ccctx-review

name	ccctx-review
description	Review ccctx changes for correctness, regressions, security, tests, documentation, and release readiness. Use when asked to review pull requests, diffs, commits, branches, or proposed changes in this repository, especially changes touching Go CLI behavior, profile storage, Claude Code credential handling, exec isolation, docs, CI, or project skills.

ccctx Review

Review as a maintainer of this repository. Prioritize findings over summary.

Context To Read

• README.md
• CONTRIBUTING.md
• docs/architecture.md
• docs/development.md
• llms.txt
• internal/ccctx/app.go
• internal/ccctx/app_test.go
• files changed in the diff

Review Workflow

1. Inspect the diff first with git diff or the provided patch.
2. Read surrounding implementation before deciding whether a change is wrong.
3. Run targeted tests when possible, then go test ./... and go vet ./... for broader changes.
4. If skills or Claude plugin files changed, run claude plugin validate . --strict when the claude CLI is available.
5. Report findings first, ordered by severity, with file and line references.
6. Include test gaps or residual risk after findings.

Invariants

• Saved Claude config, credential, and pi auth snapshots are sensitive. Do not print tokens or full credential payloads.
• Profile and credential files must use owner-only permissions.
• Switching pi auth must preserve non-Anthropic provider entries.
• Switching sessions must persist the previously active profile before replacing the active Claude config.
• Stale current markers must not override the identity detected from the live Claude config.
• exec mode must not overwrite the user's real ~/.claude.json.
• exec mode must persist refreshed Claude Code and pi Anthropic auth state back into the selected saved profile.
• Non-macOS credential behavior must remain safe with the no-op store.
• CLI output and command aliases are user-facing compatibility surface.

Output Shape

Use this order:

1. Findings with severity and path:line references.
2. Open questions or assumptions.
3. Verification performed.
4. Brief summary only after the findings.

If there are no findings, say so clearly and name any untested risk.