Brainforge Knowledge

doordash-packaging-guide

6 min read

Doordash — Packaging Guide

Owner: Brylle Girang
Created: 2026-04-01
Related: Doordash Workflow Plan · PR Scan Guide · SLA

Purpose

Once the weekly PR scan produces a classified list of updates, this guide tells L&D how to turn each update into a learning unit. The goal is not to summarize the PR — it is to produce something a team member can read, understand, and apply to their work immediately.

Education science principles (apply to all tiers)

These three principles ground every packaged update. They come from the claude-education-skills library and apply regardless of tier.

1. Cognitive load first

(memory-learning-science/cognitive-load-analyser.md)

  • Strip each update to its single critical behavior change. One update = one behavior.
  • Remove extraneous detail. The “why it was built,” the implementation details, and the engineering context belong in the Forge vault (the PR itself is the reference). The delivery unit carries only what changes how the team member acts tomorrow.
  • If a PR contains more than one behavior change, split it into two separate updates. Do not batch unlike behaviors into one delivery.

2. Explicit instruction structure

(explicit-instruction/explicit-instruction-sequence-builder.md)

Every update — regardless of tier — should be structured as:

  • I Do: L&D (or the automation) demonstrates the updated behavior. Show it; don’t describe it. Use a before/after example, a short screen recording, or a Zoom Clip.
  • We Do: A guided example the learner can follow alongside. A written walkthrough or annotated screenshot works.
  • You Do: One real micro-task the learner can complete in their actual work. Not a quiz — a real action. “Next time you run [X skill], do [Y].”

For Tier 1 digest entries, We Do and You Do can be condensed into a single “try this” line. For Tier 3, all three phases should be explicit.

3. Implementation intentions

(wellbeing-motivation-agency/implementation-intention-designer.md)

End every update with an if-then trigger:

“When I next [specific real-work context], I will [new behavior].”

Examples:

  • “When I next run an EP audit, I will use the updated skill which now auto-saves to the vault.”
  • “When I next create a Linear ticket from a transcript, I will check that the Context section follows the new template.”

This bridges reading to doing. Without the trigger, team members intend to use the update but forget when the moment arrives.

Tier 1 — Weekly digest

Format: Single Slack post, published every Thursday.
What it contains: All Tier 1 updates from the week, batched together.
Length: Aim for the whole digest to be readable in 2 minutes.

Template

📦 This week in the platform — [DATE]

[Repeat the block below for each Tier 1 update]

*[UPDATE TITLE — e.g. "Renamed: meeting-prep skill"]*
What changed: [One sentence. Before → After.]
Try it: [One concrete action. Start with a verb. "Next time you use X, do Y."]

---

[Next update title]
What changed: ...
Try it: ...

---

Questions? Drop them in #ai-learning or DM Brylle.

Notes

  • No headers, no bold sections, no bullet-point walls. Short paragraphs only.
  • If there are no Tier 1 items this week, still post the digest with: “Quiet week — nothing new to ship. The platform team is heads down on [initiative name if known].”
  • Do not include Tier 2 or Tier 3 items in the digest. They get their own posts.

Tier 2 — Standalone Slack post

Format: Single Slack post per update (not batched).
When to post: Within 5 business days of the PR being detected in the scan (see SLA).
Length: Readable in 3–5 minutes.

Template

🔄 Workflow update: [UPDATE TITLE]

*What changed:*
[2–3 sentences. What the old behavior was, what the new behavior is, and why it matters for the team's work.]

*How to use it:*
[Guided walkthrough. Step by step. Use numbered list. Include a before/after example if possible.]
1. [Step]
2. [Step]
3. [Step]

*Try it now:*
[One real micro-task. Specific enough that someone can do it in the next 10 minutes.]

*Full context:* [Link to the PR or Forge page if a permanent reference was created]

Questions? #ai-learning or DM Brylle.

Notes

  • Create a Forge page for Tier 2 updates that involve a workflow people will reference repeatedly (i.e., the post will go stale but the Forge page won’t). Link to it in the post.
  • If the update is service-line-specific, post to the SL channel first, then cross-post a brief notice to the primary channel.
  • Use a Zoom Clip (2 min max) if a written walkthrough isn’t sufficient to show the new behavior clearly.

Tier 3 — Forge page + Zoom Clip

Format: Permanent Forge page + 2–4 min Zoom Clip.
When to deliver: Scheduled with the relevant team — no fixed SLA.
Length: Forge page is comprehensive; Zoom Clip is the guided demo only.

Forge page structure

# [UPDATE TITLE]
 
**Published:** [DATE]  
**Audience:** [All / Service line]  
**Related PR:** [link]  
**Zoom Clip:** [link]
 
---
 
## What changed
 
[1–2 paragraphs. Context: what existed before, what is new, why this matters for delivery quality.]
 
## How it works
 
[Step-by-step guide with screenshots or annotated examples. Full walkthrough — not condensed.]
 
## Applied task
 
[One task grounded in real work. More detailed than Tier 2 — this is a new system, so the task should cover the core use case end-to-end.]
 
When I next [trigger], I will [new behavior].
 
## FAQ
 
[Add questions as they come up in #ai-learning after the delivery.]
 
## Reference
 
- PR: [link]
- Linear ticket: [link if applicable]

Zoom Clip guidance

  • Record in Cursor (or the tool being demonstrated) — not a slide deck.
  • Show the I Do stage only: L&D doing the full new workflow once, narrated.
  • Target 2 minutes. Cap at 4 minutes. Anything longer means the scope is too large — split it.
  • Upload to Zoom Clips and embed the link in the Forge page.

Packaging checklist (use before posting)

Before any delivery goes out, confirm:

  • Single behavior change. No compound “and also” updates.
  • I Do / We Do / You Do structure present (or condensed appropriately for Tier 1).
  • Implementation intention included (“When I next… I will…”).
  • Audience correctly identified — all-team or specific SL.
  • Format matches the tier (digest / Slack post / Forge page).
  • Changelog row ready to log after posting.

What not to do

  • Do not summarize the PR. The PR is not the delivery — it is the source. The delivery is what changes how someone works.
  • Do not use passive voice for the “Try it” or “You Do” instructions. “The skill can be invoked by…” → “Run the skill by…”.
  • Do not ship a Tier 3 without a Zoom Clip. Written-only is not sufficient for a net-new system.
  • Do not post without an implementation intention. It is the single most important line in any delivery.

Graph View

Backlinks