This video explains how to decide which of Pi's four config folders (agents, skills/prompts, extensions) a given workflow belongs in, using context cost as the deciding factor.
Eric MichaudWatchTranscript found
Quick learning frame
Read this before watching.
A model becomes useful when it is wrapped in a harness: tools, state, permissions, memory, routing, and verification.
New playlist item from Eric Michaud; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Skill you build: Choosing the right Pi customization primitive for a task so you minimize main-agent context bloat while keeping repeatable workflows fast.
Watch for the shift from claim to mechanism. The learning value is the point where the transcript reveals a repeatable action, tool boundary, context move, review habit, or artifact.
Concept diagram
Where this video fits.
01User intent
02Model role
03Tool surface
04State and memory
05Verification loop
06Reusable operating rule
Deep lesson
Turn this video into working knowledge.
1,745 cleaned transcript words reviewed across 496 timed caption segments.
Thesis
PI Agents vs Skills vs Extensions!? teaches a practical agent harness move: This video explains how to decide which of Pi's four config folders (agents, skills/prompts, extensions) a given workflow belongs in, using context cost as the deciding factor.
The goal is not to remember the video. The goal is to extract the operating principle, tie it to timestamped evidence, test how far the claim transfers, and make something reusable.
0:10
Customization, not theming
“between what should be an agent, skill, prompt, and what should be an extension. Yesterday, I went through subagents because my Pi agent was getting really heavy, and I didn't want all of my agent system prompting and...”
The real win in tailoring an agent harness is structuring agents, skills, prompts, and extensions correctly, not adjusting layout or theme; the goal is keeping only the context each task actually needs. Open your own .pi config and list which workflows currently live as raw rules in the main agent that could be moved out.
3:46
Extensions remove prompting
“confusing in terms of like prompts, skills, agents, all that sort of thing. And it all comes comes to maybe like context, because extensions are bits of TypeScript, so it's like code. And context is really the part...”
Sub-agents run in parallel to offload context, but extensions are TypeScript tools whose implementation is never passed into the prompt, so they eliminate the prompting entirely until called (e.g. Graphify querying a graph cache). Identify one mechanical, repeated task you keep re-explaining in English and sketch it as an extension instead of a prompt.
5:22
Think vs do test
“bit more experimenting. Some of my current workflows could probably remain as skills, or some could be turned into sub agents. Some could definitely be extensions. So I'm going to go through each of these, and if it's...”
Map workflows by context: skills/prompts tell the model how to think (loaded only when called by name+description), extensions are the machinery/tools, and sub-agents are specialists; if it's a tool manual or mostly 'what to do', make it an extension. Go through each of your current skills and tag each as keep-as-skill, convert-to-extension, or convert-to-sub-agent using the 'how to think vs what to do' rule.
01
User intent
Start with this video's job: This video explains how to decide which of Pi's four config folders (agents, skills/prompts, extensions) a given workflow belongs in, using context cost as the deciding factor. Treat "User intent" as the outcome you are trying to make visible, not a topic label. Anchor it to 0:10, where the video says: “between what should be an agent, skill, prompt, and what should be an extension. Yesterday, I went through subagents because my Pi agent was getting really heavy, and I didn't want all of my agent system prompting and...”
02
Model role
Use "Model role" to locate the part of the agent harness mechanism the video is demonstrating. Ask what changes in your real setup if this claim is true. Anchor it to 3:46, where the video says: “confusing in terms of like prompts, skills, agents, all that sort of thing. And it all comes comes to maybe like context, because extensions are bits of TypeScript, so it's like code. And context is really the part...”
03
Tool surface
Turn "Tool surface" into the reusable artifact for this lesson: A one-page agent harness map with tool boundaries, state ownership, and proof signals. This is where watching becomes something you can inspect and reuse.
04
State and memory
Use "State and memory" as the application surface. Decide whether the idea touches a browser flow, a local file, a model choice, a source document, a UI, or a review step.
05
Verification loop
Use "Verification loop" to prove the lesson. The evidence should connect back to the video title, transcript anchors, and a concrete output, not a generic best-practice claim.
06
Reusable operating rule
Use "Reusable operating rule" to carry the idea forward: save the prompt, checklist, diagram, or operating rule that would make the next agent run better.
Example
Source-backed artifact packet
Convert the video into a scoped artifact request that includes the transcript claim, mechanism, acceptance criteria, and proof. The output should be a one-page agent harness map with tool boundaries, state ownership, and proof signals..
Example
Agent harness proof brief
Separate what the speaker claims, what the demo actually proves, and what still needs outside verification before you adopt the agent harness pattern.
Example
Teach-back module
Transform the lesson into a definition, a User intent -> Model role -> Tool surface -> State and memory -> Verification loop -> Reusable operating rule diagram, one misconception, one practice exercise, and a check-for-understanding question.
Do not learn it wrong
Treating the title as the lesson without checking what the transcript actually says.
treating model choice as architecture
ignoring tool permissions
missing verification evidence
Letting the lesson drift into generic agent definitions.
Letting the lesson drift into model leaderboard claims.
Letting the lesson drift into tool list without operating boundaries.
Do not count this as learned until these are true.
01
State the transcript-backed claim in your own words: This video explains how to decide which of Pi's four config folders (agents, skills/prompts, extensions) a given workflow belongs in, using context cost as the deciding factor.
02
Explain the practical stakes without hype: New playlist item from Eric Michaud; queued for transcript-backed review, topic mapping, and a practical learning artifact.
03
Map the idea onto the User intent -> Model role -> Tool surface -> State and memory -> Verification loop -> Reusable operating rule sequence and name the weakest link.
04
Produce the artifact and include the evidence that proves it: A one-page agent harness map with tool boundaries, state ownership, and proof signals.
Put it into practice
Give this grounded prompt to Codex or Claude after watching.
You are helping me turn one specific YouTube video into real, durable learning.
Source video:
- Title: PI Agents vs Skills vs Extensions!?
- URL: https://www.youtube.com/watch?v=rikKlsZc5PQ
- Topic: AI Strategy
- My current learning frame: Audit your own Pi setup and reclassify three bloated main-agent rules into a skill, a sub-agent, and an extension using the video's context-cost mapping, then run /reload to test the extension live.
- Why this matters: New playlist item from Eric Michaud; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Transcript anchors from this exact video:
- 0:10 / Evidence 1: "between what should be an agent, skill, prompt, and what should be an extension. Yesterday, I went through subagents because my Pi agent was getting really heavy, and I didn't want all of my agent system prompting and..."
- 1:41 / Evidence 2: "harness, just like Claude Code and Codex, right? Except it comes very, very minimal out of the box. Whereas Claude Code has all these features and extensions and tools and stuff like that that they add, all of..."
- 3:46 / Evidence 3: "confusing in terms of like prompts, skills, agents, all that sort of thing. And it all comes comes to maybe like context, because extensions are bits of TypeScript, so it's like code. And context is really the part..."
- 5:22 / Evidence 4: "bit more experimenting. Some of my current workflows could probably remain as skills, or some could be turned into sub agents. Some could definitely be extensions. So I'm going to go through each of these, and if it's..."
- 7:23 / Evidence 5: "starter vault just like the environment you see here. This exact setup you can get through the premium membership on my school community. Just go to tools and agents along with a bunch of other different like apps..."
Video-aware target:
- Prompt lane: Agent harness
- Mechanism to extract: Identify what surrounding harness makes the model more useful than chat alone.
- Artifact to produce: A one-page agent harness map with tool boundaries, state ownership, and proof signals.
- Artifact must include: model role; tools; state/memory; permission boundary; verification proof
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: Identify what surrounding harness makes the model more useful than chat alone. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A one-page agent harness map with tool boundaries, state ownership, and proof signals.
5. Include:
- a plain-English definition of the core idea
- a diagram or structured model using this sequence: User intent -> Model role -> Tool surface -> State and memory -> Verification loop -> Reusable operating rule
- answers to these source questions: What does the video claim the agent can do? | What surrounding system makes that claim plausible? | What proof is shown instead of merely asserted?
- 3 concrete examples that apply the video idea to real agentic work, such as a repo-editing harness; a local research assistant; a recurring refresh agent
- 2 failure modes the video helps prevent, chosen from the transcript evidence and these likely risks: treating model choice as architecture; ignoring tool permissions; missing verification evidence
- a checklist for the next real workflow, focused on: tool boundaries, state ownership, done signal, recovery path
- one practical exercise with a clear done signal: Map one current coding workflow as a harness and mark the first missing proof signal.
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 Agents vs Skills vs Extensions!?", not a generic AI Strategy essay.
- Tie each harness element to a transcript anchor that names a tool, state boundary, permission, model behavior, or verification step.
- 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 agent definitions; model leaderboard claims; tool list without operating boundaries.
- 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.
Misconceptions
What to stop believing.
Every new AI tool deserves a trial.
Every tool has integration cost. Start from workflow pain, not novelty.
If an agent can do it once, it is automated.
Automation means repeatable, monitored, recoverable, and reviewable.
Practice studio
Learning only counts when you make something.
01
Transcript evidence map
Separate what the video actually says from what you already believe about the topic.
3 source-backed takeaways with timestamps, confidence, and a transfer note.02
One useful artifact
Apply the video to a real workflow and produce a one-page agent harness map with tool boundaries, state ownership, and proof signals..
A reusable artifact with a done signal and one verification step.03
Agent harness teach-back card
Explain the agent harness mechanism to someone who has not watched the video yet.
A 90-second explanation, one diagram, one example, and one misconception to avoid.
Recall check
Answer first, then reveal — without rewatching.
What does the creator mean when he says customizing Pi is 'not theming and layout,' and what is the actual goal of the customization?
How does an extension differ from a skill in terms of what gets passed into the model's prompt, using Graphify as the example?
What test does the creator use to decide whether a workflow should stay a skill or become an extension?
Source shelf
Use the video as a doorway, then verify with primary sources.
