Agent skill
proposal-review
Facilitate methodical review of proposals (technical designs, product specs, feature requests). Use when asked to "review this proposal", "give feedback on this doc", "help me review this RFC", or when presented with a document that needs structured feedback. Handles markdown files, GitHub gists/issues/PRs, and other text formats. Chunks proposals intelligently, predicts reviewer reactions, and produces feedback adapted to the proposal's format.
Install this agent skill to your Project
npx add-skill https://github.com/petekp/agent-skills/tree/main/skills/proposal-review
SKILL.md
Proposal Review
Methodically review proposals by chunking content, predicting feedback, and producing actionable output for the proposer.
Workflow
┌─────────────────────────────────────────────────────────────┐
│ 1. INTAKE: Read entire proposal, identify source format │
├─────────────────────────────────────────────────────────────┤
│ 2. CHUNK: Split into reviewable sections (smart hybrid) │
├─────────────────────────────────────────────────────────────┤
│ 3. REVIEW LOOP: For each chunk: │
│ • Present chunk content │
│ • Predict 3-4 likely reactions │
│ • Use AskUserQuestion for feedback │
│ • Record response │
├─────────────────────────────────────────────────────────────┤
│ 4. SYNTHESIZE: Compile feedback, infer overall sentiment │
├─────────────────────────────────────────────────────────────┤
│ 5. OUTPUT: Generate feedback document matching source │
└─────────────────────────────────────────────────────────────┘
Phase 1: Intake
Read the entire proposal. Identify:
- Source format: Local file, GitHub PR/issue/gist, Google Doc export, etc.
- Structure: Headers, sections, numbered lists, or flowing prose
- Length: Estimate chunk count (aim for 3-8 chunks for typical proposals)
Do not summarize or share initial impressions. Proceed directly to chunking.
Phase 2: Chunking Strategy
Use smart hybrid chunking:
| Proposal Structure | Chunking Approach |
|---|---|
| Clear headers/sections | One chunk per major section |
| Large section (>500 words) | Split at natural paragraph breaks |
| Small adjacent sections (<100 words each) | Merge into single chunk |
| Numbered lists of items | Group 3-5 related items per chunk |
| Flowing prose without structure | Split at topic transitions (~300-400 words) |
Chunk ordering: Present in document order unless there's a clear dependency (e.g., "Alternatives" before "Proposed Solution" if alternatives inform the solution review).
Phase 3: Review Loop
For each chunk:
3.1 Present the Chunk
Quote or summarize the chunk content. For longer chunks, quote key sentences and summarize the rest. Use a clear header like:
### Chunk 2 of 5: Technical Architecture
3.2 Predict Reactions
Generate 3-4 predicted reactions spanning these categories:
| Category | Example Predictions |
|---|---|
| Clarification | "This is unclear—what does X mean?", "How does this interact with Y?" |
| Concern | "This scope seems too large", "Have you considered Z risk?" |
| Approval | "This approach makes sense", "Good tradeoff analysis" |
| Suggestion | "Consider alternative A", "This needs more detail on B" |
Select predictions that feel most relevant to this specific chunk. Not every chunk needs all categories.
3.3 Collect Feedback
Use AskUserQuestion with:
- Predicted reactions as options (2-4 most likely)
- User can select one OR provide custom feedback via "Other"
- Keep option labels concise (under 10 words), use description for detail
Example:
question: "What's your reaction to this technical architecture section?"
header: "Architecture"
options:
- label: "Looks good"
description: "The proposed architecture is sound and well-reasoned"
- label: "Scope concern"
description: "This feels too ambitious for the timeline"
- label: "Need clarification"
description: "Some technical details are unclear or missing"
- label: "Consider alternative"
description: "There may be a simpler or better approach"
3.4 Record Response
Store each response with:
- Chunk identifier (number + title)
- Selected option or custom text
- Any quoted content the feedback references
Phase 4: Synthesis
After all chunks reviewed:
- Group feedback by theme: Consolidate similar concerns across chunks
- Infer overall sentiment: Based on feedback distribution:
- Mostly approvals → Positive with minor suggestions
- Mixed → Conditional support, needs revisions
- Mostly concerns → Significant issues to address
- Identify patterns: Note if same concern appears multiple times
Then ask:
question: "Would you like me to include suggested next steps for the proposer?"
header: "Next Steps"
options:
- label: "Yes, include action items"
description: "Generate concrete next steps based on feedback"
- label: "No, just the feedback"
description: "Keep output to observations and reactions only"
Phase 5: Output Generation
Adapt output format to source:
| Source | Output Format |
|---|---|
| GitHub PR | PR review comment with quoted lines and threaded feedback |
| GitHub Issue | Comment with sections matching issue structure |
| Markdown file | Companion *-feedback.md with inline references |
| Google Doc | Structured comment list with section references |
| Generic/unknown | Structured markdown with clear sections |
Output Structure
## Feedback Summary
**Overall**: [Inferred sentiment - 1 sentence]
## Section-by-Section Feedback
### [Section Name]
[Feedback with quotes where relevant]
### [Section Name]
...
## Key Themes
- [Theme 1]: [Consolidated feedback]
- [Theme 2]: ...
## Next Steps (if requested)
- [ ] [Action item 1]
- [ ] [Action item 2]
Tone Guidelines
- Direct but constructive
- Quote specific text when critiquing
- Frame concerns as questions when possible ("Have you considered..." vs "This won't work")
- Acknowledge what works, not just what doesn't
Edge Cases
Very short proposals (<300 words): Skip chunking, review as single unit with 4-5 predicted reactions.
Very long proposals (>3000 words): Cap at 8-10 chunks. Merge aggressively or offer to focus on specific sections.
Unclear structure: Ask user which sections matter most before chunking.
Multiple proposals: Review one at a time. Ask user for order preference if not obvious.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
multi-model-meta-analysis
Synthesize outputs from multiple AI models into a comprehensive, verified assessment. Use when: (1) User pastes feedback/analysis from multiple LLMs (Claude, GPT, Gemini, etc.) about code or a project, (2) User wants to consolidate model outputs into a single reliable document, (3) User needs conflicting model claims resolved against actual source code. This skill verifies model claims against the codebase, resolves contradictions with evidence, and produces a more reliable assessment than any single model.
capture-learning
Analyze recent conversation context and capture learnings to project knowledge files (for project-specific insights) or skills/commands/subagents (for cross-project patterns). Use when the user asks to "capture this learning", "update the docs with this", "remember this for next time", "document this issue", "add this to CLAUDE.md", "save this knowledge", or "update project knowledge". Also triggers after resolving build/setup issues, discovering non-obvious patterns, or completing debugging sessions with valuable insights.
optimize-agent-docs
Build a retrieval-optimized knowledge layer over agent documentation in dotfiles (.claude, .codex, .cursor, .aider). Use when asked to "optimize docs", "improve agent knowledge", "make docs more efficient", or when documentation has accumulated and retrieval feels inefficient. Generates a manifest mapping task-contexts to knowledge chunks, optimizes information density, and creates compiled artifacts for efficient agent consumption.
agent-changelog
Compile an agent-optimized changelog by cross-referencing git history with plans and documentation. Use when asked to "update changelog", "compile history", "document project evolution", or proactively after major milestones, architectural changes, or when stale/deprecated information is detected that could confuse coding agents.
literate-guide
Create a narrative guide to a codebase or feature in the style of Knuth's Literate Programming — code and prose interwoven as a single essay, ordered for human understanding rather than compiler needs. Use when the user asks to 'explain this codebase as a story', 'write a literate guide', 'create a narrative walkthrough', 'tell the story of this code', 'Knuth-style documentation', 'weave a guide for this feature', or when they want deep, readable documentation that treats the program as literature. Also trigger when someone wants a document that a thoughtful reader could follow from start to finish and come away understanding both WHAT the code does and WHY every design choice was made.
autonomous-agent-readiness
Assess a codebase's readiness for autonomous agent development and provide tailored recommendations. Use when asked to evaluate how well a project supports unattended agent execution, assess development practices for agent autonomy, audit infrastructure for agent reliability, or improve a codebase for autonomous agent workflows. Triggers on requests like "assess this project for agent readiness", "how autonomous-ready is this codebase", "evaluate agent infrastructure", or "improve development practices for agents".
Didn't find tool you were looking for?