smithery/mystilleef

review-mode

Critically reviews roadmaps, code, and strategies. Use when you need a technical review to identify flaws, risks, and improvements.

Installation

$ npx skills add smithery/mystilleef --skill review-mode

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from smithery/mystilleef.

npx skills add smithery/mystilleef

Browse all from smithery/mystilleef

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 3,824 B
  • docs SUMMARY.md 88 B

History

  1. First recorded snapshot · 0 installs

SKILL.md

Review mode

GOAL: Conduct critical technical reviews of roadmaps, code, and strategies following standard review protocols.

WHEN: Invoke this skill when the user requests a critical review of a roadmap or proposed changes.

NOTE: This skill operates strictly in read-only mode to ensure safety during analysis.

Efficiency directives

  • Optimize all operations for agent, token, and context efficiency
  • Optimize for minimal output
  • Batch operations on file groups, avoid individual file processing
  • Target only relevant files
  • Reduce token usage

Confirmation directives

After reporting the review decision, use the ask_user tool to offer 4 options:

  1. Revise plan - Invoke the prepare-mode skill to address findings
  2. Quick build - Invoke the build-mode skill for rapid execution
  3. Implement - Invoke the implement-mode skill for thorough

execution

  1. Abort - Cancel the workflow and wait for the next instruction

Set the default response based on the review recommendation:

  • If recommendation indicates REVISE or REJECT, make option 1 the

default.

  • If recommendation indicates APPROVE, make option 2 the default.

Workflow

Step 1: Enforce read-only

  • Invoke the readonly-mode skill.
  • Capture status (SUCCESS, WARN, ERROR).
  • Handle status:

- ERROR: Halt and report. - SUCCESS/WARN: Continue.

Step 2: Perceive

  • Read the roadmap, code, or context provided for review.

Step 3: Analyze

  • Apply the 6-step reasoning engine:

1. Analyze: Context, objectives, and proposed changes. 2. Evaluate: Risks from Security, QA, and Ops perspectives. 3. Identify: Missing edge cases, logical flaws, and debt. 4. Revise: Understanding based on deep-dive analysis. 5. Incorporate: KBase patterns and project constraints. 6. Retry: If analysis yields insufficient confidence.

  • Execute Multi-perspective analysis across five viewpoints:

- Security: Vulnerabilities, permissions, data handling. - QA: Test coverage, testability, regression risks. - Architecture: Design patterns, scalability, maintainability. - Performance: Latency, resource usage, optimization. - DevOps: Deployment, monitoring, infrastructure impact.

Step 4: Critique & assess risk

  • Compare against KBase and best practices.
  • Re-evaluate the risk level (TRIVIAL, LOW, MEDIUM, HIGH).

Step 5: Report

  • Output the structured review decision.

Step 6: Confirmation

  • Use the ask_user tool for confirmation with 4 options.
  • Await user response before further action.
  • DONE

Review report format

Review checklist:

  1. Security: [Findings/None]
  2. QA: [Findings/None]
  3. Architecture: [Findings/None]
  4. Performance: [Findings/None]
  5. DevOps: [Findings/None]

Risk re-assessment:

  • Level: [TRIVIAL/LOW/MEDIUM/HIGH]
  • Justification: [Reasoning]

Decision:

  • Recommendation: [APPROVE ✅ / REVISE 🔄 / REJECT ❌]
  • Blockers: [Critical Issues]
  • Concerns: [Moderate Issues]
  • Next Steps: [Actionable advice]

Output

Files created/modified:

  • None (Read-only operation).
  • .gemini_readonly - Ensured at the start.

Status communication:

First line of output indicates user's decision:

  • REVISE: user wants to revise the plan - user chose revision
  • BUILD: user wants to build the plan - user chose quick build
  • IMPLEMENT: user wants to implement the plan - user chose

implementation

  • ABORT: user cancelled workflow - user aborted process

Following lines: complete review report text