dream

A read-only command that reviews a memory directory and recent history, then creates a proposal for improving that memory.

In plain words
What is it for?
Use it to run a memory curator, structural check, or pattern review and produce an artifact for a separate apply step.
Why use it?
It can identify duplicate, contradictory, outdated, or unreferenced memory without changing the stored memory directly.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/conorbronsdon/agent-memory-kit/dream
Clone the repo
git clone --depth 1 https://github.com/conorbronsdon/agent-memory-kit
Per session 27 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,020 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00027 $0.01020
Opus 5 $0.00014 $0.00510
Sonnet 5 $0.00005 $0.00204
Haiku 4.5 $0.00003 $0.00102

Measured 2d ago against content hash 0e0f2dd7d1a5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dream scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

commands/dream.md · 92 lines

How it starts

The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/dream — memory curator pass

Read-only curator over your memory + recent history. Produces a reviewable proposal artifact. It never edits memory directly — that's /dream-apply. Background: docs/curation.md.

Usage

/dream                # default curator: rot
/dream rot            # explicit
/dream lint           # structural integrity: index drift, duplicates, contradictions, dead references
/dream pattern        # if prompts/pattern.md exists

1. Resolve the curator

If $ARGUMENTS is empty or rot, use rot. Otherwise check prompts/{name}.md exists; if not, list available curators and stop.

2. Resolve the memory dir

In order: $AGENT_MEMORY_DIR./memory if it exists → Claude Code per-project path ~/.claude/projects/<encoded-cwd>/memory (cwd with : \ / replaced by -). Call it $MEMORY_DIR.

3. Timestamp + artifact dir

TS=$(date -u +%Y-%m-%dT%H-%M-%SZ)
DREAM_DIR="$MEMORY_DIR/.dreams/$TS"
mkdir -p "$DREAM_DIR"

4. Load the curator prompt

Read prompts/{curator}.md in full — it's your role + output schema for the rest of this command.

5. Gather inputs (read-only)

Per the curator's required inputs (rot needs all of these; lint needs only the memory dir plus cheap existence checks, per its prompt):

  • ls $MEMORY_DIR/*.md and read each project_* / reference_* (skip user_/feedback_ unless the curator asks).
  • Read state/*.md (decisions, blockers, current — whichever exist).
  • Select session logs by filename date, never by mtime, and read each:
    CUT=$(date -u -d '14 days ago' +%F 2>/dev/null || date -u -v-14d +%F)
    ls sessions/*.md | sed 's|.*/||' | grep -E '^[0-9]{4}-[0-9]{2}-[0-9]{2}\.md$' | sort | awk -v c="$CUT.md" '$0 >= c'
    
    (Why not find -mtime -14: a git history rewrite resets mtime on every tracked file, after which it matches the whole directory and silently blows up the input set. Dated filenames sort lexically, so the compare stays exact through any rewrite.)
  • git log --since="14 days ago" --oneline.

Read the full file on GitHub · 92 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 92 lines · 0 tokens per session scan A 0e0f2dd7d1a5

Subscribe to this mod's changes

dream is a command published in the GitHub repository conorbronsdon/agent-memory-kit (2 stars, last pushed 28d ago), licensed MIT. It adds 27 tokens to every session and 1,020 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.