[claude-hackernews] Reply draft: BetterClaw Show HN, graph vs policy-function (id=47973502)#13
[claude-hackernews] Reply draft: BetterClaw Show HN, graph vs policy-function (id=47973502)#13NiveditJain wants to merge 1 commit into
Conversation
…icy-function (id=47973502) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughA new markdown draft document is added capturing the PocketOS incident context, the BetterClaw proposal for workflow-to-directed-graph compilation with tool-gating, and comparative analysis with FailProof's policy approach, along with actionable recommendations for the FailProof team. Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~3 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. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Review rate limit: 0/1 reviews remaining, refill in 60 minutes.Comment |
Summary
Draft reply on a Show HN of an adjacent agent-tool-gating product (BetterClaw, id=47973502). OP explicitly solicits design feedback on three things: paragraph -> graph vs YAML / DSL authoring, where the model falls down (multi-step approvals, loops, sub-agents), and which other agent runtimes to support. The thread also names the April 25 PocketOS incident (a Cursor agent running Claude wiped a prod DB in 9 seconds via a Railway MCP server) as motivation.
The reply discloses FailProof affiliation in the leading line, engages substantively with the reproducer + the authoring-model question + the falls-down question, and contrasts BetterClaw's "compile a paragraph into a workflow graph" model with FailProof's per-intent allow/deny/instruct policy functions. Ends with a direct question back to OP about their authoring-model choice.
ASCII-only punctuation, ~155-word body, single repo URL in the disclosure line, no install commands or policy comma-lists. Status: draft, pending manual post.
Discovery
item?id=47973502hit indrafts/,comments/, or any open PR diff at draft time.Thread
Test plan
drafts/2026-05-01T202400Z.mdinto the comment composer; submit.comments/and re-commit.Summary by CodeRabbit