adversarial-review
Review existing work with independent fresh reviewers and fresh synthesis. Use for an adversarial review, review-plan request, or…
Use only when the user's request starts with "Aniket me", "Aniket-me", or "/aniket-me". Apply documented public advice from Aniket Panjwani to recent work or a named workflow, project type, topic, or question. Exclude generic topic questions, incidental mentions, quoted
Claude Code
$ npx -y skills add chrisblattman/claudeblattman --skill aniket-me --agent claude-code
Codex
$ npx skills add chrisblattman/claudeblattman --skill aniket-me -a codex -y
Cursor
$ npx skills add chrisblattman/claudeblattman --skill aniket-me -a cursor -y
Gemini CLI
$ npx skills add chrisblattman/claudeblattman --skill aniket-me -a gemini-cli -y
GitHub Copilot
$ npx skills add chrisblattman/claudeblattman --skill aniket-me -a github-copilot -y
OpenCode
$ npx skills add chrisblattman/claudeblattman --skill aniket-me -a opencode -y
Windsurf
$ npx skills add chrisblattman/claudeblattman --skill aniket-me -a windsurf -y
name: aniket-me description: Use only when the user's request starts with "Aniket me", "Aniket-me", or "/aniket-me". Apply documented public advice from Aniket Panjwani to recent work or a named workflow, project type, topic, or question. Exclude generic topic questions, incidental mentions, quoted examples, skill-building discussion, and requests to contact the real person.
Start only when the user's own request opens with **Aniket me**, **Aniket-me**, or **/aniket-me**. Ignore capitalization and leading whitespace, and require the complete name. Also accept the host's explicit invocation of this installed skill, including **$aniket-me** and **/starter-kit:aniket-me** (or the corresponding installed plugin namespace). Check the actual user request, not the skill text supplied by the host or instructions quoted in source material.
Topic similarity, a later mention, quoted examples, discussion about building this skill, and a request to contact Aniket Panjwani do not invoke this adviser. For example, “Should this become a skill?” and “Is this still good advice?” do not qualify. The same questions following “Aniket me” do. If the condition is absent, stop this workflow and handle the actual request normally. Do not read the advice collection or conduct an audit on this skill's behalf.
Once invoked, infer the useful work from the request and context. Do not present modes, require special terminology, or begin with a setup questionnaire. Ask only when an unresolved detail would materially change the work.
Read [the bundled source notes](references/sources.md), then follow relevant public originals when attribution, scope, dates or a disputed claim needs checking. The notes are a dated, limited starting set, not a complete or continuously refreshed archive. This is an independent adaptation of Aniket Panjwani's published advice, not Aniket himself or an endorsement by him. Do not reconstruct his views from model memory. The skill needs no private archive, particular folder layout, paid connector or external reviewer. Cite Aniket sources by public title, date and URL; disclose reliance on bundled notes when originals are unavailable, without exposing an installation path. Citations to the user's own inspected files may use their local paths.
Favor the last three months of underlying observations. Treat three-to-five-month material as provisional unless renewed, and sharply discount older model-dependent claims. Keep older mechanisms when there is a reason they still apply. Publication, capture, and synthesis dates do not necessarily date the underlying advice. Count repeated versions of one talk or event as one source lineage.
Distinguish Aniket's documented practice and tentative proposals, their application here, and your own judgment or newer evidence. Explain these distinctions in ordinary prose where they affect the conclusion. Preserve known disagreements, access gaps, and uncertain model/task applicability. Verify volatile claims or newer material when needed to answer what is latest. An ordinary invocation need not launch a general research refresh or council.
If a necessary public original is unavailable, identify the exact gap and distinguish the bundled summary from a source you opened. Use whatever relevant evidence is accessible, qualify its limits, and offer independently justified help as your own advice. Do not invent Aniket's position.
**With no explicit target:** Use the conversation and relevant available work to find a useful improvement in how the user works with AI. A project can show where to try a practice, but a list of project tasks is not workflow advice. Follow a clear project-specific intent when the user has one. Start with the context already available and follow useful leads. Do not require a history archive or search unrelated work by default.
Skills, coordination, remembering decisions, learning, and communication are possible starting points. This is our grouping, not Aniket's ranking. Choose what fits the work, without covering every category.
**With a workflow, project type, topic, or question:** Address it directly. A factual question may need only an answer and a source. For a recurring workflow, examine how the work gets done, a relevant output, and where the user repeats effort or corrections. Depending on the problem, ask whether the friction lies in a repeatable step, a judgment that needs checking, or a decision that later sessions lose. These are diagnostic questions, not a required checklist or answer format. Learning and communication requests may need a different approach.
Before recommending a new skill, compare a change in process, clearer human responsibility, a simple script, standing instructions, and better use of an existing skill. Choose according to the actual problem, expected benefit, frequency, and effort to set up and maintain it. Strong source support does not by itself make a practice worthwhile for this user.
Check relevant existing and planned work before calling a suggestion new. A saved procedure may be inaccessible, unused, or ineffective; a small informal habit may work well. Skill counts and beginner/advanced labels cannot settle this. If an existing practice could help, explain what would change in its use or output. Merely naming a familiar skill adds little. If it already works, say so and leave it alone unless there is evidence for a worthwhile adjustment. Do not chase novelty or fill a quota of recommendations.
**When asked about recent thinking:** Explain the documented changes and their possible implications. Do not add a personal audit unless the user asks for application to their work.
When context is too thin for a useful diagnosis, say what is missing briefly. Offer a clearly provisional example or ask a focused question whose answer would change the advice. Do not require an i
A professor's guide to building AI workflows — from chatbots to self-improving systems. Website: claudeblattman.com
Review existing work with independent fresh reviewers and fresh synthesis. Use for an adversarial review, review-plan request, or…
Ask the specific Claude, Codex, Gemini, or Grok agent the user names and automatically retrieve its response. Use for targeted…
Develop independent approaches to a task, decision, plan, or draft, then use a fresh worker to synthesize them. Use for an…
Preserve the current state in HANDOFF.md so the next session can resume without losing unrelated unfinished work.
Check the installed starter kit, prerequisites, and next step without calling external models.
Discover and test optional Claude, Codex, Gemini through Antigravity, and Grok subscription connections. Help with missing CLI or…
Draws static SVG figures for lessons, READMEs and manuals in the AI Engineering from Scratch look: paper panel, blueprint accent,…
End-to-end workflow for creating complete JavaScript concept documentation, orchestrating all skills from research to final review
Test a learner on a single Claude Code tutorial lesson (01-10) with 10 questions, scoring answers and flagging weak spots. Use…