dream-state
Maintains architectural health by scanning, validating, and reporting on project memory structures.
Install
mkdir -p .claude/skills/dream-state-fabioc-aloha && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/15127" && unzip -o skill.zip -d .claude/skills/dream-state-fabioc-aloha && rm skill.zipInstalls to .claude/skills/dream-state-fabioc-aloha
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Architecture maintenance, connection validation, automated health diagnostics, and unconscious processingKey capabilities
- →Scan memory files for structural integrity
- →Validate frontmatter completeness
- →Generate a diagnostic report
- →Check for broken connections
- →Assess architecture health metrics
How it works
The skill scans memory files, validates their structural integrity, and produces a diagnostic report without modifying any files.
Inputs & outputs
When to use dream-state
- →Running project health checks
- →Repairing broken memory architecture connections
- →Auditing project documentation completeness
About this skill
Dream State Skill
Automated architecture maintenance — scan every memory file, validate structural integrity, produce a diagnostic report. Dream observes and measures; it never modifies.
When to Use
- User asks to "dream" or run "architecture maintenance"
- After file reorganizations that may break connections
- Before major meditation sessions (establish baseline)
- After domain learning to validate new connections
- When architecture health is uncertain before skill selection
6-Phase Protocol
| Phase | Action | Output |
|---|---|---|
| 1. Discovery | Scan .github/ for all memory files | File inventory |
| 2. Validation | Parse frontmatter, verify trifecta completeness | Issue list |
| 3. Repair | Auto-fix via consolidation mappings (renames only) | Repair log |
| 4. GK Sync | Check %OneDrive%/AI-Memory/ for cross-platform content | Sync status |
| 5. Reporting | Generate health report with statistics | Markdown report |
| 6. Display | Show results, open report in VS Code | Notification |
Health Metrics
| Metric | Healthy | Concern |
|---|---|---|
| Broken connections | 0 | Any |
| Missing frontmatter | 0 | Any |
| Incomplete trifectas | 0 | Any high-use skills |
| Brand violations | 0 (outside exceptions) | Any |
| Version drift | All match | Any mismatch |
| Review queue size | 0–5 pending | >10 pending (semantic debt accumulating) |
| Review queue age | Oldest <30d | Oldest >60d (stale reviews indicate blocked curation) |
Review-Queue Backpressure (FM10)
The review queue (.github/quality/review-queue.jsonl, defined in FM5) is a leading indicator of semantic debt. When the queue grows faster than it's resolved, the architecture is accumulating unreviewed changes.
| Signal | Threshold | Health Impact | Action |
|---|---|---|---|
| Queue size growing | >10 pending entries | needs-attention | Prioritize review session; schedule meditation |
| High-priority entries stale | Any high priority entry >14d unresolved | needs-attention | Escalate: these are heir-feedback or drift-detected entries |
| Queue empty | 0 pending | healthy | Normal state; curation is keeping pace |
| Oldest entry age | >60 days | needs-attention | Something is blocking reviews; investigate |
Per-volatility weighting: High-volatility files (instructions, skills with frequent edits) generate more queue pressure than stable reference docs. Weight queue entries by the source file's edit frequency when computing backpressure score.
Integration: Dream protocol reads review-queue.jsonl during Phase 2 (architecture scan) and includes queue metrics in the dream report's health assessment.
Ritual Hierarchy
Meditation is the foundational ritual. Dream is a diagnostic that runs subordinate to it.
| Dream | Meditation | |
|---|---|---|
| Role | Diagnostic tool | Foundational ritual |
| Mode | Automated, unconscious | Interactive, conscious |
| Creates files? | Reports only | Memory files + insights |
| Makes decisions? | Never | Yes |
| Activation | /dream, or chained after meditation | User says "meditate" |
| Analogy | Blood test | Doctor's visit |
When Dream Chains After Meditation
Dream triggers automatically after meditation when:
| Trigger | Condition |
|---|---|
| Pattern: file reorg | Meditation touched 3+ architecture files |
| Pattern: trifecta concern | Meditation created or modified skills/instructions |
| Pattern: staleness | Last dream was >7 days ago |
| Random | ~1 in 5 meditations (keeps architecture fresh) |
Connection Format
Skills connect via frontmatter applyTo patterns and semantic search:
---
name: "skill-name"
description: "What this skill does"
applyTo: '**/*pattern*,**/*files*'
---
Valid relations: applies-to, extends, requires, contradicts, supersedes
Report Template
# Dream Report — YYYY-MM-DD
## Summary
- **Files Scanned**: N
- **Trifecta Completeness**: N/N
- **Broken Links**: 0
- **Brand Violations**: 0
- **Status**: HEALTHY / ATTENTION REQUIRED
## Issues Found
| File | Issue | Severity |
|------|-------|----------|
## Recommendations
Prioritized list of manual fixes needed.
Interpreting Results
| Status | Meaning | Action |
|---|---|---|
| HEALTHY | All checks pass, no broken connections | None — architecture is sound |
| ATTENTION REQUIRED | One or more issues detected | Review issues table, prioritize by severity |
Severity levels: critical (broken trifecta), warning (missing frontmatter field), info (cosmetic).
Escalation: When Findings Need Human Judgment
Dream is a diagnostic — it never modifies files. When findings require judgment (skill consolidation, content quality decisions, contradiction resolution), the natural next step is meditation, which reads the dream report and decides what to fix.
| Finding type | Resolution path |
|---|---|
| Missing frontmatter, broken link with mapping | Author the fix in the next meditation; mechanical |
Two skills with overlapping applyTo | Meditation decides merge / split / keep; semantic |
| Skill flagged as shallow | Meditation enriches or deprecates; requires domain context |
| Trifecta where instruction contradicts skill | Meditation reconciles intent; requires understanding |
| Version drift master/heir | Meditation picks canonical; project-specific |
The handoff is the dream report file. Meditation's Resolve step explicitly evaluates whether a dream should chain after it; the inverse — meditation reading an existing dream report — happens naturally when a meditation session opens with "check the architecture first."
This pattern replaces the former lucid-dream ritual (retired v8.4.0). The escalation concept lives here because dream is the producer of the findings; meditation is the consumer that decides.
Snapshot Schema (.github/quality/dream-report.json)
dream-cli.cjs writes a single structural JSON snapshot every run. Heirs and graders depend on this shape; preserve it across versions.
| Field | Type | Purpose |
|---|---|---|
schemaVersion | number | Currently 1. Bump only with a heir migration plan. |
timestamp | ISO string | Last dream run; consumed by proactive-awareness and health-pulse. |
workspace | string | Absolute path of the dreamed workspace. |
inventory.skills / .instructions / .prompts / .agents / .total | numbers | File counts under .github/. |
trifectaIssues[] | {type, skill, detail} | Workflow skills missing matching .instructions.md or .prompt.md. |
brokenRefs[] | {sourceFile, line, target} | Markdown / config references that no longer resolve. |
health | "healthy" | "good" | "needs-attention" | Derived bucket: 0 / ≤5 / >5 issues. |
Stability rule: meditation, dream-creativity-score.cjs, and heir consumers (health-pulse, proactive-awareness.instructions.md) all read these field names directly. Adding fields is safe; renaming or removing requires a fleet-wide migration.
Chronicle Pruning
dream-cli.cjs keeps the newest 50 dream-report-YYYY-MM-DD*.md chronicles in .github/episodic/ after each run; older ones are deleted in place. Rolling architecture history lives in the JSON snapshot and brain-qa outputs — chronicles are only useful as recent narrative context for meditation.
When not to use it
- →When the user needs to make decisions
Limitations
- →Dream never modifies files
- →Dream never makes decisions
- →Output is a report, not a resolution
How it compares
This skill provides an automated, diagnostic assessment of architecture health, contrasting with manual checks or interactive decision-making processes.
Compared to similar skills
dream-state side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| dream-state (this skill) | 0 | 3mo | No flags | Beginner |
| chief-architect | 6 | 7mo | No flags | Advanced |
| ralplan | 1 | 2mo | Review | Intermediate |
| moai-workflow-thinking | 2 | 3mo | No flags | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by fabioc-aloha
View all by fabioc-aloha →You might also like
chief-architect
ananddtyagi
PERSONAL APP ARCHITECT - Strategic development orchestrator for personal productivity applications. Analyzes project context, makes architectural decisions for single-developer projects, delegates to specialized skills, and ensures alignment between user experience goals and technical implementation. Optimized for personal apps targeting 10-100 users.
ralplan
Yeachan-Heo
Iterative planning with Planner, Architect, and Critic until consensus
moai-workflow-thinking
modu-ai
Sequential Thinking MCP and UltraThink mode for deep analysis, complex problem decomposition, and structured reasoning workflows
agent-migration-plan
ruvnet
Agent skill for migration-plan - invoke with $agent-migration-plan
granola-reference-architecture
jeremylongshore
Enterprise meeting workflow architecture with Granola. Use when designing enterprise deployments, planning integrations, or architecting meeting management systems. Trigger with phrases like "granola architecture", "granola enterprise", "granola system design", "meeting system", "granola infrastructure".
phased-migration-plan
kaladron
Author or update a phased plan-*.md document for a multi-step migration or refactor. Use when a change is too large for a single commit, when coordinating database/architecture migrations across many call sites, or when the user asks for a "plan" before implementation. Covers the standard plan struc