[claude-hackernews] Reply draft: Agent-skills-eval thread, deny() vs instruct() for routing Bash DB queries via tidewave (id=48046023)#64
Open
NiveditJain wants to merge 1 commit intomainfrom
Conversation
…ueries via tidewave (id=48046023)
📝 WalkthroughWalkthroughA new Markdown draft file outlines a Hacker News "Show HN" submission on an eval harness for Claude code skills, including a substantive reply on tool routing enforcement using "deny" semantics with a concrete policy example that blocks Bash commands matching database access patterns to redirect requests through a dedicated MCP tool. ChangesHN Post Draft on Tool Routing via Deny Semantics
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. 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 current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@drafts/2026-05-08T212431Z.md`:
- Line 19: The fenced code block opened at the start of the snippet is missing a
language tag (triggers markdownlint MD040); update the opening fence used in the
draft (the triple-backtick block that contains "(disclosure: I work on FailProof
AI...)" ) to include an explicit language tag such as text (e.g., change ``` to
```text) so the block is properly tagged and MD040 is resolved.
🪄 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: bcfd33ac-de3c-4136-b7d5-05234b74b80d
📒 Files selected for processing (1)
drafts/2026-05-08T212431Z.md
|
|
||
| **My reply:** | ||
|
|
||
| ``` |
There was a problem hiding this comment.
Add a language tag to the fenced code block.
Line 19 opens a fenced block without a language, which triggers markdownlint MD040.
Suggested fix
-```
+```text
(disclosure: I work on FailProof AI: https://github.com/exospherehost/failproofai)
...
-```
+```🧰 Tools
🪛 markdownlint-cli2 (0.22.1)
[warning] 19-19: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@drafts/2026-05-08T212431Z.md` at line 19, The fenced code block opened at the
start of the snippet is missing a language tag (triggers markdownlint MD040);
update the opening fence used in the draft (the triple-backtick block that
contains "(disclosure: I work on FailProof AI...)" ) to include an explicit
language tag such as text (e.g., change ``` to ```text) so the block is properly
tagged and MD040 is resolved.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Discovery
/ask,/show,hn.algolia.com/?q=claude%20code%20agent,hn.algolia.com/?q=agent%20guardrails,hn.algolia.com/?q=claude%20code%20hooks,hn.algolia.com/?q=codex%20agent(alldateRange=pastWeek/pastMonth,sort=byDate).drafts/,comments/, and open PRs). Tilde.run sandbox thread (id=48037724) skipped under cross-thread duplicate guard since luv-63's "Stop Treating Agent Sandboxes as Cattle" already argues the intent-vs-infra layer.Thread
CLAUDE.mdtelling it to route DB queries through tidewave's MCP server, and instead doesBash(DATABASE_URL=$(grep ... .env) echo "ok"). reedlaw's id=48051949 then asks how a hook would even work for this, and concludes "prompts are not tightly coupled with capabilities".Proposed reply (full text in
drafts/2026-05-08T212431Z.md)The reply pivots on the deny()-vs-instruct() distinction:
instruct(msg)injects guidance and lets the call proceed (so the model can ignore it the same way it ignores CLAUDE.md), butdeny(msg)returns a tool-error so the bash literally does not run and the model has to take a different path. Includes one custom-policy snippet that pattern-matches DB-shaped Bash commands anddenys them with a redirect message to tidewave's MCP tool. ~140 words. ASCII-only punctuation. Single disclosure line at the top, single repo link.Workflow
origin/main(luv-64 has its own PR [claude-hackernews] Reply draft: Kirikiri Show HN, mobile supervision asymmetry vs hook layer (id=47996198) #55 covering id=47996198, so this is not bundled there).drafts/,comments/, or open-PR diff. Cross-thread duplicate guard verified - deny()-vs-instruct() framing is fresh; earlier PRs covered transport-vs-hook, MCP-surface-vs-PreToolUse, Docker-vs-intent, workflow-vs-invariant, etc.Summary by CodeRabbit
Note: This change consists primarily of internal draft materials with no direct user-facing impact.