Repository navigation
[training-plan] Add research-backed training node: Control What Your Agent Can Reach on the Network - #3536
Closed
github-actions[bot] wants to merge 1 commit into
Closed
[training-plan] Add research-backed training node: Control What Your Agent Can Reach on the Network#3536github-actions[bot] wants to merge 1 commit into
github-actions[bot] wants to merge 1 commit into
Conversation
…on the Network Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Contributor
Author
|
This pull request was automatically closed because it expired on 2026-09-19T16:07:49.782Z.
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Caution
Protected files were modified in this change.
This pull request is in
request_reviewmode and requires explicit human scrutiny before merge.Protected files:
README.mdResearch focus
network:frontmatter egress control and the Agent Workflow Firewall (AWF) — deny-by-default posture, ecosystem identifiers vs. invalid bare shorthands, wildcard domain rules, and verifying results viagh aw audit --parse/firewall.md.Gap identified
Steps 16–17 teach learners to fetch live data and add MCP tools, and several side quests reference
network.allowedin passing for security framing, but no core step teaches thenetwork:field itself as a hands-on configuration surface. Learners adding MCP tools or external API calls had no guided path to correctly scope egress, risking over-allowing (defeating the security model already introduced) or hitting confusing compile-time errors from invalid ecosystem shorthands (e.g.npminstead ofnode).Files updated
workshop/17b-network-egress-control.md— new node teachingnetwork:configuration, ecosystem identifiers, wildcard domains, and firewall log verification.workshop/README.md— added curriculum table row for the new step between Step 17 and Step 18.workshop/17-add-mcp-tools.md— updated the "Next" link to point to the new node instead of skipping straight to Step 18.Research metadata location
An XML
<research-metadata>comment block is embedded at the end ofworkshop/17b-network-egress-control.md, recording the selected focus, top source URLs (including theLLMs.txtdownload source), and a one-paragraph reasoning summary.Warning
Firewall blocked 1 domain
The following domain was blocked by the firewall during workflow execution:
registry.npmjs.orgTo allow these domains, add them to the
network.allowedlist in your workflow frontmatter:See Network Configuration for more information.