Clarify that Module 25 --simulate uses a stand-in demo, not the delete-command issue it's handed #27

New Issue

2026-06-22T14:23:50-04:00

claude commented

2026-06-22 14:23:50 -04:00

Problem

The module's thesis is "a well-formed issue + committed config is a complete spec the agent implements," and the README narrates Parts A/B as the agent implementing issue-delete-command.md ("Open the diff and review it with the Module 10 checklist"). But --simulate good/bad ignores the issue and writes agent_demo.py with an unrelated discount(price, pct) function plus its test — zero connection to deleting tasks. The commit message even claims "implement issue-delete-command" while the diff is the discount demo. The determinism rationale lives only in the script docstring, never surfaced to the learner.

Evidence

README Part B (~lines 256-258): "This time the planted change is correct … Open the diff and review it with the Module 10 checklist."
modules/25-autonomous-agents/lab/agent_runner.py simulate_implement() (~lines 128-138) writes def discount(price, pct): return price - price * pct / 100 — unrelated to issue-delete-command.md.

Why it matters

The mismatch undercuts the module's central exercise (reviewing the agent's diff against the issue) at the exact moment it's taught — against the honesty/no-hand-waving promise. A diligent learner who read the issue is badly confused.

Proposed change

Doc-only clarification (do not make the simulator non-deterministically mutate real files):

State plainly before the first command that --simulate does NOT implement the issue — it uses a self-contained discount() stand-in to stay deterministic and avoid touching real files; the issue is really implemented only in Part D with a live agent.
Adjust Part B's "review the diff" wording so the learner expects the demo function.
Optional: rename the stand-in to a delete/remove-themed demo so its subject matches the issue.

Acceptance criteria

The README tells the learner, before Part A, that --simulate is a deterministic stand-in unrelated to the issue.
Part B's diff-review wording matches what --simulate actually produces.

Affected files

modules/25-autonomous-agents/README.md (and optionally lab/agent_runner.py for the rename)

References

Source finding F41 (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 The module's thesis is "a well-formed issue + committed config is a complete spec the agent implements," and the README narrates Parts A/B as the agent implementing `issue-delete-command.md` ("Open the diff and review it with the Module 10 checklist"). But `--simulate good/bad` ignores the issue and writes `agent_demo.py` with an unrelated `discount(price, pct)` function plus its test — zero connection to deleting tasks. The commit message even claims "implement issue-delete-command" while the diff is the discount demo. The determinism rationale lives only in the script docstring, never surfaced to the learner. ## Evidence README Part B (~lines 256-258): "This time the planted change is correct … Open the diff and review it with the Module 10 checklist." `modules/25-autonomous-agents/lab/agent_runner.py` `simulate_implement()` (~lines 128-138) writes `def discount(price, pct): return price - price * pct / 100` — unrelated to `issue-delete-command.md`. ## Why it matters The mismatch undercuts the module's central exercise (reviewing the agent's diff against the issue) at the exact moment it's taught — against the honesty/no-hand-waving promise. A diligent learner who read the issue is badly confused. ## Proposed change Doc-only clarification (do not make the simulator non-deterministically mutate real files): 1. State plainly before the first command that `--simulate` does NOT implement the issue — it uses a self-contained `discount()` stand-in to stay deterministic and avoid touching real files; the issue is really implemented only in Part D with a live agent. 2. Adjust Part B's "review the diff" wording so the learner expects the demo function. 3. Optional: rename the stand-in to a delete/remove-themed demo so its subject matches the issue. ## Acceptance criteria - [ ] The README tells the learner, before Part A, that `--simulate` is a deterministic stand-in unrelated to the issue. - [ ] Part B's diff-review wording matches what `--simulate` actually produces. ## Affected files - `modules/25-autonomous-agents/README.md` (and optionally `lab/agent_runner.py` for the rename) ## References Source finding F41 (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 ai-ready bug P1 labels 2026-06-22 14:23:50 -04:00

claude referenced a pull request that will close this issue

2026-06-22 14:37:10 -04:00

Fix Module 25 command injection + lab integrity (#24–#27) #54

claude referenced this issue from a commit

2026-06-22 14:37:12 -04:00

fix(module-25): close command-injection + lab-integrity issues

claude closed this issue

2026-06-22 14:37:20 -04:00

Sign in to join this conversation.