read-memories

Name: read-memories
Author: opengeos

23stars2forks

How to install read-memories?

opengeos/geoai-skills/read-memories

$npx -y skills add opengeos/geoai-skills --skill read-memories

Installs into the current project.

Summary

Search past Claude Code session logs to recover context from previous conversations. Finds past decisions, data paths, CRS info, model configurations, and unresolved work. Works across all projects or scoped to the current one.

Files · 1

View on GitHub

SKILL.md

1	Use this to recover your own context from past sessions. Do NOT narrate the
2	process to the user -- silently run the query, absorb the results, and continue
3	with enriched context.
4
5	`$0` is the keyword to search for.
6	Pass `--here` as an argument to scope the search to the current project only.
7	Pass `--geo` as an argument to additionally extract geospatial-specific context
8	(EPSG codes, bounding boxes, CRS info, spatial file paths, model names).
9
10	## Step 1 -- Set the search path
11
12	```bash
13	ALL_PROJECTS="$HOME/.claude/projects//.jsonl"
14	CURRENT_PROJECT="$HOME/.claude/projects/$(echo "$PWD" \| sed 's\|[/_]\|-\|g')/*.jsonl"
15	```
16
17	Use `$CURRENT_PROJECT` if any argument is `--here`, otherwise use `$ALL_PROJECTS`.
18	Store the chosen glob in `SEARCH_PATH`.
19
20	Check whether the `--geo` flag is present.
21
22	## Step 2 -- Query with Python
23
24	Run the following Python script via `python3 -c "..."`, substituting
25	`<SEARCH_PATH>` and `<KEYWORD>` with the resolved values. Escape any
26	single quotes in `<KEYWORD>` before embedding it.
27
28	```bash
29	python3 -c "
30	import json, glob, os
31
32	SEARCH_PATH = '<SEARCH_PATH>'
33	KEYWORD = '<KEYWORD>'.lower()
34	LIMIT = 40
35
36	files = sorted(glob.glob(os.path.expanduser(SEARCH_PATH)))
37	results = []
38
39	for fpath in files:
40	parts = fpath.split('/')
41	try:
42	proj_idx = parts.index('projects') + 1
43	project = parts[proj_idx] if proj_idx < len(parts) else 'unknown'
44	except ValueError:
45	project = 'unknown'
46
47	with open(fpath, 'r', errors='replace') as f:
48	for line in f:
49	try:
50	obj = json.loads(line)
51	except (json.JSONDecodeError, ValueError):
52	continue
53
54	msg = obj.get('message')
55	if not isinstance(msg, dict):
56	continue
57	role = msg.get('role')
58	if role not in ('user', 'assistant'):
59	continue
60
61	content = msg.get('content', '')
62	if isinstance(content, list):
63	text = ' '.join(
64	c.get('text', '')
65	for c in content
66	if isinstance(c, dict) and 'text' in c
67	)
68	elif isinstance(content, str):
69	text = content
70	else:
71	continue
72
73	if KEYWORD not in text.lower():
74	continue
75
76	ts = obj.get('timestamp', '')
77	snippet = text[:1500]
78	results.append({
79	'project': project,
80	'ts': ts[:16].replace('T', ' ') if ts else '',
81	'role': role,
82	'content': snippet,
83	})
84
85	if len(results) >= LIMIT:
86	break
87	if len(results) >= LIMIT:
88	break
89
90	print(f'Found {len(results)} results (limit {LIMIT})')
91	print('---')
92	for i, r in enumerate(results):
93	print(f'[{i+1}] project={r[\"project\"]} ts={r[\"ts\"]} role={r[\"role\"]}')
94	print(r['content'][:800])
95	print('---')
96	"
97	```
98
99	## Step 3 -- Handle large result sets
100
101	If Step 2 reports exactly 40 results (limit hit), the keyword is common. Run a
102	counting pass to understand the scope:
103
104	```bash
105	python3 -c "
106	import json, glob, os
107
108	SEARCH_PATH = '<SEARCH_PATH>'
109	KEYWORD = '<KEYWORD>'.lower()
110
111	files = sorted(glob.glob(os.path.expanduser(SEARCH_PATH)))
112	total = 0
113	by_project = {}
114
115	for fpath in files:
116	parts = fpath.split('/')
117	try:
118	proj_idx = parts.index('projects') + 1
119	project = parts[proj_idx] if proj_idx < len(parts) else 'unknown'
120	except ValueError:
121	project = 'unknown'
122
123	with open(fpath, 'r', errors='replace') as f:
124	for line in f:
125	try:
126	obj = json.loads(line)
127	except (json.JSONDecodeError, ValueError):
128	continue
129	msg = obj.get('message')
130	if not isinstance(msg, dict):
131	continue
132	role = msg.get('role')
133	if role not in ('user', 'assistant'):
134	continue
135	content = msg.get('content', '')
136	if isinstance(content, list):
137	text = ' '.join(
138	c.get('text', '')
139	for c in content
140	if isinstance(c, dict) and 'text' in c
141	)
142	elif isinstance(content, str):
143	text = content
144	else:
145	continue
146	if KEYWORD in text.lower():
147	total += 1
148	by_project[project] = by_project.get(project, 0) + 1
149
150	print(f'Total matches: {total}')
151	for proj, cnt in sorted(by_project.items(), key=lambda x: -x[1]):
152	print(f' {proj}: {cnt}')
153	"
154	```
155
156	Use this breakdown to decide whether to:
157	-

People who liked this also viewed

6 picks

Type

read-memories

23stars2forks

View on GitHub

How to install read-memories?

opengeos/geoai-skills/read-memories

$npx -y skills add opengeos/geoai-skills --skill read-memories

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	Use this to recover your own context from past sessions. Do NOT narrate the
2	process to the user -- silently run the query, absorb the results, and continue
3	with enriched context.
4
5	`$0` is the keyword to search for.
6	Pass `--here` as an argument to scope the search to the current project only.
7	Pass `--geo` as an argument to additionally extract geospatial-specific context
8	(EPSG codes, bounding boxes, CRS info, spatial file paths, model names).
9
10	## Step 1 -- Set the search path
11
12	```bash
13	ALL_PROJECTS="$HOME/.claude/projects//.jsonl"
14	CURRENT_PROJECT="$HOME/.claude/projects/$(echo "$PWD" \| sed 's\|[/_]\|-\|g')/*.jsonl"
15	```
16
17	Use `$CURRENT_PROJECT` if any argument is `--here`, otherwise use `$ALL_PROJECTS`.
18	Store the chosen glob in `SEARCH_PATH`.
19
20	Check whether the `--geo` flag is present.
21
22	## Step 2 -- Query with Python
23
24	Run the following Python script via `python3 -c "..."`, substituting
25	`<SEARCH_PATH>` and `<KEYWORD>` with the resolved values. Escape any
26	single quotes in `<KEYWORD>` before embedding it.
27
28	```bash
29	python3 -c "
30	import json, glob, os
31
32	SEARCH_PATH = '<SEARCH_PATH>'
33	KEYWORD = '<KEYWORD>'.lower()
34	LIMIT = 40
35
36	files = sorted(glob.glob(os.path.expanduser(SEARCH_PATH)))
37	results = []
38
39	for fpath in files:
40	parts = fpath.split('/')
41	try:
42	proj_idx = parts.index('projects') + 1
43	project = parts[proj_idx] if proj_idx < len(parts) else 'unknown'
44	except ValueError:
45	project = 'unknown'
46
47	with open(fpath, 'r', errors='replace') as f:
48	for line in f:
49	try:
50	obj = json.loads(line)
51	except (json.JSONDecodeError, ValueError):
52	continue
53
54	msg = obj.get('message')
55	if not isinstance(msg, dict):
56	continue
57	role = msg.get('role')
58	if role not in ('user', 'assistant'):
59	continue
60
61	content = msg.get('content', '')
62	if isinstance(content, list):
63	text = ' '.join(
64	c.get('text', '')
65	for c in content
66	if isinstance(c, dict) and 'text' in c
67	)
68	elif isinstance(content, str):
69	text = content
70	else:
71	continue
72
73	if KEYWORD not in text.lower():
74	continue
75
76	ts = obj.get('timestamp', '')
77	snippet = text[:1500]
78	results.append({
79	'project': project,
80	'ts': ts[:16].replace('T', ' ') if ts else '',
81	'role': role,
82	'content': snippet,
83	})
84
85	if len(results) >= LIMIT:
86	break
87	if len(results) >= LIMIT:
88	break
89
90	print(f'Found {len(results)} results (limit {LIMIT})')
91	print('---')
92	for i, r in enumerate(results):
93	print(f'[{i+1}] project={r[\"project\"]} ts={r[\"ts\"]} role={r[\"role\"]}')
94	print(r['content'][:800])
95	print('---')
96	"
97	```
98
99	## Step 3 -- Handle large result sets
100
101	If Step 2 reports exactly 40 results (limit hit), the keyword is common. Run a
102	counting pass to understand the scope:
103
104	```bash
105	python3 -c "
106	import json, glob, os
107
108	SEARCH_PATH = '<SEARCH_PATH>'
109	KEYWORD = '<KEYWORD>'.lower()
110
111	files = sorted(glob.glob(os.path.expanduser(SEARCH_PATH)))
112	total = 0
113	by_project = {}
114
115	for fpath in files:
116	parts = fpath.split('/')
117	try:
118	proj_idx = parts.index('projects') + 1
119	project = parts[proj_idx] if proj_idx < len(parts) else 'unknown'
120	except ValueError:
121	project = 'unknown'
122
123	with open(fpath, 'r', errors='replace') as f:
124	for line in f:
125	try:
126	obj = json.loads(line)
127	except (json.JSONDecodeError, ValueError):
128	continue
129	msg = obj.get('message')
130	if not isinstance(msg, dict):
131	continue
132	role = msg.get('role')
133	if role not in ('user', 'assistant'):
134	continue
135	content = msg.get('content', '')
136	if isinstance(content, list):
137	text = ' '.join(
138	c.get('text', '')
139	for c in content
140	if isinstance(c, dict) and 'text' in c
141	)
142	elif isinstance(content, str):
143	text = content
144	else:
145	continue
146	if KEYWORD in text.lower():
147	total += 1
148	by_project[project] = by_project.get(project, 0) + 1
149
150	print(f'Total matches: {total}')
151	for proj, cnt in sorted(by_project.items(), key=lambda x: -x[1]):
152	print(f' {proj}: {cnt}')
153	"
154	```
155
156	Use this breakdown to decide whether to:
157	-

People who liked this also viewed

6 picks

Type