Fix Module 7 Part C: don't run the new clear/count commands 'while the agents run' — they don't exist yet #12

New Issue

2026-06-22T14:23:42-04:00

claude commented

2026-06-22 14:23:42 -04:00

Problem

Part C step 3 tells the learner to prove isolation by running python cli.py clear and python cli.py count from a third terminal "while [the agents] run" — but those are the very commands the two agents are still authoring. They don't exist until each agent finishes, so running them during the work prints "unknown command" instead of the intended demonstration. Isolation is further undercut because a fresh worktree starts with no tasks.json (gitignored runtime state), so both apps are empty and the "they don't even share data" claim is never actually shown.

Evidence

modules/07-worktrees-running-agents-in-parallel/README.md (~lines 272-281): "Let both work at the same time. While they run, prove the isolation from a third terminal:" followed by python cli.py clear / python cli.py count.
Starter cli.py ships only add/list/done and prints unknown command: {command}; clear/count are added by the agent prompt files. The README itself calls tasks.json gitignored runtime state.

Why it matters

The climactic proof-of-isolation moment can print "unknown command" before the agent writes the command — fragile at the exact payoff, and the data-isolation claim is never demonstrated.

Proposed change

Demonstrate isolation DURING the run with commands that already exist plus distinct data:

In each worktree, add a distinct task (e.g. python cli.py add "from worktree A" vs …B) and python cli.py list, showing each worktree's tasks.json is independent.
Move any run of the new clear/count commands to AFTER the agents finish.
Update the Check-for-understanding line (~line 357) that is keyed to clear-vs-count so it matches the new add/list demonstration.

Acceptance criteria

The during-run isolation demo uses only commands that exist at that moment.
The demo actually shows independent tasks.json per worktree (distinct data).
Check-for-understanding matches the revised demo.

Affected files

modules/07-worktrees-running-agents-in-parallel/README.md

References

Source finding F20 (realVotes 3/3).

Filed from an adversarial multi-agent course review (217 raw findings → 54 adversarially-verified survivors). Scoped for manual review; intentionally not auto-assigned to an agent.

## Problem Part C step 3 tells the learner to prove isolation by running `python cli.py clear` and `python cli.py count` from a third terminal "while [the agents] run" — but those are the very commands the two agents are still authoring. They don't exist until each agent finishes, so running them during the work prints "unknown command" instead of the intended demonstration. Isolation is further undercut because a fresh worktree starts with no `tasks.json` (gitignored runtime state), so both apps are empty and the "they don't even share data" claim is never actually shown. ## Evidence `modules/07-worktrees-running-agents-in-parallel/README.md` (~lines 272-281): "Let both work at the same time. While they run, prove the isolation from a third terminal:" followed by `python cli.py clear` / `python cli.py count`. Starter `cli.py` ships only `add`/`list`/`done` and prints `unknown command: {command}`; clear/count are added by the agent prompt files. The README itself calls `tasks.json` gitignored runtime state. ## Why it matters The climactic proof-of-isolation moment can print "unknown command" before the agent writes the command — fragile at the exact payoff, and the data-isolation claim is never demonstrated. ## Proposed change Demonstrate isolation DURING the run with commands that already exist plus distinct data: 1. In each worktree, add a distinct task (e.g. `python cli.py add "from worktree A"` vs `…B`) and `python cli.py list`, showing each worktree's `tasks.json` is independent. 2. Move any run of the *new* clear/count commands to AFTER the agents finish. 3. Update the Check-for-understanding line (~line 357) that is keyed to clear-vs-count so it matches the new add/list demonstration. ## Acceptance criteria - [ ] The during-run isolation demo uses only commands that exist at that moment. - [ ] The demo actually shows independent `tasks.json` per worktree (distinct data). - [ ] Check-for-understanding matches the revised demo. ## Affected files - `modules/07-worktrees-running-agents-in-parallel/README.md` ## References Source finding F20 (realVotes 3/3). --- *Filed from an adversarial multi-agent course review (217 raw findings → 54 adversarially-verified survivors). Scoped for manual review; intentionally not auto-assigned to an agent.*

claude added the P1 bug ai-ready labels 2026-06-22 14:23:42 -04:00

claude referenced a pull request that will close this issue

2026-06-22 14:52:21 -04:00

Fix broken git demos/labs in Modules 4, 7, 14 (#2,#3,#4,#12) #55

claude referenced this issue from a commit

2026-06-22 14:52:22 -04:00

fix(modules-4,7,14): make the git demos/labs actually do what the prose claims

claude closed this issue

2026-06-22 14:52:31 -04:00

Sign in to join this conversation.