skills/spec-review/SKILL.md
Specification Review Protocol
npx skillsauth add liza-mas/liza spec-reviewInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Review technical specifications for inconsistencies, gaps, contradictions, and ambiguities. This skill is for finding specification issues, not proposing design changes.
Use this skill when:
The user should provide:
specs/ if it exists.If not provided, ask:
To review your specifications, I need:
1. Where are the spec files? (e.g., specs/, docs/)
2. Any specific concerns to focus on?
If token limits force truncation mid-analysis, stop and report:
Map the specification corpus
Classify documents
Read all documents completely
Review against these categories. For each issue found, record:
file.md:123, file.md:50-55, file.md#section, or quoted snippet (≤15 words) if line numbers unavailable| Type | Definition | |------|------------| | Inconsistency | Same concept defined differently in different places | | Gap | Missing information needed for implementation | | Contradiction | Two statements that cannot both be true | | Ambiguity | Statement that can be interpreted multiple ways | | Missing Reference | Cross-reference to non-existent section or document | | Undefined Term | Term used without definition |
| Severity | Definition | |----------|------------| | Critical | Blocks implementation or causes runtime failure | | High | Causes significant confusion or likely bugs | | Medium | Reduces spec quality, may cause minor issues | | Low | Nitpick, polish, or style issue |
Apply these checks to the specification corpus:
Generate report in this format:
# Specification Review: [Project/Component Name]
## Summary
- Critical: N
- High: N
- Medium: N
- Low: N
---
## Critical Issues
### [Issue Title]
- **Location:** file.md:123 or file.md#section
- **Type:** [Issue Type]
- **Description:** [What's wrong]
- **Suggestion:** [How to fix]
---
## High Issues
[Same format]
---
## Medium Issues
[Same format]
---
## Low Issues
[Same format]
---
## Recommendations
[Overall observations about spec quality and suggested improvements]
### Agent State Missing from State Machine
- **Location:** roles.md:330 vs state-machines.md:102-108
- **Type:** Inconsistency
- **Description:** roles.md shows agents register with `status: STARTING`, but state-machines.md only defines IDLE, WORKING, WAITING, HANDOFF. STARTING is not a valid state.
- **Suggestion:** Add STARTING to state machine with transition STARTING → IDLE, or change registration to use IDLE.
### Grace Period Duration Undefined
- **Location:** state-machines.md:263
- **Type:** Undefined Term
- **Description:** Validation rule references "grace period" but duration is never defined.
- **Suggestion:** Define explicitly (e.g., "60 seconds") or reference related timing constant.
### Backoff Timing Mismatch
- **Location:** design.md:50-55 vs implementation.md:120-125
- **Type:** Contradiction
- **Description:** design.md specifies exponential backoff (10s, 20s, 40s), but implementation.md shows fixed 5s delay.
- **Suggestion:** Align documents — update implementation to match design or document the simplification.
development
Coordinate Pairing-mode doer/reviewer sessions through a Markdown blackboard. Use when the user invokes /adversarial-pairing with role and blackboard-path arguments or asks multiple pairing agents to coordinate plan review, implementation, staged code review, and follow-up review rounds without Liza multi-agent mode.
data-ai
Analyze Liza agents logs
development
Code Review Protocol
tools
Analyze Liza `.liza/agent-prompts/` and `.liza/agent-outputs/` from a context-engineering perspective: prompt payload shape, context budget use, cacheability, duplicated or missing context, instruction hierarchy, tool-output pressure, role-specific context fit, and prompt-output feedback loops. Use when diagnosing agent context bloat, prompt drift, poor agent handoffs, repeated misunderstandings, excessive tool output, or whether Liza agents received the right information at the right time.