[claude-hackernews] Reply draft: Harness Show HN, hook layer cuts permission-prompt UX cost at parallel-N (id=47948379)#47
[claude-hackernews] Reply draft: Harness Show HN, hook layer cuts permission-prompt UX cost at parallel-N (id=47948379)#47NiveditJain wants to merge 1 commit intomainfrom
Conversation
…n-prompt UX cost at parallel-N (id=47948379) Top-level reply on the Harness Show HN (parallel Claude Code agents across worktrees). The harness exposes "waiting on permission prompt" as a red worktree-tab status. Reply engages the design choice and points out that an auto-deny PreToolUse hook composes with the prompt surface to reduce the per-prompt context-switch tax that scales linearly with N parallel agents.
📝 WalkthroughWalkthroughAdds a single timestamped Markdown draft for a Show HN post about managing parallel Claude Code agents across Git worktrees. The draft includes an unauthenticated reply with a permission-prompt scaling argument, targeted insights for a team on parallel-agent policy configuration, and detailed notes on thread context and pitch constraints. ChangesDraft Show HN Post
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Review rate limit: 4/5 reviews remaining, refill in 12 minutes. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@drafts/2026-05-04T043235Z.md`:
- Around line 11-15: The fenced code block starting with ``` on the reply block
should include a language tag to satisfy markdownlint MD040; change the opening
fence from ``` to ```text (or another appropriate language) so the block is
fenced as ```text and lint passes, e.g., update the block that begins with the
disclosure "(disclosure: I work on FailProof AI...)" to use ```text as its
opener.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 6130d866-04f5-441a-b118-7c7491aca9d2
📒 Files selected for processing (1)
drafts/2026-05-04T043235Z.md
| ``` | ||
| (disclosure: I work on FailProof AI: https://github.com/exospherehost/failproofai) | ||
|
|
||
| The red-status-equals-permission-prompt design makes visible what the harness layer is paying for in human attention: with N parallel worktrees, every prompt is a context switch, and the cost scales with N. A lot of those prompts are recoverable a layer below: anything you would always click "deny" on can be a PreToolUse hook that returns deny() with a reason, so the agent gets the rejection inline without ever turning a tab red. Composes with what you have - the hook short-circuits the obvious-no calls, the prompt surface handles the genuinely ambiguous ones. Curious whether you considered per-worktree hook config, so an experimental worktree could ship with a stricter policy set than a "trusted" one. | ||
| ``` |
There was a problem hiding this comment.
Add a language tag to the fenced reply block to satisfy markdownlint.
Line 11 opens a fenced code block without a language, which triggers MD040. Add a language label (for example text) to keep lint clean.
Suggested diff
-```
+```text
(disclosure: I work on FailProof AI: https://github.com/exospherehost/failproofai)
The red-status-equals-permission-prompt design makes visible what the harness layer is paying for in human attention: with N parallel worktrees, every prompt is a context switch, and the cost scales with N. A lot of those prompts are recoverable a layer below: anything you would always click "deny" on can be a PreToolUse hook that returns deny() with a reason, so the agent gets the rejection inline without ever turning a tab red. Composes with what you have - the hook short-circuits the obvious-no calls, the prompt surface handles the genuinely ambiguous ones. Curious whether you considered per-worktree hook config, so an experimental worktree could ship with a stricter policy set than a "trusted" one.</details>
<details>
<summary>🧰 Tools</summary>
<details>
<summary>🪛 markdownlint-cli2 (0.22.1)</summary>
[warning] 11-11: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
</details>
</details>
<details>
<summary>🤖 Prompt for AI Agents</summary>
Verify each finding against the current code and only fix it if needed.
In @drafts/2026-05-04T043235Z.md around lines 11 - 15, The fenced code block
starting with on the reply block should include a language tag to satisfy markdownlint MD040; change the opening fence from to text (or another appropriate language) so the block is fenced as text and lint passes, e.g.,
update the block that begins with the disclosure "(disclosure: I work on
FailProof AI...)" to use ```text as its opener.
</details>
<!-- fingerprinting:phantom:triton:hawk:1b55a2c7-5e47-4a32-b5d2-ccdbc8015892 -->
<!-- d98c2f50 -->
<!-- This is an auto-generated comment by CodeRabbit -->
1 participant
Summary
Top-level reply draft for the Harness Show HN by
frenchie4111(id=47948379), a parallel-Claude-Code-sessions-across-worktrees harness. The reply engages the design choice that surfaces "waiting on permission prompt" as a red worktree-tab status, and notes that an auto-deny PreToolUse hook layer composes with that prompt surface to reduce the per-prompt context-switch tax that scales linearly with the number of parallel agents.Discovery path
Browser-only sweep through:
/news,/newest,/ask,/show,/beston news.ycombinator.comhn.algolia.comsearches:agent deleted,force push agent,claude code yolo,cursor deleted,claude code broke,mcp server,guardrails,hooks,agent sandbox,ai agent permissions,agent wiped,claude code running wild,claude code hooksExisting PR coverage filtered out 30+ candidate threads (full ID list is in the local PR title scan). The remaining concrete-failure / Show-HN-of-adjacent-product candidates that were not duplicates:
iClaw(id=47933750, Apple-Intelligence consumer agent — audience overlap with FailProof was weak),Enoch(id=47994468, agentic-research control plane — not a guardrail product),Harness(id=47948379, this thread). Picked Harness because the OP's design choice (per-worktree red-status permission-prompt indicator) is directly comparable to FailProof's PreToolUse layer, OP solicits feedback, and the thread is open with no engagement to crowd out.Thread links
Draft
drafts/2026-05-04T043235Z.md— Status: draft (pending manual post). Body is ~125 words, ASCII-punctuation only, single disclosure line, single PreToolUse hook concept named, no install commands, no feature dump, repo URL appears once (in the disclosure line).Anti-pitch checks
Test plan
drafts/2026-05-04T043235Z.md.[dead]/[flagged].comments/.Summary by CodeRabbit