tools/sage-claude-plugin/skills/review/SKILL.md
Independent artifact review via sub-agent delegation. Evaluates completeness, consistency, and quality with severity classification.
npx skillsauth add xoai/sage 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.
RULES (apply to every step — non-negotiable):
Review an artifact with independent evaluation via sub-agent delegation.
If not specified, scan .sage/work/ and .sage/docs/ for recent
artifacts. Present them:
Sage: Available for review:
[1] .sage/work/20260316-checkout/brief.md (updated today) [2] .sage/work/20260316-checkout/spec.md (updated today) [3] .sage/docs/ux-audit-homepage.md (updated yesterday)
Which artifact should I review?
If the user specifies an artifact, proceed directly.
Before delegating, gather three pieces of information:
.sage/decisions.md.
The quality criteria are in that skill's ## Quality Criteria section
(look in sage/skills/[skill]/SKILL.md or sage/core/workflows/[workflow].workflow.md)Delegation is MANDATORY when Task tool is available. Do NOT skip delegation because:
If Task tool is NOT available (e.g., Antigravity platform), proceed with self-review but announce it: "Sage: Task tool not available. Performing self-review — note this is not independent evaluation. Consider a fresh-session /review for critical artifacts."
Tell the user: "Sage: Delegating to a review sub-agent for independent evaluation. The reviewer gets a fresh context window without my reasoning from this session."
Use the Task tool to spawn a sub-agent with this prompt:
You are independently reviewing a Sage project artifact. You were
NOT involved in producing this work — evaluate it with fresh eyes.
CONTEXT PACKAGE:
1. PERSONA: Read sage/core/agents/reviewer.persona.md for mindset.
2. ARTIFACT: Read the artifact at: [ARTIFACT PATH]
3. CRITERIA: Read quality criteria from: [SKILL/WORKFLOW PATH],
section titled "## Quality Criteria"
4. DECISIONS: Read .sage/decisions.md for last 5 entries.
5. LEARNINGS: Call sage_memory_search(query: "[MEMORY QUERY]", limit: 5)
If this tool is not available, check .sage-memory/ folder.
EVALUATE the artifact against EACH quality criterion specifically.
CLASSIFY each finding by severity:
- CRITICAL: Blocks proceeding. Must fix before next step.
- MAJOR: Significant gap. Should fix before next step.
- MINOR: Improvement opportunity. Can fix later.
PRESENT YOUR REVIEW AS:
## Review: [artifact name]
### Critical Issues
[If none, say "None found." Do not omit this section.]
### Major Issues
[If none, say "None found." Do not omit this section.]
### Minor Issues / Improvements
[Specific observations with suggested actions]
### Strengths
[Specific observations — not generic praise]
### Verdict
PASS — ready to proceed [minor notes if any]
NEEDS REVISION — [specific items to address, with severity]
FAIL — [significant gaps, recommend returning to earlier step]
Share the sub-agent's review with the user.
Critical findings block approval. If the review contains CRITICAL issues, do NOT present [A] Accept as the first option:
Sage: Review found critical issues that must be addressed: [critical findings summary]
[R] Address critical issues first [D] Discuss — let's talk about specific findings [A] Accept anyway — I understand the risks
If no critical issues:
Sage: Review complete. [verdict summary]
[A] Accept findings — proceed with suggested next step [R] Revise — I'll address the issues found [D] Discuss — let's talk about specific findings
Append review findings to .sage/decisions.md.
$ARGUMENTS
tools
Captures agent mistakes, corrections, and discovered gotchas so they are not repeated. Use when: (1) a command or operation fails unexpectedly, (2) the user corrects the agent, (3) the agent discovers non-obvious behavior through debugging, (4) an API or tool behaves differently than expected, (5) a better approach is found for a recurring task. Also searches past learnings before starting tasks to avoid known pitfalls. Activate alongside the sage-memory skill — they share the same MCP backend but serve different purposes (sage-memory = codebase knowledge, sage-self-learning = agent mistakes and gotchas).
development
Typed knowledge graph stored in sage-memory. Use when creating or querying structured entities (Person, Project, Task, Event, Document), linking related objects, checking dependencies, planning multi-step actions as graph transformations, or when skills need to share structured state. Trigger on "remember that X is Y", "what do I know about", "link X to Y", "show dependencies", "what blocks X", entity CRUD, cross-skill data access, or any request involving structured relationships between things.
tools
Integrates sage-memory into Sage workflows. Teaches the agent when to remember (store findings during work), when to recall (search memory at session start and task start), and how to learn (structured knowledge capture via sage learn). Use when the user mentions memory, remember, recall, learn, capture knowledge, onboard to codebase, or when starting any session where sage-memory MCP tools are available.
tools
Captures agent mistakes, corrections, and discovered gotchas so they are not repeated. Use when: (1) a command or operation fails unexpectedly, (2) the user corrects the agent, (3) the agent discovers non-obvious behavior through debugging, (4) an API or tool behaves differently than expected, (5) a better approach is found for a recurring task. Also searches past learnings before starting tasks to avoid known pitfalls. Activate alongside the sage-memory skill — they share the same MCP backend but serve different purposes (sage-memory = codebase knowledge, sage-self-learning = agent mistakes and gotchas).