[codex] Clarify CodeRabbit auth recovery flow#2
Conversation
📝 WalkthroughWalkthroughUpdates the CodeRabbit plugin configuration and documentation: shifts authentication from install-time to use-time, bumps the plugin version to 1.1.1, reduces default prompt options from three to one, and enhances authentication and error-handling instructions in the skill documentation. Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~5 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 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. Comment |
- Require explicit failure reporting on any CLI error (auth, missing CLI, network, timeout) instead of falling back to a manual review - Use "issues" in the user-facing summary for clearer phrasing - Trim defaultPrompt to the single primary prompt and bump to 1.1.1 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
🧹 Nitpick comments (3)
plugins/coderabbit/skills/coderabbit-review/SKILL.md (2)
59-59: Consider clarifying "retry once network is available".Reads as if the agent should autonomously keep retrying. Since the surrounding guidance is to report failure to the user rather than silently recover (except for auth), suggest rephrasing to make it clear this is user-facing resolution guidance, not an instruction for the agent to loop.
✏️ Suggested wording tweak
-- If an `error` event is returned, or the CLI fails for any other reason (auth failure, missing CLI, network error, timeout), do not fall back to a manual review. Report the exact failure and tell the user how to resolve it (e.g. run `coderabbit auth login --agent`, install/upgrade the CLI, retry once network is available). +- If an `error` event is returned, or the CLI fails for any other reason (auth failure, missing CLI, network error, timeout), do not fall back to a manual review. Report the exact failure to the user along with the concrete resolution step (e.g. run `coderabbit auth login --agent`, install/upgrade the CLI, or retry after the network is restored).Also note the auth-failure case here partially overlaps with the Prerequisites recovery flow (which tells the agent to auto-run
coderabbit auth login --agent). Worth reconciling so it's unambiguous whether a mid-review auth failure should be auto-recovered or surfaced to the user.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@plugins/coderabbit/skills/coderabbit-review/SKILL.md` at line 59, Update the sentence in SKILL.md that currently reads "If an `error` event is returned, or the CLI fails for any other reason (auth failure, missing CLI, network error, timeout), do not fall back to a manual review. Report the exact failure and tell the user how to resolve it (e.g. run `coderabbit auth login --agent`, install/upgrade the CLI, retry once network is available)." to clarify that "retry once network is available" is user-facing guidance (e.g., "retry when network connectivity is restored") and not an instruction for the agent to loop; also reconcile the auth-failure clause with the Prerequisites recovery flow by explicitly stating whether the agent should attempt `coderabbit auth login --agent` automatically on mid-review auth failures or instead surface the auth error to the user (make the behavior unambiguous and consistent with the Prerequisites section).
67-71: Terminology switch to "issues" is consistent here, but two upstream strings still say "findings".The Result Format section is internally consistent, but for end-to-end consistency with the PR's rename:
- Line 8: "summarize the findings" — user-facing skill intro.
- Line 23 in
plugin.json:longDescriptionmentions "actionable findings".Line 57 ("Collect
findingevents") should stay as-is sincefindingis the NDJSON event type emitted by the CLI, not user-facing copy.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@plugins/coderabbit/skills/coderabbit-review/SKILL.md` around lines 67 - 71, Update the remaining user-facing occurrences of "findings" to "issues": change the phrase "summarize the findings" in SKILL.md to "summarize the issues" and update the plugin.json longDescription value that currently mentions "actionable findings" to use "actionable issues" (leave the NDJSON event reference "finding" on the "Collect `finding` events" line unchanged). Ensure wording and punctuation match existing style and that only user-facing copy is altered.plugins/coderabbit/.codex-plugin/plugin.json (1)
3-3: Version bump and defaultPrompt trim look good.1.1.1 appropriately reflects a docs/config-only change, and narrowing
defaultPromptto the primary prompt matches the PR intent.Nit:
interface.longDescriptionon line 23 still says "actionable findings". Since the PR renames user-facing "findings" → "issues" in SKILL.md, consider updating this string for consistency.Also applies to: 36-38
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@plugins/coderabbit/.codex-plugin/plugin.json` at line 3, Update the user-facing wording in the JSON interface strings: replace occurrences of "actionable findings" with "actionable issues" in the interface.longDescription property and the other similar description strings (the other entries around the same block, e.g., the secondary description lines). Locate the JSON keys named interface.longDescription (and the adjacent description fields in the same object) and edit their text to use "issues" instead of "findings" so the plugin wording matches the SKILL.md rename.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@plugins/coderabbit/.codex-plugin/plugin.json`:
- Line 3: Update the user-facing wording in the JSON interface strings: replace
occurrences of "actionable findings" with "actionable issues" in the
interface.longDescription property and the other similar description strings
(the other entries around the same block, e.g., the secondary description
lines). Locate the JSON keys named interface.longDescription (and the adjacent
description fields in the same object) and edit their text to use "issues"
instead of "findings" so the plugin wording matches the SKILL.md rename.
In `@plugins/coderabbit/skills/coderabbit-review/SKILL.md`:
- Line 59: Update the sentence in SKILL.md that currently reads "If an `error`
event is returned, or the CLI fails for any other reason (auth failure, missing
CLI, network error, timeout), do not fall back to a manual review. Report the
exact failure and tell the user how to resolve it (e.g. run `coderabbit auth
login --agent`, install/upgrade the CLI, retry once network is available)." to
clarify that "retry once network is available" is user-facing guidance (e.g.,
"retry when network connectivity is restored") and not an instruction for the
agent to loop; also reconcile the auth-failure clause with the Prerequisites
recovery flow by explicitly stating whether the agent should attempt `coderabbit
auth login --agent` automatically on mid-review auth failures or instead surface
the auth error to the user (make the behavior unambiguous and consistent with
the Prerequisites section).
- Around line 67-71: Update the remaining user-facing occurrences of "findings"
to "issues": change the phrase "summarize the findings" in SKILL.md to
"summarize the issues" and update the plugin.json longDescription value that
currently mentions "actionable findings" to use "actionable issues" (leave the
NDJSON event reference "finding" on the "Collect `finding` events" line
unchanged). Ensure wording and punctuation match existing style and that only
user-facing copy is altered.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository: coderabbitai/coderabbit/.coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: f1436d6c-e4f8-406e-ad90-dc121fdd5104
📒 Files selected for processing (3)
.agents/plugins/marketplace.jsonplugins/coderabbit/.codex-plugin/plugin.jsonplugins/coderabbit/skills/coderabbit-review/SKILL.md
|
@coderabbitai based on everything you know about codex plugins (and coding agent plugins/skills), what are all the wording tweaks you'd suggest for this PR? |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chainLet me read the actual file contents to give you precise, line-level suggestions. 🏁 Script executed: Repository: coderabbitai/codex-plugin
|
| # | File | Location | Current | Suggested |
|---|---|---|---|---|
| 1 | SKILL.md | Opening sentence | "summarize the findings" | "summarize the issues found" |
| 2 | SKILL.md | Frontmatter description | "wants autonomous fix-review cycles" | "requests fix-review cycles" |
| 3 | SKILL.md | Silence rule | "scope selection" | "diff scoping" |
| 4 | SKILL.md | Silence rule | "prerequisite action is required" | "prerequisite is needed" |
| 5 | SKILL.md | Prerequisites #1 |
"git worktree" | "git repository" |
| 6 | SKILL.md | Prerequisites #3 |
"Check auth" | "Verify authentication" |
| 7 | SKILL.md | Auth recovery | "Run the login flow yourself" | "Initiate the login flow" |
| 8 | SKILL.md | Review Commands | "pass the files that exist with -c" | "pass them with -c" |
| 9 | SKILL.md | Output Handling | "intermediary" | "intermediate" |
| 10 | SKILL.md | Output Handling | "even if output is quiet" | "even if no output is produced" |
| 11 | SKILL.md | Output Handling | "wait budget is exhausted" | "10-minute window has elapsed" |
| 12 | SKILL.md | Result Format | "severity/category label" | "severity label" |
| 13 | SKILL.md | Result Format | "concrete fix direction" | "concrete suggested fix" |
| 14 | plugin.json | shortDescription | leading space before "Run" | remove leading space |
| 15 | plugin.json | longDescription | "actionable findings" | "actionable issues" |
|
@coderabbitai great finds. go though the whole coderabbit plugin with a holistic view of all one could use the current coderabbit cli for (ALL possible intents in ALL contexts) and suggest additions here. |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chainLet me read all plugin files and check the latest CodeRabbit CLI documentation simultaneously. 🏁 Script executed: Repository: coderabbitai/codex-plugin 🌐 Web query: 💡 Result: Official “all commands” documentation for the CodeRabbit CLI is the CLI Command Reference page. It lists the full command set, review modes, and flags. [1] Commands (as documented)
If you meant “all PR commands” (the Sources: CLI Command Reference [1], CodeRabbit changelog (CLI v0.3.4 auto-update) [2], PR review commands reference [3].
🆕 Missing skills (new files to create)1.
|
| # | Type | Gap | Addition |
|---|---|---|---|
| 1 | New skill | PR comment auto-fix loop | coderabbit-fix skill |
| 2 | New skill | Config generation/editing | coderabbit-configure skill |
| 3 | Review commands | --prompt-only mode undocumented |
Add with when-to-use guidance vs --agent |
| 4 | Review commands | --dir <path> flag missing |
Add to Review Commands section |
| 5 | Auth recovery | API key path missing | Add --api-key fallback in Prerequisites |
| 6 | Auth | logout + org intents not handled |
Add to Prerequisites or new auth section |
| 7 | CLI maintenance | coderabbit update not covered |
Add to Prerequisites |
| 8 | Output Handling | Rate limit errors not handled | Add rate-limit event handling + retry guidance |
| 9 | Frontmatter | Narrow trigger phrases | Expand description with common phrasings |
| 10 | Output Handling | --interactive mode never clarified |
Add "do not invoke autonomously" guardrail |
Summary
Clarifies the CodeRabbit review skill's authentication recovery instructions.
What Changed
Updates the
coderabbit-reviewskill so the agent is explicitly told to:coderabbit auth status --agentresult as a step to handlecoderabbit auth login --agentcoderabbit auth status --agentWhy
The prior wording only said to run the login command if auth was missing. This left too much room for the agent to stop at the auth error instead of actively recovering and retrying.
Impact
Agents using the CodeRabbit review skill should handle missing authentication more reliably and continue the workflow without unnecessary user interruption.
Validation
Verified the repo diff and committed only the
plugins/coderabbit/skills/coderabbit-review/SKILL.mdchange.Summary by CodeRabbit
Bug Fixes
Changes