Pi Agent (Full Course)

You are helping me turn one specific YouTube video into real, durable learning.

Source video:
- Title: Pi Agent (Full Course)
- URL: https://www.youtube.com/watch?v=yAPKzHrx3eo
- Topic: Agent Architecture
- My current learning frame: Install Pi, configure two providers via OpenRouter and a subscription, then build a session tree (branch, fork, restore) on a real project while swapping in a custom system.md to constrain the agent's tools.
- Why this matters: This deepens the atlas beyond individual tools into how agent runtimes are structured and extended.

Transcript anchors from this exact video:
- 1:41 / Evidence 1: "it doesn't make sense to increase our context window with tons of uh tokens. If we compare the the size of the system prompts with other tools like cloud code, it's like orders of magnitude less than um..."
- 4:29 / Evidence 2: "security around your PI agent. There's many other ways to do that like using sandboxes using some other guards that you can use. So if you want security you can do that uh through some extensions as well."
- 10:10 / Evidence 3: "that you want. There's another important point here. You can also use local models and it's going to start happening more and more especially now that we have open source models that can run locally and for AI..."
- 11:57 / Evidence 4: "there. So everything that you configure in pi or install is going to be inside of this home.py/ aent folder. All the extensions, skills, prompts, teams, sessions, models or JSON or settings JSON that you have for PI..."
- 18:32 / Evidence 5: "put everything related to tech guidelines, important source files or documentations that you want to bring to the agent like architecture or design decisions that you have. It's also the place where you can instruct the agent on..."
- 20:03 / Evidence 6: "you're going to completely replace the original system prompt of PI. That's really great because it's really flexible. You can also call that through the system prompt argument in PI. And here I'm just like saying pay you..."
- 22:48 / Evidence 7: "basically create them as markdown files inside of PI agent prompts and they have this support for argument and you can use the argument inside of the markdown just like a template. So if we load that in..."

Video-aware target:
- Prompt lane: Context/search
- Mechanism to extract: Extract how context is found, filtered, refreshed, and handed to the agent before it acts.
- Artifact to produce: A context retrieval map with source inventory, indexing/search path, query rules, freshness checks, and agent handoff.
- Artifact must include: source inventory; index/search layer; query rule; freshness check; agent handoff; proof behavior

Your task:
1. Use the transcript anchors above as the primary source packet. If you add outside context, label it clearly as outside context and keep it secondary.
2. Create a source-check table with columns: timestamp, claim, transcript support, what the demo proves, confidence, and what still needs verification.
3. Extract the actual teachable mechanism from the video: Extract how context is found, filtered, refreshed, and handed to the agent before it acts. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A context retrieval map with source inventory, indexing/search path, query rules, freshness checks, and agent handoff.
5. Include:
   - a plain-English definition of the core idea
   - a diagram or structured model using this sequence: Work question -> Source inventory -> Index/search layer -> Retrieval rule -> Agent context -> Answer/proof -> Maintenance
   - answers to these source questions: What source is searched or indexed? | What query/retrieval rule is demonstrated? | How does the agent use the retrieved context?
   - 3 concrete examples that apply the video idea to real agentic work, such as codebase memory; personal wiki retrieval; Elastic search context engineering
   - 2 failure modes the video helps prevent, chosen from the transcript evidence and these likely risks: dumping all context; stale memory; retrieval with no proof trail
   - a checklist for the next real workflow, focused on: sources, query, freshness, handoff, citation/proof
   - one practical exercise with a clear done signal: Write three retrieval queries for one real project and define what each must return.
6. Add a "learning transfer" section: what changes in my workflow tomorrow if I actually learned this?
7. Add a "source check" section that cites which transcript anchor supports each major takeaway.

Quality bar:
- Make this specific to "Pi Agent (Full Course)", not a generic Agent Architecture essay.
- Cite the transcript wherever the prompt names a task boundary, review habit, context move, or verification standard.
- Prefer operational examples, failure modes, and reusable artifacts over broad definitions.
- Call out uncertainty instead of smoothing over weak evidence.
- Avoid these generic drifts: generic context-window advice; memory hype without retrieval rules; source claims without freshness checks.
- If evidence is weak or missing, stop and say what transcript segment or timestamp needs review instead of guessing.
- Finish with a concise artifact I could paste into my learning app.