Field Notes - akash-coded/aws-bedrock-agentcore-strands GitHub Wiki

Field Notes

What happened when somebody ran this method on a real project. Dated, first person, one page section per note.

The tutorial and the manual say what should happen. This page is for what did. A field note is worth more than a review: it says which step was harder than the lesson made it sound, which number was wrong for your industry, and what you would do differently. Nothing here needs a pull request, and nothing here is edited for tone.

The test for a note: would a team starting the same kind of project next month be glad you wrote it?


How to add one

  1. Copy the template below to the bottom of this page, newest first. Edit this page in the browser; the Wiki edits workflow copies it back into the repository.
  2. Keep the project anonymous if you need to, but keep the numbers real. A note that says "a bank" with a real bar and a real bill is more useful than a named client with rounded figures.
  3. Link the lesson or how-to you were following, so the next reader can see where the method and the project parted ways.
  4. If your note ends in a number the Cost Log or the Error Index should carry, add it there too, and link back.

Prefer a conversation to a page? Post it in Show and tell and link it from here. The best ones get curated into Community Answers.

The template

## <A one-line title: what the project was, in the past tense> · <YYYY-MM>

**Context.** <Industry, team size, your role, and which phase you were in. Two sentences.>
**Following.** <The lesson, how-to or manual step you were using, as a link.>

**What the method said.** <One paragraph.>

**What actually happened.** <One paragraph. Numbers welcome: the bar, the bill, the days to cut over.>

**What I would do differently.** <Three bullets at most.>

**Evidence.** <A link, a screenshot, a figure. Optional but persuasive.>

— <name or handle>, <role>

Notes

None yet. Yours goes here.

The first entry sets the tone for the page. If you ran the ninety-minute team session, the twelve-week reading group or a cohort, that counts: say what landed, what did not, and what you cut.

⚠️ **GitHub.com Fallback** ⚠️