gemini-review-capacity-recovery

Handle Gemini adversarial-review runs that emit repeated 429 capacity errors before either recovering with a real verdict or failing unavailable.

5 stars

Best use case

gemini-review-capacity-recovery is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Handle Gemini adversarial-review runs that emit repeated 429 capacity errors before either recovering with a real verdict or failing unavailable.

Teams using gemini-review-capacity-recovery should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/gemini-review-capacity-recovery/SKILL.md --create-dirs "https://raw.githubusercontent.com/vamseeachanta/workspace-hub/main/.agents/skills/software-development/gemini-review-capacity-recovery/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/gemini-review-capacity-recovery/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How gemini-review-capacity-recovery Compares

Feature / Agentgemini-review-capacity-recoveryStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Handle Gemini adversarial-review runs that emit repeated 429 capacity errors before either recovering with a real verdict or failing unavailable.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

Related Guides

SKILL.md Source

# Gemini Review Capacity Recovery

Use when `gemini exec` is part of a plan/code adversarial review wave and the command prints repeated `429 RESOURCE_EXHAUSTED` / `MODEL_CAPACITY_EXHAUSTED` blocks.

## Why this exists
A Gemini run that looks failed at first glance may still recover and produce a substantive review verdict later in the same command output. If you stop early or mark it unavailable too soon, you can lose valid review evidence.

## Trigger pattern
Typical output starts with one or more retry blocks like:
- `Attempt N failed with status 429`
- `RESOURCE_EXHAUSTED`
- `MODEL_CAPACITY_EXHAUSTED`

This does **not** necessarily mean the run is unusable.

## Required workflow
1. Prefer foreground Gemini review for compact rerun prompts when possible.
2. Let the command complete if the timeout budget allows.
3. Inspect the **tail** of the full captured output for a real review block:
   - `Verdict:`
   - `Retrieval adequacy:`
   - `Key findings:`
   - `Main blockers to fix:`
4. If those appear, treat the run as a successful substantive review even if multiple 429 retry blocks occurred first.
5. Save a normal canonical artifact under `scripts/review/results/...-gemini.md`.
6. Only mark Gemini `UNAVAILABLE` if the command ends without a substantive review after retries.

## Artifact decision rule

### If Gemini recovers and produces a verdict
Save a normal artifact, for example:
- `scripts/review/results/YYYY-MM-DD-plan-<issue>-gemini.md`

Include the actual review findings, not the retry noise.

### If Gemini never produces a verdict
Save an unavailable artifact, for example:
- `scripts/review/results/YYYY-MM-DD-plan-<issue>-gemini.md`

With:
- `Verdict: UNAVAILABLE`
- concrete failure reason
- note that repeated capacity retries did not end in a substantive review
- path to the raw CLI log if captured

## Practical notes
- Do not classify a run from the first screenful of output.
- Gemini foreground runs may look noisy but still succeed.
- Capacity exhaustion is different from tool/config failure; treat them separately.
- If a provider-specific review artifact already exists as `UNAVAILABLE` and a later rerun recovers with a real verdict, replace the placeholder with the substantive artifact.

## Example reusable summary
"Gemini printed repeated 429 capacity retries but eventually emitted a valid `Verdict:` block. Save the substantive review artifact and do not mark the provider unavailable."

Related Skills

plan-review-prompt-refresh-after-plan-edits

5
from vamseeachanta/workspace-hub

Refresh reviewer prompt files from the latest on-disk plan before every adversarial re-review. Prevents Codex/Gemini from critiquing stale plan text after local edits.

tdd-verification-and-adversarial-review

5
from vamseeachanta/workspace-hub

Verify pre-written TDD tests pass, conduct adversarial code review on committed diffs, and route findings to existing issues

multi-role-agent-contract-review-pipeline

5
from vamseeachanta/workspace-hub

Execute a 4-role agent team (Planner/Architect/Reviewer/Integrator) pipeline for self-reviewing knowledge artifacts before delivery

multi-repo-stale-lock-recovery

5
from vamseeachanta/workspace-hub

Diagnose and recover from stale git lock files in multi-repo workspaces, especially with submodules

multi-repo-pull-with-untracked-conflict-recovery

5
from vamseeachanta/workspace-hub

Diagnose and resolve multi-repo pull failures caused by untracked files conflicting with remote changes

multi-repo-git-recovery

5
from vamseeachanta/workspace-hub

Diagnose and recover from stale lock files, diverged branches, and untracked file conflicts across multiple repos in a workspace ecosystem

gsd-adversarial-review-pattern

5
from vamseeachanta/workspace-hub

Catch hidden test failures by running adversarial review on sparse-data edge cases before final push

git-stale-lock-recovery

5
from vamseeachanta/workspace-hub

Diagnose and recover from git stale lock files and corrupted rebase states in multi-repo workflows

git-large-file-staging-conflict-recovery

5
from vamseeachanta/workspace-hub

Recover from pre-commit hook blocks on oversized files and corrupted rebase states during bulk repo syncs

freetaxusa-form-entry-recovery

5
from vamseeachanta/workspace-hub

Handle session timeouts and modal dialogs when entering tax forms in FreeTaxUSA

adversarial-code-review-tdd

5
from vamseeachanta/workspace-hub

Systematic adversarial review pattern to identify breaking assumptions in already-passing test suites

adversarial-code-review-for-committed-diffs

5
from vamseeachanta/workspace-hub

Systematic process for reviewing already-committed code changes to catch type inconsistencies, edge cases, and docstring gaps