[claude-hackernews] Reply draft: Kirikiri Show HN, mobile supervision asymmetry vs hook layer (id=47996198)

[NiveditJain]

Summary

• Proposes a top-level reply to Show HN: Kirikiri - A mobile IDE for Claude Code (iOS, open source) (id=47996198, by Husena, ~24h old, 0 comments at draft time).
• Engages with the OP's explicit ask ("Anyone already using Claude Code on mobile - what's broken for you?") on the supervision-asymmetry seam: per-tool-call accept/deny is fine on desktop but degrades on a phone, where ergonomics push users toward yolo-mode or session-wide approve. Tap-buttons solve the input side; they don't solve "I greenlit one Bash and the agent rm-rf'd the wrong path while I was scrolling Twitter." Suggests pushing that one specific gate to a PreToolUse hook so per-call approvals stay opt-in.
• ASCII-only punctuation, ~146 words, single disclosure, single failure mode (rm -rf) tied to a single hook concept. No install commands, no comma-listed policy names, no feature catalog or scope/version talk.

Discovery URL

• Show HN listing: https://news.ycombinator.com/show
• Specific thread (item permalink): https://news.ycombinator.com/item?id=47996198

Discovery path

/ask -> /show -> /newest -> hn.algolia.com searches for: "agent deleted" (24h), "claude code hooks" (week), "agent production deleted" (week), "claude code agent" (24h), "agentic coding" (24h), "ai agent safety" (24h), "guardrail agent" (week), "ai agent broke" (week), "Show HN agent" (week), "Show HN claude code" (week), "Tell HN" (week), "claude permissions" (week), "destroyed" (week). Cross-checked the 44 open-PR thread IDs in this repo and 1 logged comment ID before settling on this thread; many adjacent Show HN's were already covered (Smithy AI, AgentPort, BetterClaw, Trent, Snyk agent-scan, etc.).

Duplicate check (three-surface scan)

• `drafts/`, `comments/` on this branch: no hit for id=47996198.
• Open PRs on this repo: `gh pr diff` over all open PRs returned no match for `item?id=47996198`.
• Cross-thread body-fingerprint guard: the angle (mobile/SSH supervision asymmetry) is new for this repo. Prior FailProof drafts engage at the static-vs-runtime seam (Snyk, Smithery, Trent, Spec27, TrainForgeTester) or the workflow-shape-vs-invariant-shape seam (BetterClaw); none address phone-shell ergonomics.

Test plan

• Reviewer confirms the reply text is ASCII-only (no em/en dashes, no curly quotes, no Unicode arrows, no fancy ellipsis).
• Reviewer confirms thread (id=47996198) is still accepting replies (composer textarea rendered) before posting.
• Reviewer confirms the disclosure line lists the right repo URL: https://github.com/exospherehost/failproofai
• Reviewer confirms one failure mode named (rm -rf) and one hook concept (PreToolUse hard deny); no install commands or comma-listed policy names.
• After posting on HN, append the comment permalink to the HN: line in `drafts/2026-05-04T122844Z.md` (or move to `comments/` if that's the convention you prefer for posted replies). Do NOT proactively edit before the user asks.

Summary by CodeRabbit

• Chores
  • Added internal draft documentation containing notes and insights on mobile development considerations.

[coderabbitai]

📝 Walkthrough

Walkthrough

A new reply draft document for a Show HN discussion about Kirikiri (a mobile IDE) is added to the drafts directory. The draft proposes a PreToolUse hook with hard deny rules for risky commands as a solution to supervision asymmetry on mobile platforms, along with an insight suggestion for the FailProof team and contextual notes about engagement and positioning.

Changes

Show HN Reply Draft on Mobile Supervision

Layer / File(s)	Summary
Draft Content & Core Proposal drafts/2026-05-04T122844Z.md (lines 1–46)	Reply structure with metadata (status, HN link, OP summary), verbatim reply body describing supervision asymmetry in mobile (per-tool-call approvals collapsing into "yolo mode"), and a proposed PreToolUse hook strategy with hard deny for rm -rf-matching paths.
Team Insights & Findings drafts/2026-05-04T122844Z.md (lines 47–55)	"Insight for the FailProof team" section proposing a docs/blog note on hook-layer policy ergonomics for headless/mobile supervision. Supplementary notes covering thread traction, voice constraints, failure-mode specificity, and discovery context.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

• [claude-hackernews] Restore drafts/ vs comments/ split: route writes to drafts/ #6: Routes agent writes into drafts/<timestamp>.md structure; directly aligned with the drafts workflow this PR extends.
• [claude-hackernews] Reconcile drafts/ vs comments/ split; add cron no-op alert #4: Reconciles the drafts/ vs. comments/ filesystem workflow and .gitignore behavior; this PR adds a file within that established drafts pattern.
• [claude-hackernews] Switch HN automation to drafts-only mode #2: Introduced the drafts-only workflow and tracking of the drafts/ directory; this PR creates a timestamped draft file within that system.

Poem

🐰 A draft takes shape in drafts/ so bright,
PreToolUse hooks guard rm -rf with might,
On mobile screens, asymmetry fades away,
The FailProof team learns what hooks convey—
Wisdom hops forward, one reply at a time! 🌙✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title directly and specifically summarizes the main change: adding a reply draft for a Kirikiri Show HN post focusing on mobile supervision asymmetry and hook-layer solutions, with the HN item ID included.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ 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 help to get the list of available commands and usage tips.}

[coderabbitai]

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/2026-05-04T122844Z.md`:
- Line 35: The fenced code block containing "(disclosure: I work on FailProof
AI: https://github.com/exospherehost/failproofai)" is missing a language tag;
update the opening triple-backtick for that block to include a language
identifier (e.g., change ``` to ```text) so markdownlint MD040 is satisfied and
the CI/docs lint stays clean.

🪄 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: b5e4635a-16df-41ae-9759-4b9bcd0a317d

📥 Commits

Reviewing files that changed from the base of the PR and between ebbce06 and cb3bf38.

📒 Files selected for processing (1)

• drafts/2026-05-04T122844Z.md

[coderabbitai]

⚠️ Potential issue | 🟡 Minor | ⚡ Quick win

Add a language tag to the fenced block (MD040)

Line 35 should declare a code fence language to satisfy markdownlint and keep CI/docs lint clean.

Suggested fix

-```
+```text
 (disclosure: I work on FailProof AI: https://github.com/exospherehost/failproofai)
@@
-```
+```

📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change

```

🧰 Tools

🪛 markdownlint-cli2 (0.22.1)

[warning] 35-35: Fenced code blocks should have a language specified

(MD040, fenced-code-language)

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@drafts/2026-05-04T122844Z.md` at line 35, The fenced code block containing
"(disclosure: I work on FailProof AI:
https://github.com/exospherehost/failproofai)" is missing a language tag; update
the opening triple-backtick for that block to include a language identifier
(e.g., change ``` to ```text) so markdownlint MD040 is satisfied and the CI/docs
lint stays clean.