cm-deep-search

Name: cm-deep-search
Author: tody-agent

47stars23forks

How to install cm-deep-search?

tody-agent/codymaster/cm-deep-search

$npx -y skills add tody-agent/codymaster --skill cm-deep-search

Installs into the current project.

Summary

Optional power-up — detects oversized codebases/docs and suggests tobi/qmd for local semantic search. Bridges cm-continuity (working memory) with long-term document retrieval. Zero-config detection, non-intrusive suggestion.

Files · 1

View on GitHub

SKILL.md

1	# Deep Search — Semantic Memory Power-Up
2
3	> When your project outgrows AI's context window, bring the search engine to your docs.
4	> Optional integration with [tobi/qmd](https://github.com/tobi/qmd) — BM25 + Vector + LLM re-ranking, 100% local.
5
6	## When to Trigger
7
8	This skill is NOT invoked directly. It is triggered automatically by other skills when they detect an oversized project.
9
10	### Detection Thresholds
11
12	During codebase scan (Phase 1a of `cm-brainstorm-idea`, Step 2 of `cm-dockit`, etc.), check:
13
14	```
15	TRIGGER if ANY of these are true:
16	→ docs/ folder contains >50 markdown files
17	→ Project has >200 source files total
18	→ User mentions "meeting notes", "historical PRDs", "old specs"
19	→ User asks "find that file that talked about X from before"
20	→ cm-dockit just generated >30 doc files
21	```
22
23	### What to Say (Non-Intrusive)
24
25	When threshold is met, suggest naturally — DO NOT block or force:
26
27	```markdown
28	💡 Pro Tip: Deep Search
29
30	This project has [X doc files / Y source files] — quite large for AI to read directly.
31	You can install [qmd](https://github.com/tobi/qmd) to create semantic search
32	across all your documentation, helping AI find the right context faster.
33
34	Quick install:
35	\`\`\`bash
36	npm install -g @tobilu/qmd
37	qmd collection add ./docs --name project-docs
38	qmd context add qmd://project-docs "Project documentation for [project-name]"
39	qmd embed
40	\`\`\`
41
42	Then AI can search using: `qmd query "your question"`
43	```
44
45	---
46
47	## Setup Guide (when user agrees to install)
48
49	### Step 1: Install
50
51	```bash
52	# Node.js
53	npm install -g @tobilu/qmd
54
55	# Or Bun
56	bun install -g @tobilu/qmd
57	```
58
59	### Step 2: Index project docs
60
61	```bash
62	# Add collections
63	qmd collection add ./docs --name docs
64	qmd collection add ./src --name source --mask "*/.{ts,tsx,js,jsx,py,go,rs}"
65
66	# Add context (helps AI understand each collection's purpose)
67	qmd context add qmd://docs "Technical documentation for [project-name]"
68	qmd context add qmd://source "Source code for [project-name]"
69
70	# Create vector embeddings
71	qmd embed
72	```
73
74	### Step 3: Setup MCP Server (for Codex/Cursor/Antigravity)
75
76	Add to MCP config:
77
78	```json
79	{
80	"mcpServers": {
81	"qmd": {
82	"command": "qmd",
83	"args": ["mcp"]
84	}
85	}
86	}
87	```
88
89	Or run HTTP mode for shared server:
90
91	```bash
92	qmd mcp --http --daemon
93	```
94
95	### Step 4: Verify
96
97	```bash
98	# Check index
99	qmd status
100
101	# Test search
102	qmd query "authentication flow"
103	```
104
105	---
106
107	## Usage with CodyMaster Skills
108
109	### With `cm-brainstorm-idea` (Phase 1: DISCOVER)
110
111	When AI needs to understand the full picture of a large project:
112
113	```bash
114	# Find all docs related to the topic being brainstormed
115	qmd query "user authentication redesign" --json -n 10
116
117	# Get full content of important docs
118	qmd get "docs/architecture.md" --full
119	```
120
121	### With `cm-planning` (Phase A: Brainstorm)
122
123	When searching for specs, PRDs, or past decisions related to the feature being planned:
124
125	```bash
126	qmd query "payment integration decisions" --files --min-score 0.4
127	```
128
129	### With `cm-dockit` (Post-generation)
130
131	After `cm-dockit` finishes generating docs, index them so AI can search from any session:
132
133	```bash
134	qmd collection add ./docs --name project-knowledge
135	qmd embed
136	```
137
138	### With `cm-continuity` (Tier 4: External Memory)
139
140	`cm-continuity` manages working memory (500 words). `qmd` extends it with long-term semantic search:
141
142	```
143	Tier 1: Sensory Memory → temporary variables in session (not persisted)
144	Tier 2: Working Memory → CONTINUITY.md (~500 words)
145	Tier 3: Long-Term Memory → learnings.json, decisions.json
146	Tier 4: External Semantic → qmd (optional, text search for large docs)
147	Tier 5: Structural Code → CodeGraph (optional, AST graph for code — see cm-codeintell)
148	```
149
150	> qmd finds text across docs/code. CodeGraph finds symbols, call graphs, and impact.
151	> They complement each other — use both for maximum intelligence on large projects.
152
153	---
154
155	## 🛑 Staleness Prevention
156
157	The biggest risk of Semantic Search is stale index / new source. If AI reads outdated docs and generates incorrect code, the consequences are severe.
158
159	CodyMaster handles this with 3 mechanisms:
160
161	### 1. The "Post-Execution" Sync
162	Whenever AI completes a task that changes/creates a large number of files (e.g., `cm-dockit` generates docs, `cm-execution` refactors source code):
163	```bash
164	# This runs quickly because qmd only embeds changed files (incremental)
165	qmd embed
166	```
167	> AI Rule: If the project has qmd enabled, AI must automatically run `qmd embed` via terminal before finishing a task.
168
169	### 2. The "Pre-Flight" Check
170	Before starting `cm-brainstorm-idea` or `cm-planning` on a project using qmd, AI calls the MCP tool to perform a health check:
171	```json
172	// AI auto-runs this MCP tool
173	{
174	"name": "status",
175	"arguments": {}
176	}
177	`

People who liked this also viewed

6 picks

Type

cm-deep-search

47stars23forks

View on GitHub

How to install cm-deep-search?

tody-agent/codymaster/cm-deep-search

$npx -y skills add tody-agent/codymaster --skill cm-deep-search

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Deep Search — Semantic Memory Power-Up
2
3	> When your project outgrows AI's context window, bring the search engine to your docs.
4	> Optional integration with [tobi/qmd](https://github.com/tobi/qmd) — BM25 + Vector + LLM re-ranking, 100% local.
5
6	## When to Trigger
7
8	This skill is NOT invoked directly. It is triggered automatically by other skills when they detect an oversized project.
9
10	### Detection Thresholds
11
12	During codebase scan (Phase 1a of `cm-brainstorm-idea`, Step 2 of `cm-dockit`, etc.), check:
13
14	```
15	TRIGGER if ANY of these are true:
16	→ docs/ folder contains >50 markdown files
17	→ Project has >200 source files total
18	→ User mentions "meeting notes", "historical PRDs", "old specs"
19	→ User asks "find that file that talked about X from before"
20	→ cm-dockit just generated >30 doc files
21	```
22
23	### What to Say (Non-Intrusive)
24
25	When threshold is met, suggest naturally — DO NOT block or force:
26
27	```markdown
28	💡 Pro Tip: Deep Search
29
30	This project has [X doc files / Y source files] — quite large for AI to read directly.
31	You can install [qmd](https://github.com/tobi/qmd) to create semantic search
32	across all your documentation, helping AI find the right context faster.
33
34	Quick install:
35	\`\`\`bash
36	npm install -g @tobilu/qmd
37	qmd collection add ./docs --name project-docs
38	qmd context add qmd://project-docs "Project documentation for [project-name]"
39	qmd embed
40	\`\`\`
41
42	Then AI can search using: `qmd query "your question"`
43	```
44
45	---
46
47	## Setup Guide (when user agrees to install)
48
49	### Step 1: Install
50
51	```bash
52	# Node.js
53	npm install -g @tobilu/qmd
54
55	# Or Bun
56	bun install -g @tobilu/qmd
57	```
58
59	### Step 2: Index project docs
60
61	```bash
62	# Add collections
63	qmd collection add ./docs --name docs
64	qmd collection add ./src --name source --mask "*/.{ts,tsx,js,jsx,py,go,rs}"
65
66	# Add context (helps AI understand each collection's purpose)
67	qmd context add qmd://docs "Technical documentation for [project-name]"
68	qmd context add qmd://source "Source code for [project-name]"
69
70	# Create vector embeddings
71	qmd embed
72	```
73
74	### Step 3: Setup MCP Server (for Codex/Cursor/Antigravity)
75
76	Add to MCP config:
77
78	```json
79	{
80	"mcpServers": {
81	"qmd": {
82	"command": "qmd",
83	"args": ["mcp"]
84	}
85	}
86	}
87	```
88
89	Or run HTTP mode for shared server:
90
91	```bash
92	qmd mcp --http --daemon
93	```
94
95	### Step 4: Verify
96
97	```bash
98	# Check index
99	qmd status
100
101	# Test search
102	qmd query "authentication flow"
103	```
104
105	---
106
107	## Usage with CodyMaster Skills
108
109	### With `cm-brainstorm-idea` (Phase 1: DISCOVER)
110
111	When AI needs to understand the full picture of a large project:
112
113	```bash
114	# Find all docs related to the topic being brainstormed
115	qmd query "user authentication redesign" --json -n 10
116
117	# Get full content of important docs
118	qmd get "docs/architecture.md" --full
119	```
120
121	### With `cm-planning` (Phase A: Brainstorm)
122
123	When searching for specs, PRDs, or past decisions related to the feature being planned:
124
125	```bash
126	qmd query "payment integration decisions" --files --min-score 0.4
127	```
128
129	### With `cm-dockit` (Post-generation)
130
131	After `cm-dockit` finishes generating docs, index them so AI can search from any session:
132
133	```bash
134	qmd collection add ./docs --name project-knowledge
135	qmd embed
136	```
137
138	### With `cm-continuity` (Tier 4: External Memory)
139
140	`cm-continuity` manages working memory (500 words). `qmd` extends it with long-term semantic search:
141
142	```
143	Tier 1: Sensory Memory → temporary variables in session (not persisted)
144	Tier 2: Working Memory → CONTINUITY.md (~500 words)
145	Tier 3: Long-Term Memory → learnings.json, decisions.json
146	Tier 4: External Semantic → qmd (optional, text search for large docs)
147	Tier 5: Structural Code → CodeGraph (optional, AST graph for code — see cm-codeintell)
148	```
149
150	> qmd finds text across docs/code. CodeGraph finds symbols, call graphs, and impact.
151	> They complement each other — use both for maximum intelligence on large projects.
152
153	---
154
155	## 🛑 Staleness Prevention
156
157	The biggest risk of Semantic Search is stale index / new source. If AI reads outdated docs and generates incorrect code, the consequences are severe.
158
159	CodyMaster handles this with 3 mechanisms:
160
161	### 1. The "Post-Execution" Sync
162	Whenever AI completes a task that changes/creates a large number of files (e.g., `cm-dockit` generates docs, `cm-execution` refactors source code):
163	```bash
164	# This runs quickly because qmd only embeds changed files (incremental)
165	qmd embed
166	```
167	> AI Rule: If the project has qmd enabled, AI must automatically run `qmd embed` via terminal before finishing a task.
168
169	### 2. The "Pre-Flight" Check
170	Before starting `cm-brainstorm-idea` or `cm-planning` on a project using qmd, AI calls the MCP tool to perform a health check:
171	```json
172	// AI auto-runs this MCP tool
173	{
174	"name": "status",
175	"arguments": {}
176	}
177	`

People who liked this also viewed

6 picks

Type