personal memory agent
Remove iq option from category config, use providers system
The iq field in category JSON was informational-only and didn't affect
model routing. All category follow-ups already use the providers config
system via observe.describe.* context pattern.
- Remove iq field from meeting.json and messaging.json
- Remove iq handling code from observe/describe.py
- Update docs and tests to reflect simplified schema
- Add note in SCREEN_CATEGORIES.md pointing to providers config
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
+6
-18
+3
-3
docs/SCREEN_CATEGORIES.md
+3
-3
docs/SCREEN_CATEGORIES.md
···
14
14
{
15
15
"description": "One-line description for categorization prompt",
16
16
"followup": true,
17
-
"output": "markdown",
18
-
"iq": "lite"
17
+
"output": "markdown"
19
18
}
20
19
```
21
20
···
24
23
| `description` | Yes | - | Single-line description used in the categorization prompt |
25
24
| `followup` | No | `false` | Whether to run follow-up analysis for this category |
26
25
| `output` | No | `"markdown"` | Response format: `"json"` or `"markdown"` |
27
-
| `iq` | No | `"lite"` | Model tier: `"lite"`, `"flash"`, or `"pro"` |
26
+
27
+
Model selection is handled via the providers configuration in `journal.json`. Each category uses the context pattern `observe.describe.<category>` for routing. See [JOURNAL.md](JOURNAL.md) for details on configuring providers per context.
28
28
29
29
### 2. `<category>.txt` (required if `followup: true`)
30
30
+1
-1
observe/categories/__init__.py
+1
-1
observe/categories/__init__.py
···
4
4
"""Category prompts and formatters for screen description.
5
5
6
6
This package contains:
7
-
- <category>.json: Metadata (description, followup, output, iq)
7
+
- <category>.json: Metadata (description, followup, output)
8
8
- <category>.txt: Vision prompt template (required if followup=true)
9
9
- <category>.py: Optional formatter for rich markdown output
10
10
+1
-2
observe/categories/meeting.json
+1
-2
observe/categories/meeting.json
+1
-2
observe/categories/messaging.json
+1
-2
observe/categories/messaging.json
-6
observe/describe.py
-6
observe/describe.py
···
53
53
- description (required): Single-line description for categorization prompt
54
54
- followup (optional, default: false): Whether to run follow-up analysis
55
55
- output (optional, default: "markdown"): Response format if followup=true
56
-
- iq (optional, default: "lite"): Model tier for follow-up ("lite", "flash", "pro")
57
56
58
57
If followup=true, a matching .txt file contains the follow-up prompt.
59
58
···
87
86
# Store the category context for later resolution
88
87
# The model will be resolved at runtime via generate()
89
88
metadata["context"] = f"observe.describe.{category}"
90
-
91
-
# Store iq for informational purposes (actual tier comes from CONTEXT_DEFAULTS
92
-
# or config override, but category iq can influence CONTEXT_DEFAULTS registration)
93
-
if "iq" not in metadata:
94
-
metadata["iq"] = "lite" # Default tier for categories
95
89
96
90
# Load prompt if followup is enabled
97
91
if metadata["followup"]:
-4
tests/test_describe_config.py
-4
tests/test_describe_config.py
···
37
37
assert "context" in metadata, f"Category {category} missing 'context'"
38
38
assert metadata["context"].startswith("observe.describe.")
39
39
40
-
# Every category should have iq field (tier name)
41
-
assert "iq" in metadata, f"Category {category} missing 'iq'"
42
-
assert metadata["iq"] in ("lite", "flash", "pro")
43
-
44
40
45
41
def test_followup_categories_have_prompts():
46
42
"""Test that categories with followup=true have prompt loaded."""