Agent benefit review en
Agent Skills that teach and implement the **New Work with Skills** methodology
npx -y skills add roebi/new-work-skills --skill agent-benefit-review-enAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Guides a user and a chat agent through a structured monthly review focused on the benefits gained from using the agent. Use this skill at the end of each month when the user wants to reflect on value created, time saved, friction reduced, knowledge gained, or next improvements. Activate for phrases like: "monthly review", "agent benefit review", "did this help me", "value of the agent this month", "retrospective with the agent", or "end of month review".
The file declares its own license as CC BY-NC-SA 4.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
6.5 KB, ~1.2k tokens by cl100k_base, as published. Nobody here has run it
Agent Benefit Review
A structured monthly conversation between user and chat agent.
Purpose: - Reflect on the concrete benefit created by using the agent. - Identify what worked well. - Capture missed opportunities. - Decide one improvement for next month. - Agree the next review date. - Create a reusable markdown archive.
The goal is practical clarity.
Keep the tone calm and concise.
The review should take about 10--15 minutes.
Session setup
Before the review begins capture both names.
Agent asks:
"Before we begin:
What is your name?"
Wait.
Then ask:
"What name should I use for me, the Agent in this review?"
Examples: - <Agent Name>{=html} - Software Architect - Software Developer
Store:
User Name: <User Name>{=html}
Agent Name: <Agent Name>{=html}
Date: <YYYY-MM-DD>{=html}
Month Reviewed: <YYYY-MM>{=html}
Then continue:
"Thanks.
Monthly review for <Month Reviewed>{=html}.
Participants: - User: <User Name>{=html} - Agent:
<Agent Name>{=html}
Goal: understand what benefit was created through working together this month.
We go step by step."
When to use this skill
Activate when: - The calendar month ends - The user wants a retrospective - The user asks whether the agent is worth using - The user wants measurable benefit - The user wants to improve collaboration
Trigger phrases: - monthly review - end of month - benefit review - agent retrospective - was the agent useful - value created this month
Conversation principles
The agent should: - Ask one question at a time - Wait for answers - Summarize briefly after each section - Prefer concrete examples - Ask for numbers when available - End with one actionable improvement
1. Opening
Ask:
"Where did I help you most this month?"
Summarize briefly.
2. Concrete wins
Ask: - What is one thing you completed faster? - What is one thing you completed better? - What might not have started without the agent?
Summarize top wins.
3. Time and effort
Remark: - Remember here that both know that bekoming a best trimmed Agent having Skills use time.
4. Quality and confidence
Ask: - Did i as Agent ask for Requirements akively, or do i just fall in this low performance loop to not ask about requirements and give fast output. In other Words: Garbage In - Garbage out. Did output quality improve if i have better Requirements? - Did confidence improve?
Summarize.
5. Learning and growth
Ask: - What did you learn? - Did the agent help you think differently? - Which skill improved?
Summarize.
6. Friction and missed opportunities
Ask: - What should it do better? - Where did you forget to use it?
Summarize.
7. Value score
Ask:
"Goes the relavive value up, same or down by using the agent this month compared to last moth?"
Then: - Why that relative value? - What give value up next month?
8. One improvement for next month
Ask:
"What is one specific way we should work differently next month?"
Capture one improvement. And save it in Memory.
9. Final summary
Create:
YYYY-MM-DD Monthly Agent Benefit Review --- Summary
Date: <YYYY-MM-DD>{=html}
Month reviewed: <YYYY-MM>{=html}
Participants: - User: <User Name>{=html} - Agent:
<Agent Name>{=html}
Top wins: - ...
Used time remark: - ...
Quality improvements: - ...
Learning: - ...
Friction: - ...
Value score: - ... of Values up, same, down
Next month improvement: - ...
10. Next appointment
Before export ask:
"Our review is complete.
Before I write the markdown file:
let us schedule the next end-of-month review.
Suggested next appointment: <YYYY-MM-last-day>{=html}
Do you agree?"
If no:
Ask:
"What date should we use?"
Store in Memory, may update existing Memory Entry:
Next appointment: <chosen date>{=html}
Status: agreed
Confirm:
"Confirmed.
Next monthly review: <User Name>{=html} + <Agent Name>{=html}
Date: <chosen date>{=html}"
11. Export
Only after the next appointment is agreed.
Generate one markdown file.
Filename format:
YYYYMMDD_<Agent-Name>{=html}_<User-Name>{=html}_Benefit_Review.md
containing
Rules: - Date numeric - Spaces become hyphens - Only letters, digits,
hyphen, underscore - Keep .md
The markdown file must contain: - title - date - month reviewed - participants - summary - value score - next month improvement - next appointment
Final message:
"Your monthly review is complete.
Next appointment agreed: <chosen date>{=html}
Markdown file ready:
<filename>{=html}"
Optional long-term tracking
Month Top benefit Score Improvement
Jan
Feb
Mar
Useful questions: - Is value increasing? - Which workflows repeat? - Where is the best ROI? - Which friction stays unresolved?
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.