dream
Private overnight operator mode. Routes interactive Dream requests to the shared `ao overnight` engine for setup, bedtime runs, and morning reports. Triggers: "$dream", "overnight", "bedtime run", "morning report", "dream setup", "dream report", "dream council", "dreamscape".
Best use case
dream is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Private overnight operator mode. Routes interactive Dream requests to the shared `ao overnight` engine for setup, bedtime runs, and morning reports. Triggers: "$dream", "overnight", "bedtime run", "morning report", "dream setup", "dream report", "dream council", "dreamscape".
Teams using dream should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/dream/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How dream Compares
| Feature / Agent | dream | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Private overnight operator mode. Routes interactive Dream requests to the shared `ao overnight` engine for setup, bedtime runs, and morning reports. Triggers: "$dream", "overnight", "bedtime run", "morning report", "dream setup", "dream report", "dream council", "dreamscape".
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Dream Skill `$dream` is the interactive surface for the same Dream engine used by `ao overnight`. ## Compounding Loop (v2) Dream v2 runs a bounded outer loop of `INGEST -> REDUCE -> MEASURE` iterations until a halt condition fires: wall-clock budget exhausted, plateau (K sub-epsilon deltas in a row), regression beyond a per-metric floor, or metadata integrity failure. Each iteration is atomic and checkpointed so any rollback leaves the corpus clean. Dream is strictly knowledge-only. Anti-goals (hard constraints): - NEVER mutates source code. - NEVER invokes `$rpi` or any code-mutating flow. - NEVER performs git operations (no commits, branches, push, rebase, checkout). - NEVER creates symlinks anywhere. - No swarm/gc fan-out inside iterations in the first slice (serial only). ## Execution Steps ### Step 1: Route the request - `setup` -> `ao overnight setup` - `curator` or local Gemma worker requests -> `ao overnight curator status|diagnose|enqueue|compact|event` - `start` or `run` -> `ao overnight start` - `report` -> `ao overnight report` ### Step 2: Setup lane Use `ao overnight setup` to inspect host constraints, runner availability, scheduler mode, and keep-awake behavior. ```bash ao overnight setup ao overnight setup --apply --runner codex --runner claude --at 01:30 ``` Default to preview. Use `--apply` only when the user explicitly wants Dream config or scheduler artifacts persisted. Setup detects Tier 1 local curator state separately from Tier 2 Dream Council runners. ### Step 2a: Local curator lane Use `ao overnight curator` when the user asks about Gemma, Ollama, the local worker, SOC trigger signals, Tier 1 drafts, or pending LOG compaction. ```bash ao overnight curator status --json ao overnight curator diagnose ao overnight curator enqueue --kind lint-wiki ao overnight curator enqueue --kind dream-seed ao overnight curator compact --dry-run ao overnight curator event --source local-soc --severity high --desired-action "review alert cluster" --budget 1 ``` The first supported local curator shape is Ollama + Gemma under `dream.local_curator.*`. Treat it as a Tier 1 draft/lint/triage lane, not as a Dream Council runner. Gemma may enqueue allowlisted knowledge jobs and emit needs-review event records; Codex and Claude remain Tier 2 review/synthesis runners; humans own promotion into durable authored memory. Do not create an unbounded model-to-model loop. Any escalation needs an explicit source, severity, desired action, escalation target, budget, and ledger entry. ### Step 3: Bedtime run lane Use `ao overnight start` for the actual private local run. ```bash ao overnight start --goal "close the loop on today's auth work" ao overnight start --goal "stabilize release follow-ups" --runner codex --runner claude --creative-lane $dream start --queue=.agents/dream/tonight.md $dream start --max-iterations=3 $dream start --warn-only=false ``` Expected behavior: - operates against the real repo-local `.agents` corpus - writes `summary.json` and `summary.md` (with per-iteration sub-summary entries for each INGEST -> REDUCE -> MEASURE pass) - persists each iteration atomically to `<output-dir>/<runID>/iterations/iter-<N>.json`; resumed runs rehydrate prior iterations from disk instead of starting from a clean slate (Micro-epic 2) - degrades honestly when soft-fail steps or keep-awake helpers are unavailable ### Step 4: Morning report lane Use `ao overnight report` to render the latest Dream result. ```bash ao overnight report ao overnight report --from .agents/overnight/latest ``` When rendering a report, answer four questions fast: 1. What state did I wake up to? 2. What ran overnight? 3. What degraded or failed? 4. What should I do first? ## Key Rules - Keep Dream settings under the shared `dream.*` control plane. - Do not promise scheduled execution on a sleeping laptop. - Do not imply tracked source-code edits overnight. - GitHub nightly is the public proof harness, not the private Dream engine. ## Delineation vs $evolve | Lane | Runs | Mutates code? | Mutates corpus? | Outer loop? | Budget | |------|------|---------------|-----------------|-------------|--------| | `$dream` | nightly, private local | **No** | **Yes (heavy)** | **Yes (convergence)** | wall-clock + plateau | | `$evolve` | daytime, operator-driven | Yes (via `$rpi`) | Yes (light) | Yes | cycle cap | Dream owns the knowledge compounding layer; `$evolve` owns the code compounding layer. Both share fitness-measurement substrate via `corpus.Compute` / `ao goals measure`. Run Dream overnight, then start each day with `$evolve` against the freshly-compounded corpus with a clean fitness baseline. ## See Also - [Dream Run Contract](../../docs/contracts/dream-run-contract.md) - [Dream Report Contract](../../docs/contracts/dream-report.md)
Related Skills
vibe
Comprehensive code validation. Runs complexity analysis then multi-model council. Answer: Is this code ready to ship? Triggers: "vibe", "validate code", "check code", "review code", "code quality", "is this ready".
validation
Full validation phase orchestrator. Vibe + post-mortem + retro + forge. Reviews implementation quality, extracts learnings, feeds the knowledge flywheel. Triggers: "validation", "validate", "validate work", "review and learn", "validation phase", "post-implementation review".
update
Reinstall all AgentOps skills globally from the latest source. Triggers: "update skills", "reinstall skills", "sync skills".
trace
Trace design decisions and concepts through session history, handoffs, and git. Triggers: "trace decision", "how did we decide", "where did this come from", "design provenance", "decision history".
test
Test generation, coverage analysis, and TDD workflow. Triggers: "test", "generate tests", "test coverage", "write tests", "tdd", "add tests", "test strategy", "missing tests", "coverage gaps".
status
Single-screen dashboard showing current work, recent validations, flywheel health, and suggested next action. Triggers: "status", "dashboard", "what am I working on", "where was I".
standards
Language-specific coding standards and validation rules. Provides Python, Go, Rust, TypeScript, Shell, YAML, JSON, and Markdown standards. Auto-loaded by $vibe, $implement, $doc, $bug-hunt, $complexity based on file types.
shared
Shared reference documents for multi-agent skills (not directly invocable)
security
Continuous repository security scanning and release gating. Triggers: "security scan", "security audit", "pre-release security", "run scanners", "check vulnerabilities".
security-suite
Composable security suite for binary and prompt-surface assurance, static analysis, dynamic tracing, repo-native redteam scans, contract capture, baseline drift, and policy gating. Triggers: "binary security", "reverse engineer binary", "black-box binary test", "behavioral trace", "baseline diff", "prompt redteam", "security suite".
scenario
Author and manage holdout scenarios for behavioral validation. Scenarios are stored in .agents/holdout/ where implementing agents cannot see them. Triggers: "$scenario", "holdout", "behavioral scenario", "create scenario", "list scenarios".
scaffold
Project scaffolding, component generation, and boilerplate setup. Triggers: "scaffold", "new project", "init project", "create project", "generate component", "setup project", "starter", "boilerplate".