Repository navigation
Warn about install permission prompts, from a real walkthrough run - #29
Merged
Merged
Conversation
docs/first-task.md was run end to end in Claude Code 2.1.316 against the published 3.4.0 plugin. Installing from GitHub, the gated fix with the verifier subagent, and the uninstall all behaved as documented. The one gap: Claude Code asks permission before the skill writes CLAUDE.md, reads the plugin's agent files, and creates agents/, and a refused write leaves the install undone. README and the walkthrough now say so, and the walkthrough summarises what the recorded step 4 session did. Fixes #11 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TgCNfzRULZvxaP1GfyqzRe
ovhirup
marked this pull request as ready for review
October 31, 2026 07:38
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
This was referenced Oct 31, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #11: the last criterion, a run of the walkthrough in a real Claude Code session.
What was run
docs/first-task.mdwas run end to end in Claude Code 2.1.316, in a throwaway home directory and project, against the published 3.4.0 plugin:claude plugin marketplace add ovhirup/stackwich+install stackwich@stackwich/stackwich:stackwichcheck_install.py(scope written with exactly what the installer writes)INSTALL: PASS, policy-rev 7, three Stackwich agentsFAILED (failures=1)total(); theverifiersubagent re-ran the tests, checked scope, and returnedGATE: PASS; the report quoted the command and counted 1 subagent. Independent check: tests pass,test_cart.pyunchangeduninstall.pythenclaude plugin uninstall/marketplace removeThe gap found
Claude Code protects
CLAUDE.mdand files outside the project. In non-bypass modes, the skill needs permission to writeCLAUDE.md, read the plugin's agent files, and createagents/. If any of those is refused, the install does not happen. The skill handled this well: it did not work around the block, it reported each check as FAIL, and it said what to allow. But neither the README nor the walkthrough warned users to expect these prompts.Change
docs/first-task.mdstep 1 now say to expect and approve those prompts, and what happens if one is refused.docs/first-task.mdnotes the recorded run and summarises what the step 4 session did.Verification
skill-lintsteps pass locally.GATE: PASSand scope check, not just the final message.Generated by Claude Code