You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: workshop/side-quest-10-01-agent-brief.md
+6-28Lines changed: 6 additions & 28 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -37,11 +37,7 @@ Replace the bracketed example below with your own one-sentence goal.
37
37
Every day, I want the agent to [summarize open pull requests and post a health report as an issue comment].
38
38
```
39
39
40
-
Quick check:
41
-
42
-
-[ ] My goal is one sentence.
43
-
-[ ] My goal describes one main action.
44
-
-[ ] My goal says where the result should appear.
40
+
Before moving on, confirm your goal is one sentence that describes one action and says where the result will appear.
45
41
46
42
<details>
47
43
<summary>Why this works</summary>
@@ -64,11 +60,7 @@ List the data the agent must collect before it can write the report. Mark uncert
64
60
65
61
Add a `?` only on the lines you are not sure about yet.
66
62
67
-
Quick check:
68
-
69
-
-[ ] I listed at least three inputs.
70
-
-[ ] Each input connects to a line in my report.
71
-
-[ ] I marked any uncertain data with a `?`.
63
+
Before moving on, confirm you have at least three inputs, each linked to a field in your report, and that you've marked uncertain items with `?`.
72
64
73
65
<details>
74
66
<summary>Why this works</summary>
@@ -91,11 +83,7 @@ CI: {status}
91
83
Health check: {one sentence}
92
84
```
93
85
94
-
Quick check:
95
-
96
-
-[ ] My output has a title or heading.
97
-
-[ ] Every placeholder maps to one input.
98
-
-[ ] I could scan this report in a few seconds.
86
+
Before moving on, confirm your skeleton has a title or heading, every placeholder maps to one of your inputs, and you can scan the whole report in a few seconds.
99
87
100
88
<details>
101
89
<summary>Why this works</summary>
@@ -116,14 +104,10 @@ Add short rules that limit write operations, such as posting comments, and tell
116
104
- If [data is missing or a prerequisite is absent], then [fallback].
117
105
```
118
106
119
-
> [!IMPORTANT]
107
+
> [!TIP]
120
108
> Skipping guardrails can lead to duplicate comments or guessed data.
121
109
122
-
Quick check:
123
-
124
-
-[ ] I included something the agent must not do.
125
-
-[ ] I set a maximum number of writes.
126
-
-[ ] I told the agent what to do when data is missing.
110
+
Before moving on, confirm your guardrails include something the agent must not do, a maximum number of writes, and a fallback for missing data.
Read the draft once. Replace vague words like "recent" with "within the last 7 days" or "important" with "labeled priority-1".
140
124
141
-
Quick check:
142
-
143
-
-[ ] My goal says exactly what happens every day.
144
-
-[ ] My inputs cover every field in the output.
145
-
-[ ] My output format matches what I want to read.
146
-
-[ ] My guardrails limit writes and guessing.
147
-
-[ ] I replaced at least one vague phrase.
125
+
Before moving on, confirm that your goal, inputs, output format, and guardrails all use concrete, observable language and that you've replaced at least one vague phrase.
0 commit comments