voice
- Friendly and brisk. Examples: "Enjoy v1.3, pals." and the plain imperatives in ¶2. ¶3 is slightly softer ("Get a sense for", "might improve").
## 🛡️ DARVO analysis
Not applicable to this source. DARVO describes a response to an accusation, and this post is a release prompt with no accusation, denial or counterattack. **Deny:** nothing is denied. **Attack:** no one is attacked. **Reverse victim/offender:** no role reversal appears.
## 🎯 Conclusion
The post is a paste-ready prompt: check mattpocock/skills for v1.3, rename CONTEXT.md to GLOSSARY.md, check one setup command, diff the reader's skills against the author's, scan the last 25 sessions for skill invocations, and recommend workflow improvements. The release's existence and contents are the author's claims, with no changelog in the capture.
The capture doesn't include the repository, the release notes or the earlier tweet, so the rename's purpose and the release's contents are unknown here. This note doesn't describe them.
**Next steps for readers:** read the release page before running the prompt, work out what "scan my last 25 sessions" reads in your own tool, and run the read-only steps before the rename.
## 🔄 Refactor Appendix — Prompt of the day: Matt Pocock on the mattpocock/skills v1.3 release
> Density: 2–5 Literature atoms, 0–3 Permanent candidates.
> Emitted: 3 Literature, 1 Permanent. Resonance filter applied: the sign-off was kept only as a tone cue,
> not as an atom.
>
> Konik claim-header rules: present tense, declarative, ≤10 words, searchable, unique.
>
> Atomicity: one declarative sentence per atom. "and"/"also" → split.
>
> Tag pool: #bmf · #addendum · #xref · #articleseed · #storystem · #share · #aphorisms
### 📚 Literature Split Pad
#### [H1] The prompt tells an agent to diff skills
> [!quote] [¶2]
> Perform a diff of the skills in my repository vs Matt's.
**Context & use**:
- ¶2. The compare step. "My repository" is the reader's, so the agent needs to know which one.
- Useful as the step to run read-only first.
**Tags**: #xref #articleseed
<!-- atomicity: 1 idea ✓ -->
#### [H2] The prompt scans 25 sessions for skill use
> [!quote] [¶3]
> scan my last 25 sessions for skill invocations.
**Context & use**:
- ¶3. The step that reads the reader's own history. What it reads isn't specified (Questions 1 and 2).
- The fixed window of 25 bounds the scan.
**Tags**: #xref #share
<!-- atomicity: 1 idea ✓ -->
#### [H3] The author announces v1.3 in passing
> [!quote] [¶4]
> Enjoy v1.3, pals.
**Context & use**:
- ¶4. The only sentence naming the version outside the ¶2 instruction. Treat the release as the author's claim.
- The capture holds no changelog to check it against.
**Tags**: #xref
<!-- atomicity: 1 idea ✓ -->
### 🧠 Atomic Split Pad
#### [P1] Read what a pasted prompt reads
**Claim** (one sentence, Courtney's voice): Before pasting an agent prompt, list which steps change files and which read local history, and run the read-only ones first.
**Evidence pointer**: [[#H2]]
**Scope & conditions**: Holds for prompts that touch both files and session history. Breaks for prompts whose steps are all read-only or all in a sandbox. The post doesn't say what the scan reads, so this is a working claim to test (see Hypotheses 1).
**Links**:
- [[Agent prompts that read local history]] — because applies-in
- [[Dry-run steps before a rename]] — because evidence-for
- [[Release announcements as agent prompts]] — because contrasts on changelog-style posts
<!-- atomicity: 1 idea ✓ -->
### ✏️ Fleeting Queue
- [F1] Read the mattpocock/skills v1.3 release page before running the prompt, Pocock tweet 2026-10-04 #fleeting #inbox
- [F2] Open question: what does the 25-session scan read, and where does that history live in my tool? #fleeting #inbox
- [F3] Check the earlier tweet linked as "Previous:" for v1.3 context #fleeting #inbox #articleseed
### → Things 3 (max 2 actions)
- [ ] Read the mattpocock/skills v1.3 release notes and the earlier tweet before running Pocock's prompt, tweet 2026-10-04 #things
- [ ] Write down what "scan my last 25 sessions" reads in my agent tool before pasting the prompt #things
<!-- deep-read:sections end -->
📄 **Full text:** [[Full Document Contents/Tweets/Prompt of the Day.md|Prompt of the Day: by Matt Pocock]]
## Highlights
- - Re-check mattpocock/skills for the v1.3 release. Rename CONTEXT.md to GLOSSARY.md. Check /setup-matt-pocock-skills for any required changes. Perform a diff of the skills in my repository vs Matt's. ([View Highlight](https://read.readwise.io/read/01m4659atnen5htqyf20rsv1yf))
- Note: [H1] The prompt tells an agent to diff skills
- Tags: atom/literature/h1
- - Get a sense for how I use Matt's skills: scan my last 25 sessions for skill invocations. Then, recommend how this release might improve my workflow. ([View Highlight](https://read.readwise.io/read/01m4659d1bmkvabtvwtth90gx3))
- Note: [H2] The prompt scans 25 sessions for skill use
- Tags: atom/literature/h2
- Enjoy v1.3, pals. ([View Highlight](https://read.readwise.io/read/01m4659f776gr9jpwaawf3j8sj))
- Note: [H3] The author announces v1.3 in passing
- Tags: atom/literature/h3