personal memory agent
open cogitate agents to sol * and improve all skills
Part 1: Open agent tool access from `sol call *` to `sol *`
- anthropic.py: Bash(sol call *) → Bash(sol *)
- google.py: run_shell_command(sol call) → run_shell_command(sol)
- openai.py: update comment, codex rules pattern expanded to sol
- heartbeat.md: replace phantom commands (sol call health status,
sol call health journal-layout, sol call health agent-runs) with
real commands (sol health, sol health logs, sol muse logs)
Part 2: Review and improve all 10 cogitate skills
- All descriptions updated to Anthropic guide pattern:
[what it does] + [when to use it] + [trigger phrases]
- Added missing trigger phrases to journal, todos, entities,
transcripts, health, onboarding descriptions
- entities: added missing strength, search, intelligence commands
- todos: added missing move command
- health: added troubleshooting section for common diagnostic issues
- body: added agent CLI boundaries section listing infrastructure
commands agents must never call
- Updated test assertions for new restriction patterns
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
+205
-41
+5
-5
.codex/rules/solstone.rules
+5
-5
.codex/rules/solstone.rules
···
1
1
# Execution policy rules for solstone agent CLI commands.
2
-
# Allows "sol call" commands to run outside the sandbox without prompting.
2
+
# Allows "sol" commands to run outside the sandbox without prompting.
3
3
# See: https://developers.openai.com/codex/exec-policy
4
4
5
5
prefix_rule(
6
-
pattern=["sol", "call"],
6
+
pattern=["sol"],
7
7
decision="allow",
8
-
justification="sol call invokes journal query and agent handoff commands",
9
-
match=["sol call todos list", "sol call entities list"],
10
-
not_match=["sol restart-convey"],
8
+
justification="sol invokes journal query, agent, and diagnostic commands",
9
+
match=["sol call todos list", "sol call entities list", "sol health", "sol muse logs"],
10
+
not_match=["solc", "solid"],
11
11
)
+5
-4
apps/calendar/muse/calendar/SKILL.md
+5
-4
apps/calendar/muse/calendar/SKILL.md
···
1
1
---
2
2
name: calendar
3
3
description: >
4
-
Manage calendar events with sol call calendar commands. List, create, update,
5
-
cancel, and move events organized by facet and day.
6
-
TRIGGER: calendar events, scheduling, appointments, meetings, event
7
-
management, moving events between facets.
4
+
Manage calendar events organized by facet and day. List, create, update,
5
+
cancel, and move events including scheduling with participants and times.
6
+
Use when the user mentions calendar events, scheduling, appointments,
7
+
meetings, or wants to create, reschedule, or cancel events.
8
+
TRIGGER: calendar, schedule, appointment, meeting, event, reschedule.
8
9
---
9
10
10
11
# Calendar CLI Skill
+65
-1
apps/entities/muse/entities/SKILL.md
+65
-1
apps/entities/muse/entities/SKILL.md
···
1
1
---
2
2
name: entities
3
-
description: Manage tracked entities with sol call entities commands. List, detect, attach, update, alias, and record observations for people, companies, projects, and tools. Track relationships and knowledge within a facet.
3
+
description: >
4
+
Manage tracked entities for people, companies, projects, and tools within
5
+
facets. Detect, attach, update, alias, search, and record observations.
6
+
Query relationship strength and get full intelligence briefings.
7
+
Use when the user asks about people, contacts, companies, or projects
8
+
tracked in the journal, or wants to add, update, or search entities.
9
+
TRIGGER: entity, person, company, project, relationship, observation,
10
+
who is, contact, knowledge graph, intelligence briefing.
4
11
---
5
12
6
13
# Entities CLI Skill
···
202
209
```bash
203
210
sol call entities observe "Alicia Chen" "Prefers design docs before implementation" -f work --source-day 20260115
204
211
```
212
+
213
+
## strength
214
+
215
+
```bash
216
+
sol call entities strength [FACET] [-n LIMIT]
217
+
```
218
+
219
+
Rank entities by relationship strength score within a facet.
220
+
221
+
- `FACET`: facet name (default: `SOL_FACET` env).
222
+
- `-n, --limit`: max results (default `10`).
223
+
224
+
Use this to find the most significant relationships in a facet.
225
+
226
+
Example:
227
+
228
+
```bash
229
+
sol call entities strength work -n 20
230
+
```
231
+
232
+
## search
233
+
234
+
```bash
235
+
sol call entities search [QUERY] [--type TYPE] [--facet FACET] [--active-days N]
236
+
```
237
+
238
+
Search entities by text, type, facet, or recent activity.
239
+
240
+
- `QUERY`: optional text query.
241
+
- `--type`: filter by entity type (e.g., `Person`, `Company`).
242
+
- `--facet`: filter by facet.
243
+
- `--active-days`: filter to entities active within N days.
244
+
245
+
Examples:
246
+
247
+
```bash
248
+
sol call entities search "Chen"
249
+
sol call entities search --type Person --facet work
250
+
sol call entities search --active-days 7
251
+
```
252
+
253
+
## intelligence
254
+
255
+
```bash
256
+
sol call entities intelligence ENTITY [-f FACET]
257
+
```
258
+
259
+
Get a full intelligence briefing for an entity — relationship history, observations, activity timeline, and cross-facet presence.
260
+
261
+
- `ENTITY`: entity id, name, or alias.
262
+
- `-f, --facet`: facet name (default: `SOL_FACET` env).
263
+
264
+
Example:
265
+
266
+
```bash
267
+
sol call entities intelligence "Alicia Chen" -f work
268
+
```
+28
-1
apps/health/muse/health/SKILL.md
+28
-1
apps/health/muse/health/SKILL.md
···
1
1
---
2
2
name: health
3
-
description: Diagnoses service health and inspects agent runs using sol health and sol muse logs commands. Provides a journal layout reference for navigating logs, agent outputs, and data files at the journal, day, and segment levels.
3
+
description: >
4
+
Diagnose solstone service health, inspect agent run logs, and check system
5
+
status. View service uptimes, crashes, queue depths, recent errors, and
6
+
agent run costs. Includes a journal layout reference for navigating data
7
+
files. Use when the user reports issues, asks about service health, agent
8
+
costs, pipeline status, or when troubleshooting capture gaps and processing
9
+
failures.
10
+
TRIGGER: health, status, is it running, something broke, service down,
11
+
errors, agent runs, costs, logs, pipeline, diagnostics, system check.
4
12
---
5
13
6
14
# Health CLI Skill
···
156
164
| Cortex | Agent JSONL in `agents/<name>/`, outputs in segment/day dirs |
157
165
| Indexer | `indexer/journal.sqlite` |
158
166
| Supervisor | `health/supervisor.log`, service logs in `YYYYMMDD/health/` |
167
+
168
+
## Troubleshooting
169
+
170
+
### `sol health` returns "Connection refused" or times out
171
+
The supervisor is not running. Check if `sol supervisor` is active. The user may need to start solstone with `sol start` or `make dev`.
172
+
173
+
### Agent run shows "error" status in `sol muse logs`
174
+
Run `sol muse log <ID> --full` to see the complete event timeline including the error. Common causes:
175
+
- API key issues (rate limits, expired keys)
176
+
- Prompt too large (context overflow)
177
+
- Network connectivity
178
+
179
+
### Missing segments or capture gaps
180
+
1. Run `sol health` to check observer service status
181
+
2. Run `sol health logs --service sense --since 2h` to check for transcription errors
182
+
3. Check if the stream is active: `sol streams`
183
+
184
+
### High agent costs
185
+
Run `sol muse logs --summary` for aggregated cost view. Filter by agent: `sol muse logs <agent-name> --summary`.
+6
-5
apps/speakers/muse/speakers/SKILL.md
+6
-5
apps/speakers/muse/speakers/SKILL.md
···
1
1
---
2
2
name: speakers
3
3
description: >
4
-
Manage the speaker identification subsystem with sol call speakers commands.
5
-
Check speaker status, detect the owner voice, identify unknown speakers, and
6
-
merge name variants. Curate the speaker library over time.
7
-
TRIGGER: speaker identification, voice recognition, owner detection, speaker
8
-
names, voice matching, unknown voices, speaker merge.
4
+
Manage the speaker identification subsystem. Check speaker status, detect
5
+
the owner voice, identify unknown speakers, merge name variants, and curate
6
+
the speaker library over time. Use when the user asks about voices in
7
+
recordings, wants to identify speakers, or manage voice recognition.
8
+
TRIGGER: speaker, voice, who was talking, identify speaker, owner voice,
9
+
unknown voice, merge speakers, voice recognition, speaker curation.
9
10
---
10
11
11
12
# Speakers CLI Skill
+6
-5
apps/support/muse/support/SKILL.md
+6
-5
apps/support/muse/support/SKILL.md
···
1
1
---
2
2
name: support
3
3
description: >
4
-
File support tickets, search the knowledge base, and give feedback via
5
-
the sol support CLI. Use this skill when the user needs help with solstone,
6
-
wants to report a bug, request a feature, or submit feedback to sol pbc.
7
-
TRIGGER: support tickets, bug reports, feature requests, feedback, help
8
-
requests, knowledge base search, system diagnostics.
4
+
File support tickets, search the knowledge base, and submit feedback to
5
+
sol pbc. Manage open tickets, attach files, check announcements, and run
6
+
local diagnostics. Use when the user needs help with solstone, wants to
7
+
report a bug, request a feature, check for known issues, or give feedback.
8
+
TRIGGER: support, bug report, feature request, feedback, help, knowledge
9
+
base, file a ticket, known issues, announcements, diagnostics.
9
10
---
10
11
11
12
# sol support
+26
-1
apps/todos/muse/todos/SKILL.md
+26
-1
apps/todos/muse/todos/SKILL.md
···
1
1
---
2
2
name: todos
3
-
description: Manage todo checklists using sol call todos commands. List, add, complete, and cancel tasks and action items organized by facet and day. Review upcoming scheduled items.
3
+
description: >
4
+
Manage todo checklists organized by facet and day. List, add, complete,
5
+
cancel, and move tasks and action items. Review upcoming scheduled items.
6
+
Use when the user mentions tasks, to-do items, action items, checklists,
7
+
or reminders, or asks to add, complete, cancel, or review todos.
8
+
TRIGGER: todo, task, action item, checklist, reminder, upcoming items.
4
9
---
5
10
6
11
# Todos CLI Skill
···
126
131
sol call todos upcoming
127
132
sol call todos upcoming -l 50 -f work
128
133
```
134
+
135
+
## move
136
+
137
+
```bash
138
+
sol call todos move LINE_NUMBER --day DAY --from SOURCE --to DEST [--consent]
139
+
```
140
+
141
+
Move an open todo from one facet to another.
142
+
143
+
- `LINE_NUMBER`: 1-based line number from `list` output (positional argument).
144
+
- `--day`: day in `YYYYMMDD` (required).
145
+
- `--from`: source facet name.
146
+
- `--to`: destination facet name.
147
+
- `--consent`: required when called by a proactive agent.
148
+
149
+
Example:
150
+
151
+
```bash
152
+
sol call todos move 3 --day 20260115 --from personal --to work --consent
153
+
```
+8
-1
apps/transcripts/muse/transcripts/SKILL.md
+8
-1
apps/transcripts/muse/transcripts/SKILL.md
···
1
1
---
2
2
name: transcripts
3
-
description: Browse and read transcript content using sol call transcripts commands. Inspect audio recordings, screen captures, and agent summaries. Check recording coverage, list segments, read transcript text with source filtering, or review monthly statistics.
3
+
description: >
4
+
Browse and read transcript content from audio recordings, screen captures,
5
+
and agent summaries. Check recording coverage, list segments, read
6
+
transcript text with source filtering, and review monthly statistics.
7
+
Use when the user asks about recordings, transcripts, what was said,
8
+
conversation content, or wants to review captured audio or screen activity.
9
+
TRIGGER: transcript, recording, audio, what was said, conversation,
10
+
segment, screen capture, recording coverage, monthly stats.
4
11
---
5
12
6
13
# Transcripts CLI Skill
+28
-1
muse/body/SKILL.md
+28
-1
muse/body/SKILL.md
···
1
1
---
2
2
name: body
3
-
description: Development guidelines, project structure, coding standards, testing, and environment for solstone. TRIGGER when contributing code, reviewing PRs, setting up development environment, or asking about project conventions.
3
+
description: >
4
+
Development guidelines, project structure, coding standards, testing, and
5
+
environment for solstone. Use when contributing code, reviewing PRs,
6
+
setting up a development environment, or asking about project conventions,
7
+
architecture, or the CLI command structure.
8
+
TRIGGER: code contribution, development setup, coding standards, project
9
+
structure, architecture, testing, make commands, PR review.
4
10
---
5
11
6
12
# Development Guidelines
···
42
48
make ci # Full CI check (format check + lint + test)
43
49
make dev # Start stack (Ctrl+C to stop)
44
50
```
51
+
52
+
## Agent CLI Boundaries
53
+
54
+
Cogitate agents have access to all `sol` commands. The following infrastructure
55
+
commands must **never** be called by agents — they manage services and data
56
+
pipelines that should only be operated by the supervisor or human operators:
57
+
58
+
- `sol supervisor` / `sol start` — service lifecycle management
59
+
- `sol dream` — full processing pipeline (only heartbeat uses `sol dream --segment` for targeted reprocessing)
60
+
- `sol import` — data injection into journal
61
+
- `sol config` — system configuration changes
62
+
- `sol cortex` — agent process manager (meta-spawning)
63
+
- `sol agents` — direct agent execution
64
+
- `sol callosum` — message bus server
65
+
- `sol observer` / `sol observe-*` — capture services
66
+
- `sol sense` — capture event dispatcher
67
+
- `sol transcribe` / `sol describe` — processing pipelines
68
+
- `sol indexer --reset` — destructive index rebuild (read-only queries via `sol indexer` are fine)
69
+
70
+
Agents should use `sol call` commands for journal interaction and `sol health` /
71
+
`sol muse logs` for diagnostics.
45
72
46
73
## Reference
47
74
+4
-4
muse/heartbeat.md
+4
-4
muse/heartbeat.md
···
23
23
24
24
## Step 1: Check system health
25
25
26
-
Run `sol call health status` and check recent health logs. Note any service
27
-
issues, capture gaps, or pipeline failures.
26
+
Run `sol health` and check recent health logs with `sol health logs --since 1h`.
27
+
Note any service issues, capture gaps, or pipeline failures.
28
28
29
29
If you find issues: update agency.md's `## system` section via
30
30
`echo '...' | sol call sol agency --write`.
31
31
32
32
## Step 2: Check journal quality
33
33
34
-
Run `sol call health journal-layout` and `sol call health agent-runs` for
35
-
the last 3 days. Look for:
34
+
Run `sol muse logs --daily -c 10` to review recent agent runs and
35
+
`sol muse logs --errors -c 10` for recent errors. Look for:
36
36
- Broken segments (transcription failures, missing agent output)
37
37
- Processing gaps (capture with no dream processing)
38
38
- Orphaned entities (zero observations after 7+ days)
+7
-1
muse/journal/SKILL.md
+7
-1
muse/journal/SKILL.md
···
1
1
---
2
2
name: journal
3
-
description: Search and browse journal content using sol call journal commands. Find, query, and look up information across transcripts, insights, events, entities, and todos. Get facet overviews and news feeds.
3
+
description: >
4
+
Search and browse journal content across transcripts, insights, events,
5
+
entities, and todos. Manage facets, get overviews, and read news feeds.
6
+
Use when the user asks to search, find, or look up journal entries, asks
7
+
about a specific day or topic, wants to browse facets, read agent outputs,
8
+
or manage facet organization (create, rename, merge, delete).
9
+
TRIGGER: search, find, look up, browse, journal, facet, news feed, agent output.
4
10
---
5
11
6
12
# Journal CLI Skill
+6
-1
muse/onboarding/SKILL.md
+6
-1
muse/onboarding/SKILL.md
···
1
1
---
2
2
name: onboarding
3
-
description: Set up a new journal — welcome choice, facet creation, and entity seeding.
3
+
description: >
4
+
Guide first-time journal setup including welcome path choice, facet
5
+
creation, and entity seeding. Use when setting up a new journal, during
6
+
initial configuration, or when the user is new and needs orientation.
7
+
TRIGGER: new journal, first time, getting started, setup, onboarding,
8
+
initial configuration, create first facets.
4
9
---
5
10
6
11
# Onboarding CLI Skill
+2
-2
tests/test_anthropic_cli.py
+2
-2
tests/test_anthropic_cli.py
···
394
394
395
395
return MockCLIRunner
396
396
397
-
def test_plan_mode_with_sol_call_allowed(self):
397
+
def test_plan_mode_with_sol_allowed(self):
398
398
provider = _anthropic_provider()
399
399
MockCLIRunner = self._mock_runner()
400
400
with (
···
408
408
)
409
409
cmd = MockCLIRunner.last_instance.cmd
410
410
assert cmd[cmd.index("--permission-mode") + 1] == "plan"
411
-
assert cmd[cmd.index("--allowedTools") + 1] == "Bash(sol call *)"
411
+
assert cmd[cmd.index("--allowedTools") + 1] == "Bash(sol *)"
412
412
413
413
def test_write_mode_bypasses_restrictions(self):
414
414
_assert_write_mode_bypasses_restrictions(self._mock_runner)
+4
-4
tests/test_cogitate_coder.py
+4
-4
tests/test_cogitate_coder.py
···
32
32
@patch("think.providers.anthropic.check_cli_binary")
33
33
@patch("think.providers.anthropic.CLIRunner")
34
34
def test_no_write_restricts_tools(self, mock_runner_cls, mock_check):
35
-
"""Without write flag, --allowedTools restricts to sol call."""
35
+
"""Without write flag, --allowedTools restricts to sol."""
36
36
provider = self._provider()
37
37
mock_instance = AsyncMock()
38
38
mock_instance.run = AsyncMock(return_value="result")
···
44
44
45
45
cmd = mock_runner_cls.call_args.kwargs["cmd"]
46
46
assert "--allowedTools" in cmd
47
-
assert "Bash(sol call *)" in cmd
47
+
assert "Bash(sol *)" in cmd
48
48
49
49
@patch("think.providers.anthropic.check_cli_binary")
50
50
@patch("think.providers.anthropic.CLIRunner")
···
159
159
160
160
@patch("think.providers.google.CLIRunner")
161
161
def test_no_write_restricts_tools(self, mock_runner_cls):
162
-
"""Without write flag, --allowed-tools restricts to sol call."""
162
+
"""Without write flag, --allowed-tools restricts to sol."""
163
163
provider = self._provider()
164
164
mock_instance = AsyncMock()
165
165
mock_instance.run = AsyncMock(return_value="result")
···
171
171
172
172
cmd = mock_runner_cls.call_args.kwargs["cmd"]
173
173
assert "--allowed-tools" in cmd
174
-
assert "run_shell_command(sol call)" in cmd
174
+
assert "run_shell_command(sol)" in cmd
175
175
176
176
@patch("think.providers.google.CLIRunner")
177
177
def test_write_true_grants_full_access(self, mock_runner_cls):
+2
-2
tests/test_google_cli.py
+2
-2
tests/test_google_cli.py
···
329
329
330
330
return MockCLIRunner
331
331
332
-
def test_yolo_mode_with_sol_call_allowed(self):
332
+
def test_yolo_mode_with_sol_allowed(self):
333
333
provider = _google_provider()
334
334
MockCLIRunner = self._mock_runner()
335
335
with patch("think.providers.google.CLIRunner", MockCLIRunner):
···
340
340
)
341
341
cmd = MockCLIRunner.last_instance.cmd
342
342
assert "--yolo" in cmd
343
-
assert cmd[cmd.index("--allowed-tools") + 1] == "run_shell_command(sol call)"
343
+
assert cmd[cmd.index("--allowed-tools") + 1] == "run_shell_command(sol)"
344
344
345
345
def test_write_mode_removes_allowed_tools(self):
346
346
_assert_write_mode_removes_allowed_tools(self._mock_runner)
+1
-1
think/providers/anthropic.py
+1
-1
think/providers/anthropic.py
···
254
254
255
255
# Restrict tool access unless write mode is enabled
256
256
if not config.get("write"):
257
-
cmd.extend(["--allowedTools", "Bash(sol call *)"])
257
+
cmd.extend(["--allowedTools", "Bash(sol *)"])
258
258
259
259
if system_instruction:
260
260
cmd.extend(["--system-prompt", system_instruction])
+1
-1
think/providers/google.py
+1
-1
think/providers/google.py
···
611
611
612
612
# Restrict tool access unless write mode is enabled
613
613
if not config.get("write"):
614
-
cmd.extend(["--allowed-tools", "run_shell_command(sol call)"])
614
+
cmd.extend(["--allowed-tools", "run_shell_command(sol)"])
615
615
616
616
# Resume from previous session if continuing
617
617
if session_id:
+1
-1
think/providers/openai.py
+1
-1
think/providers/openai.py
···
174
174
else:
175
175
prompt_text = prompt_body
176
176
177
-
# Build command — sandbox is read-only; "sol call" commands bypass
177
+
# Build command — sandbox is read-only; "sol" commands bypass
178
178
# the sandbox via exec-policy rules in .codex/rules/solstone.rules
179
179
# Write-enabled agents get full sandbox access
180
180
sandbox = "write" if config.get("write") else "read-only"