dream

Name: dream
Author: tranhieutt

69stars41forks

How to install dream?

tranhieutt/software_development_department/dream

$npx -y skills add tranhieutt/software_development_department --skill dream

Installs into the current project.

Summary

Consolidates and prunes the memory directory through a 4-phase reflective pass: Orient, Gather, Consolidate, Prune. Run after intensive sessions, when memories feel duplicated, or when MEMORY.md approaches the 200-line limit.

Files · 1

View on GitHub

SKILL.md

1	# Dream: Memory Consolidation
2
3	You are performing a dream — a reflective pass over your memory files. Synthesize what you've learned recently into durable, well-organized memories so that future sessions orient quickly.
4
5	Memory directory: find the path from your system prompt — look for the "auto memory" section which says "You have a persistent, file-based memory system at `<path>`". That is your memory directory. If no such section exists, default to `~/.claude/projects/<sanitized-cwd>/memory/` where `<sanitized-cwd>` is the current working directory with `/` replaced by `-`.
6
7	Session transcripts: look for `*.jsonl` files in the project directory inside `~/.claude/projects/` that corresponds to the current working directory. These are signal sources — grep narrowly, do NOT read whole files.
8
9	If the user passed a topic argument (e.g., `/dream feedback`), focus Phase 2–3 on that topic only. Still run Phase 1 and Phase 4 in full.
10
11	---
12
13	## Phase 1 — Orient
14
15	1. `mkdir -p` the memory directory if it doesn't exist yet.
16	2. `ls` the memory directory — note all existing topic files.
17	3. Read `MEMORY.md` (if it exists) to understand the current index. If it doesn't exist, this is a fresh start — you'll create it in Phase 4.
18	4. Skim existing topic files to build a picture of what's already recorded — so you improve rather than duplicate.
19	5. If `logs/` or `sessions/` subdirectories exist, note their presence for Phase 2.
20	6. Check for topic files missing mandatory YAML frontmatter (name, description, type) — flag these for repair in Phase 3.
21
22	Report: memory directory path, file count, MEMORY.md line count (or "new"), any frontmatter issues found.
23
24	---
25
26	## Phase 2 — Gather recent signal
27
28	Look for new information worth persisting. Do NOT exhaustively read transcripts. Look only for things you already suspect matter.
29
30	Sources in priority order:
31
32	1. Daily logs (`logs/YYYY/MM/YYYY-MM-DD.md`) if present — append-only stream, check recent entries first.
33	2. Drifted facts — scan existing memories for claims that may contradict the current codebase (file paths, function names, flags). Verify with Glob/Grep before marking as stale.
34	3. Transcript search — grep JSONL files narrowly for specific context:
35	```bash
36	grep -rn "<narrow term>" ~/.claude/projects/ --include="*.jsonl" \| tail -50
37	```
38	Useful search terms: user corrections ("don't", "no not", "stop"), confirmations ("exactly", "perfect"), role signals ("I'm a", "I own"), decisions ("we're going with", "we decided"), deadlines, external system URLs.
39
40	Signal categories to look for:
41
42	\| Category \| Type \| Trigger phrases \|
43	\|---\|---\|---\|
44	\| User corrections \| feedback \| "no", "don't", "stop doing", "not like that", "actually..." \|
45	\| Confirmed approaches \| feedback \| "yes exactly", "perfect", silent acceptance of unusual choice \|
46	\| Role / expertise \| user \| "I'm a ...", "I've been writing X for Y years", "I own the ..." \|
47	\| Project decisions \| project \| "we're going with", "we decided", deadlines, incidents \|
48	\| External pointers \| reference \| Dashboard URLs, issue trackers, Slack channels, doc links \|
49
50	What NOT to gather (strict exclusions):
51	- Code patterns, architecture, file paths, project structure — derivable from reading code.
52	- Git history or who-changed-what — `git log` / `git blame` are authoritative.
53	- Debugging solutions — the fix is in the code; context is in the commit message.
54	- Anything already in CLAUDE.md files.
55	- Ephemeral task state or current conversation context.
56	- PR lists / activity summaries — save only what was surprising or non-obvious.
57
58	---
59
60	## Phase 3 — Consolidate
61
62	Before making any file modifications, show the user a plan:
63	- List files to be created, updated, merged, or deleted.
64	- Wait for user approval before proceeding.
65
66	After approval, execute:
67
68	### 3a. Repair missing frontmatter
69	For any topic file flagged in Phase 1, add the required YAML frontmatter:
70	```markdown
71	---
72	name: {{memory name}}
73	description: "Consolidates and prunes the memory directory through a 4-phase reflective pass: Orient, Gather, Consolidate, Prune. Run after intensive sessions, when memories feel duplicated, or when MEMORY.md approaches the 200-line limit."
74	type: {{user \| feedback \| project \| reference}}
75	---
76	```
77
78	### 3b. Merge duplicates and near-duplicates
79	If two or more files cover the same topic (e.g., `feed

tranhieuttPublisher · 88 listings

Preview

tranhieutt/software_development_department

# Dream: Memory Consolidation

You are performing a dream — a reflective pass over your memory files. Synthesize what you've learned recently into durable, well-organized memories so that fut

**Memory directory:** find the path from your system prompt — look for the "auto memory" section which says "You have a persistent, file-based memory system at

**Session transcripts:** look for `*.jsonl` files in the project directory inside `~/.claude/projects/` that corresponds to the current working directory. These

Repotranhieutt/software_development_department

LicenseMIT

TypeSkills

CategoryPlanning

UpdatedMay 2026

Repo stars69

First seenJul 22, 2026

People who liked this also viewed

6 picks

Type

dream

69stars41forks

View on GitHub

How to install dream?

tranhieutt/software_development_department/dream

$npx -y skills add tranhieutt/software_development_department --skill dream

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Dream: Memory Consolidation
2
3	You are performing a dream — a reflective pass over your memory files. Synthesize what you've learned recently into durable, well-organized memories so that future sessions orient quickly.
4
5	Memory directory: find the path from your system prompt — look for the "auto memory" section which says "You have a persistent, file-based memory system at `<path>`". That is your memory directory. If no such section exists, default to `~/.claude/projects/<sanitized-cwd>/memory/` where `<sanitized-cwd>` is the current working directory with `/` replaced by `-`.
6
7	Session transcripts: look for `*.jsonl` files in the project directory inside `~/.claude/projects/` that corresponds to the current working directory. These are signal sources — grep narrowly, do NOT read whole files.
8
9	If the user passed a topic argument (e.g., `/dream feedback`), focus Phase 2–3 on that topic only. Still run Phase 1 and Phase 4 in full.
10
11	---
12
13	## Phase 1 — Orient
14
15	1. `mkdir -p` the memory directory if it doesn't exist yet.
16	2. `ls` the memory directory — note all existing topic files.
17	3. Read `MEMORY.md` (if it exists) to understand the current index. If it doesn't exist, this is a fresh start — you'll create it in Phase 4.
18	4. Skim existing topic files to build a picture of what's already recorded — so you improve rather than duplicate.
19	5. If `logs/` or `sessions/` subdirectories exist, note their presence for Phase 2.
20	6. Check for topic files missing mandatory YAML frontmatter (name, description, type) — flag these for repair in Phase 3.
21
22	Report: memory directory path, file count, MEMORY.md line count (or "new"), any frontmatter issues found.
23
24	---
25
26	## Phase 2 — Gather recent signal
27
28	Look for new information worth persisting. Do NOT exhaustively read transcripts. Look only for things you already suspect matter.
29
30	Sources in priority order:
31
32	1. Daily logs (`logs/YYYY/MM/YYYY-MM-DD.md`) if present — append-only stream, check recent entries first.
33	2. Drifted facts — scan existing memories for claims that may contradict the current codebase (file paths, function names, flags). Verify with Glob/Grep before marking as stale.
34	3. Transcript search — grep JSONL files narrowly for specific context:
35	```bash
36	grep -rn "<narrow term>" ~/.claude/projects/ --include="*.jsonl" \| tail -50
37	```
38	Useful search terms: user corrections ("don't", "no not", "stop"), confirmations ("exactly", "perfect"), role signals ("I'm a", "I own"), decisions ("we're going with", "we decided"), deadlines, external system URLs.
39
40	Signal categories to look for:
41
42	\| Category \| Type \| Trigger phrases \|
43	\|---\|---\|---\|
44	\| User corrections \| feedback \| "no", "don't", "stop doing", "not like that", "actually..." \|
45	\| Confirmed approaches \| feedback \| "yes exactly", "perfect", silent acceptance of unusual choice \|
46	\| Role / expertise \| user \| "I'm a ...", "I've been writing X for Y years", "I own the ..." \|
47	\| Project decisions \| project \| "we're going with", "we decided", deadlines, incidents \|
48	\| External pointers \| reference \| Dashboard URLs, issue trackers, Slack channels, doc links \|
49
50	What NOT to gather (strict exclusions):
51	- Code patterns, architecture, file paths, project structure — derivable from reading code.
52	- Git history or who-changed-what — `git log` / `git blame` are authoritative.
53	- Debugging solutions — the fix is in the code; context is in the commit message.
54	- Anything already in CLAUDE.md files.
55	- Ephemeral task state or current conversation context.
56	- PR lists / activity summaries — save only what was surprising or non-obvious.
57
58	---
59
60	## Phase 3 — Consolidate
61
62	Before making any file modifications, show the user a plan:
63	- List files to be created, updated, merged, or deleted.
64	- Wait for user approval before proceeding.
65
66	After approval, execute:
67
68	### 3a. Repair missing frontmatter
69	For any topic file flagged in Phase 1, add the required YAML frontmatter:
70	```markdown
71	---
72	name: {{memory name}}
73	description: "Consolidates and prunes the memory directory through a 4-phase reflective pass: Orient, Gather, Consolidate, Prune. Run after intensive sessions, when memories feel duplicated, or when MEMORY.md approaches the 200-line limit."
74	type: {{user \| feedback \| project \| reference}}
75	---
76	```
77
78	### 3b. Merge duplicates and near-duplicates
79	If two or more files cover the same topic (e.g., `feed

tranhieuttPublisher · 88 listings

Preview

tranhieutt/software_development_department

# Dream: Memory Consolidation

You are performing a dream — a reflective pass over your memory files. Synthesize what you've learned recently into durable, well-organized memories so that fut

**Memory directory:** find the path from your system prompt — look for the "auto memory" section which says "You have a persistent, file-based memory system at

**Session transcripts:** look for `*.jsonl` files in the project directory inside `~/.claude/projects/` that corresponds to the current working directory. These

Repotranhieutt/software_development_department

LicenseMIT

TypeSkills

CategoryPlanning

UpdatedMay 2026

Repo stars69

First seenJul 22, 2026

People who liked this also viewed

6 picks

Type