Improve someone else’s skill.
Start with Feedforward’s Praxis skill-making skill if you have access. You can also do this in your LLM: let it study someone else’s skill, discover what you wish it did better, and make a version that works for you.
Try it now
Copy the prompt into your AI tool of choice, then answer its questions as they come.
Paste it into any AI chat to start. For most people, this is all you need.
Everything on this page in one paste: the prompt plus the full plan, every step and what to check. Use it for longer moves, to plan ahead, to share with a colleague, or if your AI can't open links.
How it goes
What the AI will do
- Start by looking for Feedforward’s Praxis skill-making skill, then inspect how skills and skill-writing tools work in your AI environment.
- Study the existing skill, find where it falls short, interview you about improvements, and build and test a revised copy.
What to have ready
- The skill itself, a link or its name and where you use it. Rough examples of disappointing results are welcome.
Where you come in
- Explain what better would look like and choose the improvements that matter.
- Try the revised skill on real work; complete any installation step that requires your account.
What you'll get
An improved skill in the format your tool supports, with an example of how to invoke it, a comparison with the original and a clear account of installation and testing.
You're done when
You have the revised skill and know how to use it again. Its comparison with the original shows what was tested; any remaining installation or user trial is identified precisely.
These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result. ### Start with the activity and the user's context Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear. Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them. ### Check the capabilities needed for this work Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit. When useful access is missing: - Explain what you cannot reach and how that material would help. - Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot. - If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response. - Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available. ### Ask focused questions and respect the answers Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful. Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements. ### Keep evidence and independent contributions distinct Distinguish evidence, inference, and missing information. Check your work against the source material. Use independent agents when your system actually supports delegation and independent investigation or review would improve the work: - Give each agent a bounded brief and the context appropriate to its role. - Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations. - Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately. ### Produce, check, and revise the result Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them. Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result. Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized. Help me improve someone else’s skill for work I actually do. Start by suggesting Feedforward’s Praxis skill-making skill. If it is unavailable or unsuitable here, guide and perform the work in this LLM using the environment’s supported skill-writing tools. Produce a usable improved skill, not just advice about improving one. ### 1. Find the skill and learn how this environment uses it Inspect available tools, installed skills and local or connected documentation. Look for Feedforward’s Praxis skill-making skill and other skill-writing helpers; read any accessible instructions before using them. The Praxis link is not supplied here: do not invent one, confuse it with unrelated products named Praxis, or assume how it works. If it cannot be found, say so briefly and continue through the LLM route. Ask for a known Praxis link only if I want to use it and your search has not found it. Identify how this environment discovers and invokes skills, its actual supported file or package format, supporting resources and scripts, installation location or import method, and any required refresh. Use current official documentation where local instructions do not answer those questions. Distinguish an installable skill from reusable prompt text if this environment lacks native skills; help identify a compatible target and make the export for that target. Locate the original skill from the name, file or link I supplied. Read its instructions and relevant examples, templates, scripts and references. Search accessible usage conversations, issue threads, colleague feedback and project documents for the task it serves, recurring failures, accepted output examples and previous fixes. Use those records to understand the intended outcome and avoid asking me to reconstruct what you can inspect. Keep the search bounded to this skill and its work. If you cannot access it, investigate an export or upload route and ask for the skill files or text; a title alone is insufficient for claiming to improve the original. ### 2. Show what it does now and interview me about better Explain its trigger, required inputs, major actions, questions, output and tool dependencies in ordinary language. Trace at least one supplied example through the instructions. Where supported, run a representative task with the original skill and retain its actual input and output. Use a copy or test destination for actions that would change real work. If execution is unavailable, distinguish reading the instructions from observing behavior. Recommend a few specific improvements, each tied to a frustrating result, missing instruction, unnecessary user effort or useful capability the environment actually provides. For example, the skill might ask me for context it could retrieve, stop at suggestions when it could create the file, or produce a polished answer without checking the underlying figures. Interview me one focused question at a time, building on that inspection: where do I intervene, what do I repeatedly explain, what should the skill preserve, and what would make me want to use it again? Do not turn this into a questionnaire I must complete before you help. Confirm the selected changes and a concrete example of success; wait before making a revision that depends on an unresolved preference. ### 3. Build the revised skill Keep an identifiable original version and preserve attribution and any license notices. Create a separate revised copy so we can compare it with the original. Use Praxis or another available skill-writing helper according to its actual instructions where useful; otherwise author the supported files yourself. Make the revised skill self-contained for an AI that has not read this conversation. Include when to invoke it, useful input forms, relevant context searches and their purpose, the actual task sequence, missing-access handling, questions and decisions that require the user, output format, checks and stopping point. Supply required templates, examples, scripts or references when the work calls for them; do not add a scaffold for its own sake. Keep original behavior that already works. Explain the substantive changes with a short before-and-after example. If the skill produces audience-sensitive work, build its context search around evidence of that audience’s role, expressed needs and earlier responses. Treat an imagined reaction as a hypothesis. For a skill with several independent components, divide inspection or tests into bounded assignments with exact files, test inputs and expected evidence. Maintain a coverage list and reconcile findings; use successive passes when separate agents are unavailable. ### 4. Test whether it is actually better Compare original and revised versions on the same representative input against the success criteria we agreed. Check useful dimensions such as correctness, completion, output usability and how much correction or explanation I must supply. Do not infer consistent improvement from prettier instructions. Give a fresh reviewer the revised files, agreed requirements and an unfamiliar realistic input without your explanation of the changes. Include a case with missing information or inaccessible tools when that is relevant to the skill. Ask the reviewer to execute the method and report concrete failures, unintended changes and unfinished work. If fresh review is unavailable, make a separate bounded pass and disclose the shared context. Repair observed problems and repeat the affected trials. Retain actual evidence and separate executed tests from editorial inspection. ### 5. Make it available and help me use it Install or import the revised skill using the supported method when you have access and authorization. Verify that the target tool discovers it, then invoke it on a small real example. If I must perform a login, import or restart, prepare the exact files and guide that specific action; wait for the result and resume verification. If native installation is impossible, provide the complete supported export and exact remaining action, or an explicitly agreed reusable-prompt alternative. Show me a concrete invocation using the skill’s actual name and trigger. Invite me to try it on something I would really use, ask what happened, and wait for my response before assessing my experience. Inspect the output where possible and fix observed problems. If I cannot try it yet, preserve the tested package and distinguish your trials from my unperformed trial. Finish with the usable skill, where to find it, how to invoke it again, what changed and what the comparison established. State any remaining installation or testing step precisely. Do not call it installed, invoked or improved in practice unless the corresponding evidence exists. These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.
View full AI instructions
Improve someone else’s skill.
What the AI will do
- Start by looking for Feedforward’s Praxis skill-making skill, then inspect how skills and skill-writing tools work in your AI environment.
- Study the existing skill, find where it falls short, interview you about improvements, and build and test a revised copy.
What you can provide
- The skill itself, a link or its name and where you use it. Rough examples of disappointing results are welcome.
Where you'll take part
- Explain what better would look like and choose the improvements that matter.
- Try the revised skill on real work; complete any installation step that requires your account.
What you'll get
An improved skill in the format your tool supports, with an example of how to invoke it, a comparison with the original and a clear account of installation and testing.
Activity context
Start with Feedforward’s Praxis skill-making skill if you have access. You can also do this in your LLM: let it study someone else’s skill, discover what you wish it did better, and make a version that works for you.
For the user
Give the AI the skill or point it to where the skill lives. You do not need to know how skills are packaged. The AI will investigate that, recommend changes and ask about the moments that frustrate you. You choose what better means; it does the rewriting, testing and packaging.
Guidance for the AI
These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.
Start with the activity and the user's context
Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.
Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.
Check the capabilities needed for this work
Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.
When useful access is missing:
- Explain what you cannot reach and how that material would help.
- Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
- If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
- Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.
Ask focused questions and respect the answers
Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.
Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.
Keep evidence and independent contributions distinct
Distinguish evidence, inference, and missing information. Check your work against the source material.
Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:
- Give each agent a bounded brief and the context appropriate to its role.
- Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
- Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.
Produce, check, and revise the result
Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.
Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.
Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.
Your AI prompt
Help me improve someone else’s skill for work I actually do. Start by suggesting Feedforward’s Praxis skill-making skill. If it is unavailable or unsuitable here, guide and perform the work in this LLM using the environment’s supported skill-writing tools. Produce a usable improved skill, not just advice about improving one.
1. Find the skill and learn how this environment uses it
Inspect available tools, installed skills and local or connected documentation. Look for Feedforward’s Praxis skill-making skill and other skill-writing helpers; read any accessible instructions before using them. The Praxis link is not supplied here: do not invent one, confuse it with unrelated products named Praxis, or assume how it works. If it cannot be found, say so briefly and continue through the LLM route. Ask for a known Praxis link only if I want to use it and your search has not found it.
Identify how this environment discovers and invokes skills, its actual supported file or package format, supporting resources and scripts, installation location or import method, and any required refresh. Use current official documentation where local instructions do not answer those questions. Distinguish an installable skill from reusable prompt text if this environment lacks native skills; help identify a compatible target and make the export for that target.
Locate the original skill from the name, file or link I supplied. Read its instructions and relevant examples, templates, scripts and references. Search accessible usage conversations, issue threads, colleague feedback and project documents for the task it serves, recurring failures, accepted output examples and previous fixes. Use those records to understand the intended outcome and avoid asking me to reconstruct what you can inspect. Keep the search bounded to this skill and its work. If you cannot access it, investigate an export or upload route and ask for the skill files or text; a title alone is insufficient for claiming to improve the original.
2. Show what it does now and interview me about better
Explain its trigger, required inputs, major actions, questions, output and tool dependencies in ordinary language. Trace at least one supplied example through the instructions. Where supported, run a representative task with the original skill and retain its actual input and output. Use a copy or test destination for actions that would change real work. If execution is unavailable, distinguish reading the instructions from observing behavior.
Recommend a few specific improvements, each tied to a frustrating result, missing instruction, unnecessary user effort or useful capability the environment actually provides. For example, the skill might ask me for context it could retrieve, stop at suggestions when it could create the file, or produce a polished answer without checking the underlying figures.
Interview me one focused question at a time, building on that inspection: where do I intervene, what do I repeatedly explain, what should the skill preserve, and what would make me want to use it again? Do not turn this into a questionnaire I must complete before you help. Confirm the selected changes and a concrete example of success; wait before making a revision that depends on an unresolved preference.
3. Build the revised skill
Keep an identifiable original version and preserve attribution and any license notices. Create a separate revised copy so we can compare it with the original. Use Praxis or another available skill-writing helper according to its actual instructions where useful; otherwise author the supported files yourself.
Make the revised skill self-contained for an AI that has not read this conversation. Include when to invoke it, useful input forms, relevant context searches and their purpose, the actual task sequence, missing-access handling, questions and decisions that require the user, output format, checks and stopping point. Supply required templates, examples, scripts or references when the work calls for them; do not add a scaffold for its own sake. Keep original behavior that already works. Explain the substantive changes with a short before-and-after example.
If the skill produces audience-sensitive work, build its context search around evidence of that audience’s role, expressed needs and earlier responses. Treat an imagined reaction as a hypothesis. For a skill with several independent components, divide inspection or tests into bounded assignments with exact files, test inputs and expected evidence. Maintain a coverage list and reconcile findings; use successive passes when separate agents are unavailable.
4. Test whether it is actually better
Compare original and revised versions on the same representative input against the success criteria we agreed. Check useful dimensions such as correctness, completion, output usability and how much correction or explanation I must supply. Do not infer consistent improvement from prettier instructions.
Give a fresh reviewer the revised files, agreed requirements and an unfamiliar realistic input without your explanation of the changes. Include a case with missing information or inaccessible tools when that is relevant to the skill. Ask the reviewer to execute the method and report concrete failures, unintended changes and unfinished work. If fresh review is unavailable, make a separate bounded pass and disclose the shared context. Repair observed problems and repeat the affected trials. Retain actual evidence and separate executed tests from editorial inspection.
5. Make it available and help me use it
Install or import the revised skill using the supported method when you have access and authorization. Verify that the target tool discovers it, then invoke it on a small real example. If I must perform a login, import or restart, prepare the exact files and guide that specific action; wait for the result and resume verification. If native installation is impossible, provide the complete supported export and exact remaining action, or an explicitly agreed reusable-prompt alternative.
Show me a concrete invocation using the skill’s actual name and trigger. Invite me to try it on something I would really use, ask what happened, and wait for my response before assessing my experience. Inspect the output where possible and fix observed problems. If I cannot try it yet, preserve the tested package and distinguish your trials from my unperformed trial.
Finish with the usable skill, where to find it, how to invoke it again, what changed and what the comparison established. State any remaining installation or testing step precisely. Do not call it installed, invoked or improved in practice unless the corresponding evidence exists.
Done when
You have the revised skill and know how to use it again. Its comparison with the original shows what was tested; any remaining installation or user trial is identified precisely.
These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.
These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.