@goondocks/myco 0.27.17 → 0.27.19

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (8) hide show
  1. package/dist/ui/assets/{index-DM3JaGqM.js → index-GRKrkWAI.js} +68 -68
  2. package/dist/ui/index.html +1 -1
  3. package/package.json +6 -6
  4. package/skills/myco/SKILL.md +29 -18
  5. package/skills/myco/references/cli-usage.md +43 -88
  6. package/skills/myco/references/reconfiguration.md +47 -59
  7. package/skills/myco/references/vault-status.md +21 -40
  8. package/skills/myco/references/wisdom.md +2 -2
package/dist/ui/index.html CHANGED
@@ -5,7 +5,7 @@
5
5
  <link rel="icon" type="image/svg+xml" href="/favicon.svg" />
6
6
  <meta name="viewport" content="width=device-width, initial-scale=1.0" />
7
7
  <title>Myco</title>
8
- <script type="module" crossorigin src="/assets/index-DM3JaGqM.js"></script>
8
+ <script type="module" crossorigin src="/assets/index-GRKrkWAI.js"></script>
9
9
  <link rel="stylesheet" crossorigin href="/assets/index-D80qRqAU.css">
10
10
  </head>
11
11
  <body>
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@goondocks/myco",
3
- "version": "0.27.17",
3
+ "version": "0.27.19",
4
4
  "description": "Collective agent intelligence — Claude Code plugin",
5
5
  "type": "module",
6
6
  "bin": {
@@ -73,11 +73,11 @@
73
73
  "zod": "^4.4.3"
74
74
  },
75
75
  "optionalDependencies": {
76
- "@goondocks/myco-darwin-arm64": "0.27.17",
77
- "@goondocks/myco-darwin-x64": "0.27.17",
78
- "@goondocks/myco-linux-x64": "0.27.17",
79
- "@goondocks/myco-linux-arm64": "0.27.17",
80
- "@goondocks/myco-windows-x64": "0.27.17"
76
+ "@goondocks/myco-darwin-arm64": "0.27.19",
77
+ "@goondocks/myco-darwin-x64": "0.27.19",
78
+ "@goondocks/myco-linux-x64": "0.27.19",
79
+ "@goondocks/myco-linux-arm64": "0.27.19",
80
+ "@goondocks/myco-windows-x64": "0.27.19"
81
81
  },
82
82
  "devDependencies": {
83
83
  "@goondocks/myco-shared": "*",
package/skills/myco/SKILL.md CHANGED
@@ -1,6 +1,6 @@
1
1
  ---
2
2
  name: myco
3
- description: Use when making design decisions, debugging non-obvious issues, encountering gotchas, wondering why code is structured a certain way, or when you need context about prior work on the same feature or component. Myco captures the reasoning, trade-offs, and lessons behind the codebase — things the code itself doesn't show. Also use when the user mentions vault, spores, sessions, team knowledge, institutional memory, or prior decisions.
3
+ description: Use when making design decisions, debugging non-obvious issues, encountering gotchas, wondering why code is structured a certain way, delegating work to another agent, or when you need context about prior work on the same feature or component. Myco captures the reasoning, trade-offs, and lessons behind the codebase — things the code itself doesn't show. Also use when the user mentions vault, spores, sessions, team knowledge, institutional memory, or prior decisions.
4
4
  ---
5
5
 
6
6
  # Myco — Collective Agent Intelligence
@@ -15,6 +15,7 @@ Use Myco tools proactively in these situations — don't wait to be asked:
15
15
  - **When debugging a non-obvious issue** — search for the error message, component name, or symptom. A prior session may have hit the same problem and documented the root cause.
16
16
  - **When wondering why code is structured a certain way** — decisions and trade-offs behind the architecture are captured as spores.
17
17
  - **When continuing work on a feature** — check session history and plan progress for context on what's been done and what's pending.
18
+ - **Before delegating work to another agent, subagent, teammate, worker session, or other spawned process** — refresh the current Cortex instructions and pass them along verbatim so delegated work sees the same Myco guidance.
18
19
  - **After discovering a gotcha, making a key decision, or fixing a tricky bug** — save it so future sessions benefit from the knowledge.
19
20
  - **When starting work on a branch** — context is injected automatically at session start, but you can call `myco_search` and then follow each result's `retrieve` hint for deeper context.
20
21
 
@@ -106,6 +107,14 @@ Canopy map:
106
107
  node .agents/myco-cli.cjs tool call myco_cortex --json --input '{"op":"canopy_map"}'
107
108
  ```
108
109
 
110
+ Current generated instructions:
111
+
112
+ ```bash
113
+ node .agents/myco-cli.cjs tool call myco_cortex --json --input '{"op":"instructions"}'
114
+ ```
115
+
116
+ When delegating work, include the returned instructions verbatim in the delegated prompt alongside the task-specific instructions. Do not assume the returned instructions have a particular heading or section name; they are generated project guidance and may change over time.
117
+
109
118
  Canopy entry returned by search:
110
119
 
111
120
  ```json
@@ -165,6 +174,8 @@ Session association is derived by the daemon; the MCP client does not pass it.
165
174
 
166
175
  List plans, retrieve a single plan's full content by ID, save a plan, or delete one.
167
176
 
177
+ Before creating a new plan or spec, or when existing plans may already cover the work, list plans and read any relevant ones before drafting something new. You do not need to run a plan lookup before every small implementation edit.
178
+
168
179
  ```json
169
180
  { "op": "list", "status": "active" }
170
181
  ```
@@ -306,7 +317,7 @@ The vault should get sharper over time, not just bigger. Every session should le
306
317
 
307
318
  1. `myco_cortex` op `"canopy_map"` for project layout, then `myco_search` with your branch and key files
308
319
  2. `myco_sessions` filtered by branch to see prior session summaries
309
- 3. `myco_plans` to check if there's an active plan
320
+ 3. If you are planning new work or the task may overlap existing specs, use `myco_plans` to check for active plans
310
321
  4. `myco_plans` op `"save"` after generating or revising a plan that should persist in Myco
311
322
 
312
323
  ### After fixing a tricky bug
@@ -323,26 +334,26 @@ The vault should get sharper over time, not just bigger. Every session should le
323
334
 
324
335
  ## Reconfiguration
325
336
 
326
- To change LLM providers, models, or digest settings on an existing vault, see `references/reconfiguration.md`. It covers the exact CLI commands, flag names, and order of operations (setup-llm → restart → rebuild if needed → verify).
337
+ To change embedding settings or Cortex injection behavior on an existing vault, see `references/reconfiguration.md`. It covers the current CLI commands, flag names, and order of operations (setup-llm → restart → rebuild if needed → verify).
327
338
 
328
339
  ## Maintenance
329
340
 
330
341
  For the full CLI reference with all flags, see `references/cli-usage.md`.
331
342
 
332
- All CLI commands use `node` with the CLI script inside the plugin root. Run commands as:
343
+ Prefer the project-resolved launcher from the initialized project root:
333
344
 
334
345
  ```
335
- node <plugin-root>/dist/src/cli.js <command> [args]
346
+ node .agents/myco-cli.cjs <command> [args]
336
347
  ```
337
348
 
338
- Where `<plugin-root>` is the agent's plugin root environment variable (e.g., the value of `CLAUDE_PLUGIN_ROOT` or `CURSOR_PLUGIN_ROOT`).
349
+ That launcher honors project and worktree runtime pins. Use a plugin-root `dist/src/cli.js` path only when you are deliberately operating inside an installed plugin bundle and `.agents/myco-cli.cjs` is unavailable.
339
350
 
340
351
  ### Reprocessing sessions
341
352
 
342
353
  If observations were lost due to a bug, or if you want to re-extract observations with a different LLM, run the `reprocess` command:
343
354
 
344
355
  ```
345
- node <plugin-root>/dist/src/cli.js reprocess
356
+ node .agents/myco-cli.cjs reprocess
346
357
  ```
347
358
 
348
359
  This re-reads all session transcripts, re-extracts observations, and re-indexes everything. Existing spores are preserved — new observations are additive.
@@ -354,9 +365,9 @@ Options:
354
365
  ### Digest management
355
366
 
356
367
  ```
357
- node <plugin-root>/dist/src/cli.js digest # Run incremental digest cycle
358
- node <plugin-root>/dist/src/cli.js digest --tier 3000 # Reprocess a specific tier (clean slate)
359
- node <plugin-root>/dist/src/cli.js digest --full # Reprocess all tiers from scratch
368
+ node .agents/myco-cli.cjs digest # Run incremental digest cycle
369
+ node .agents/myco-cli.cjs digest --tier 3000 # Reprocess a specific tier (clean slate)
370
+ node .agents/myco-cli.cjs digest --full # Reprocess all tiers from scratch
360
371
  ```
361
372
 
362
373
  ### Vault intelligence
@@ -364,8 +375,8 @@ node <plugin-root>/dist/src/cli.js digest --full # Reprocess all tiers fro
364
375
  Supersession happens automatically on every spore write. For vault-wide cleanup, see `references/cli-usage.md` for full flags:
365
376
 
366
377
  ```
367
- node <plugin-root>/dist/src/cli.js agent # Run the intelligence agent
368
- node <plugin-root>/dist/src/cli.js agent --dry-run # Preview without writing
378
+ node .agents/myco-cli.cjs agent # Run the intelligence agent
379
+ node .agents/myco-cli.cjs agent --dry-run # Preview without writing
369
380
  ```
370
381
 
371
382
  For patterns on when to manually supersede or consolidate, see `references/wisdom.md`.
@@ -373,10 +384,10 @@ For patterns on when to manually supersede or consolidate, see `references/wisdo
373
384
  ### Other maintenance commands
374
385
 
375
386
  ```
376
- node <plugin-root>/dist/src/cli.js version # Check plugin version
377
- node <plugin-root>/dist/src/cli.js rebuild # Re-index all records
378
- node <plugin-root>/dist/src/cli.js stats # Check vault health
379
- node <plugin-root>/dist/src/cli.js verify # Test provider connectivity
380
- node <plugin-root>/dist/src/cli.js config get intelligence.llm.model
381
- node <plugin-root>/dist/src/cli.js config set intelligence.llm.model phi4
387
+ node .agents/myco-cli.cjs version # Check plugin version
388
+ node .agents/myco-cli.cjs rebuild # Re-index all records
389
+ node .agents/myco-cli.cjs stats # Check vault health
390
+ node .agents/myco-cli.cjs verify # Test embedding/provider connectivity
391
+ node .agents/myco-cli.cjs setup-llm --show
392
+ node .agents/myco-cli.cjs setup-llm --embedding-provider ollama --embedding-model bge-m3
382
393
  ```
package/skills/myco/references/cli-usage.md CHANGED
@@ -1,10 +1,10 @@
1
1
  # Myco CLI Reference
2
2
 
3
- All CLI commands are invoked as:
3
+ In an initialized project, prefer the project-resolved launcher:
4
4
 
5
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js <command> [flags]
5
+ node .agents/myco-cli.cjs <command> [flags]
6
6
 
7
- Where `${CLAUDE_PLUGIN_ROOT}` resolves to the plugin root directory (set by the agent automatically).
7
+ It honors project and worktree runtime pins. Use a plugin-root `dist/src/cli.js` path only when `.agents/myco-cli.cjs` is unavailable and you are deliberately operating inside an installed plugin bundle.
8
8
 
9
9
  ---
10
10
 
@@ -16,42 +16,30 @@ Initializes a new Myco vault. Skipped automatically if the vault is already init
16
16
 
17
17
  | Flag | Type | Description |
18
18
  |------|------|-------------|
19
- | `--vault <path>` | string | Custom vault directory (supports `~/` expansion) |
20
- | `--llm-provider <name>` | string | `ollama`, `lm-studio`, or `anthropic` |
21
- | `--llm-model <name>` | string | Model name |
22
- | `--llm-url <url>` | string | Provider base URL |
23
- | `--embedding-provider <name>` | string | `ollama` or `lm-studio` |
19
+ | `--project <path>` | string | Project root to initialize |
20
+ | `--grove <name\|id>` | string | Grove to bind this project to |
21
+ | `--worktree` | boolean | Bootstrap hook files in a git worktree |
22
+ | `--non-interactive` | boolean | Run without prompts |
23
+ | `--embedding-provider <name>` | string | Embedding provider for new vaults |
24
24
  | `--embedding-model <name>` | string | Embedding model name |
25
25
  | `--embedding-url <url>` | string | Embedding provider base URL |
26
- | `--user <name>` | string | Username for team-enabled vault |
27
- | `--team` | boolean | Enable team collaboration mode |
28
- | `--tiers <csv>` | string | Comma-separated digest tier list (e.g., `1500,3000,5000`) |
29
- | `--inject-tier <number>` | number | Tier to auto-inject at session start |
30
- | `--context-window <number>` | number | Context window size for digest operations |
31
26
 
32
27
  **Example:**
33
28
 
34
29
  ```sh
35
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js init \
36
- --llm-provider ollama \
37
- --llm-model qwen2.5-coder:14b \
30
+ node .agents/myco-cli.cjs init \
38
31
  --embedding-provider ollama \
39
32
  --embedding-model bge-m3
40
33
  ```
41
34
 
42
35
  ---
43
36
 
44
- ### `setup-llm` — Change LLM and embedding provider settings
37
+ ### `setup-llm` — Change embedding provider settings
45
38
 
46
- Reconfigures intelligence backend without reinitializing the vault.
39
+ Reconfigures Grove-tier embedding settings without reinitializing the vault. LLM configuration is managed by the Myco agent harness; `setup-llm` ignores legacy `--llm-*` flags.
47
40
 
48
41
  | Flag | Type | Description |
49
42
  |------|------|-------------|
50
- | `--llm-provider <name>` | string | Provider name (`ollama`, `lm-studio`, `anthropic`) |
51
- | `--llm-model <name>` | string | Model name |
52
- | `--llm-url <url>` | string | Provider base URL |
53
- | `--llm-context-window <number>` | number | Context window in tokens |
54
- | `--llm-max-tokens <number>` | number | Max output tokens |
55
43
  | `--embedding-provider <name>` | string | Embedding provider name |
56
44
  | `--embedding-model <name>` | string | Embedding model name |
57
45
  | `--embedding-url <url>` | string | Embedding base URL |
@@ -63,52 +51,19 @@ Note: changing the embedding model requires running `rebuild` afterward to re-em
63
51
 
64
52
  ```sh
65
53
  # Show current settings
66
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js setup-llm --show
54
+ node .agents/myco-cli.cjs setup-llm --show
67
55
 
68
- # Switch to Anthropic
69
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js setup-llm \
70
- --llm-provider anthropic \
71
- --llm-model claude-opus-4-5
56
+ # Change embedding model
57
+ node .agents/myco-cli.cjs setup-llm \
58
+ --embedding-provider ollama \
59
+ --embedding-model bge-m3
72
60
  ```
73
61
 
74
62
  ---
75
63
 
76
- ### `setup-digest` — Configure digest and capture settings
77
-
78
- Controls the continuous reasoning system: metabolism timing, tier configuration, and per-operation token budgets.
64
+ ### `setup-digest` — Deprecated
79
65
 
80
- | Flag | Type | Description |
81
- |------|------|-------------|
82
- | `--enabled <true\|false>` | boolean | Enable or disable digest |
83
- | `--tiers <csv>` | string | Comma-separated token tier list |
84
- | `--inject-tier <number\|null>` | number\|null | Auto-inject tier at session start (`null` to disable) |
85
- | `--provider <name\|null>` | string\|null | Digest-specific provider (`null` = inherit from main LLM) |
86
- | `--model <name\|null>` | string\|null | Digest-specific model (`null` = inherit) |
87
- | `--base-url <url\|null>` | string\|null | Digest provider base URL (`null` = inherit) |
88
- | `--context-window <number>` | number | Context window for digest |
89
- | `--keep-alive <duration>` | string | Keep model loaded (Ollama only, e.g., `30m`) |
90
- | `--gpu-kv-cache <true\|false>` | boolean | GPU KV cache offload (LM Studio only) |
91
- | `--active-interval <seconds>` | number | Metabolism active processing interval |
92
- | `--dormancy-threshold <seconds>` | number | Time before entering dormancy |
93
- | `--max-notes <number>` | number | Max substrate notes per digest cycle |
94
- | `--extraction-tokens <number>` | number | Max tokens for spore extraction |
95
- | `--summary-tokens <number>` | number | Max tokens for session summaries |
96
- | `--title-tokens <number>` | number | Max tokens for session titles |
97
- | `--classification-tokens <number>` | number | Max tokens for artifact classification |
98
- | `--show` | boolean | Display current settings and exit |
99
-
100
- **Example:**
101
-
102
- ```sh
103
- # Show current digest config
104
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js setup-digest --show
105
-
106
- # Use a faster model just for digest, with shorter active interval
107
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js setup-digest \
108
- --model qwen2.5:7b \
109
- --active-interval 120 \
110
- --dormancy-threshold 600
111
- ```
66
+ Digest and Cortex configuration now live in the normal scoped config/UI surfaces. `setup-digest` prints a deprecation message and points users back to `setup-llm` for embedding settings.
112
67
 
113
68
  ---
114
69
 
@@ -120,16 +75,16 @@ Direct access to vault config via dot-path notation. Values are parsed as JSON f
120
75
 
121
76
  ```sh
122
77
  # Read a value
123
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js config get intelligence.llm.model
78
+ node .agents/myco-cli.cjs config get cortex.instructions.inject_on_session_start
124
79
 
125
80
  # Write a value
126
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js config set intelligence.llm.model qwen2.5-coder:14b
81
+ node .agents/myco-cli.cjs config set cortex.instructions.inject_on_session_start true
127
82
 
128
83
  # Write a non-string value (parsed as JSON)
129
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js config set digest.enabled true
84
+ node .agents/myco-cli.cjs config set cortex.spores.max_per_prompt 3
130
85
  ```
131
86
 
132
- Restart the daemon after changes: `node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js restart`
87
+ Restart the daemon after changes that affect runtime behavior: `node .agents/myco-cli.cjs restart`
133
88
 
134
89
  ---
135
90
 
@@ -155,12 +110,12 @@ Output is JSON:
155
110
 
156
111
  ### `verify` — Test LLM and embedding connectivity
157
112
 
158
- Sends a test prompt to the LLM and a test embed to the embedding provider. Exits 0 if both pass, 1 if either fails.
113
+ Sends a test embed to the configured embedding provider. Exits 0 if it passes, 1 if it fails. LLM configuration is managed by the Myco agent harness.
159
114
 
160
115
  No flags.
161
116
 
162
117
  ```sh
163
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js verify
118
+ node .agents/myco-cli.cjs verify
164
119
  ```
165
120
 
166
121
  ---
@@ -172,7 +127,7 @@ Shows session/spore/plan counts, spore type breakdown, vector count, and daemon
172
127
  No flags.
173
128
 
174
129
  ```sh
175
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js stats
130
+ node .agents/myco-cli.cjs stats
176
131
  ```
177
132
 
178
133
  Typical output:
@@ -205,13 +160,13 @@ Components: `processor`, `embeddings`, `hooks`, `lifecycle`, `daemon`, `lineage`
205
160
 
206
161
  ```sh
207
162
  # Show last 20 lines
208
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js logs -n 20
163
+ node .agents/myco-cli.cjs logs -n 20
209
164
 
210
165
  # Follow errors from the processor
211
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js logs -f -l error -c processor
166
+ node .agents/myco-cli.cjs logs -f -l error -c processor
212
167
 
213
168
  # Show logs from a specific window
214
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js logs \
169
+ node .agents/myco-cli.cjs logs \
215
170
  --since 2025-01-01T10:00:00Z \
216
171
  --until 2025-01-01T11:00:00Z
217
172
  ```
@@ -225,7 +180,7 @@ node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js logs \
225
180
  Runs semantic search (primary) with FTS fallback across sessions, spores, and plans.
226
181
 
227
182
  ```sh
228
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js search "why did we choose sqlite over postgres"
183
+ node .agents/myco-cli.cjs search "why did we choose sqlite over postgres"
229
184
  ```
230
185
 
231
186
  ---
@@ -235,7 +190,7 @@ node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js search "why did we choose sqlite over
235
190
  Shows all results with similarity scores and no threshold filtering. Useful for tuning embedding thresholds.
236
191
 
237
192
  ```sh
238
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js vectors "session lifecycle hooks"
193
+ node .agents/myco-cli.cjs vectors "session lifecycle hooks"
239
194
  ```
240
195
 
241
196
  ---
@@ -249,10 +204,10 @@ node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js vectors "session lifecycle hooks"
249
204
 
250
205
  ```sh
251
206
  # Show latest session
252
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js session
207
+ node .agents/myco-cli.cjs session
253
208
 
254
209
  # Show a specific session by partial ID
255
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js session ac5220
210
+ node .agents/myco-cli.cjs session ac5220
256
211
  ```
257
212
 
258
213
  ---
@@ -266,7 +221,7 @@ Sends SIGTERM to the running daemon, waits for it to exit, and spawns a fresh in
266
221
  No flags.
267
222
 
268
223
  ```sh
269
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js restart
224
+ node .agents/myco-cli.cjs restart
270
225
  ```
271
226
 
272
227
  Run this after any daemon code changes to pick up new behavior.
@@ -280,7 +235,7 @@ Re-indexes all records. Superseded and archived spores are skipped.
280
235
  No flags.
281
236
 
282
237
  ```sh
283
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js rebuild
238
+ node .agents/myco-cli.cjs rebuild
284
239
  ```
285
240
 
286
241
  Run this after changing the embedding model (via `setup-llm`) to regenerate all embeddings with the new model.
@@ -302,13 +257,13 @@ When `--tier` or `--full` is used, the cycle reads all records (ignoring the las
302
257
 
303
258
  ```sh
304
259
  # Run an incremental cycle (same as what the metabolism timer does)
305
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js digest
260
+ node .agents/myco-cli.cjs digest
306
261
 
307
262
  # Reprocess tier 3000 from scratch
308
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js digest --tier 3000
263
+ node .agents/myco-cli.cjs digest --tier 3000
309
264
 
310
265
  # Full rebuild of all tiers
311
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js digest --full
266
+ node .agents/myco-cli.cjs digest --full
312
267
  ```
313
268
 
314
269
  ---
@@ -327,10 +282,10 @@ Runs the intelligence agent to process unprocessed session data, extract observa
327
282
 
328
283
  ```sh
329
284
  # Run the default intelligence task
330
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js agent
285
+ node .agents/myco-cli.cjs agent
331
286
 
332
287
  # Preview what would be changed
333
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js agent --dry-run
288
+ node .agents/myco-cli.cjs agent --dry-run
334
289
  ```
335
290
 
336
291
  Note: `--dry-run` still runs LLM calls (to evaluate) — it just skips the writes. Use it to review before running on a vault for the first time.
@@ -350,13 +305,13 @@ Re-reads session transcripts, re-extracts observations with the current LLM, and
350
305
 
351
306
  ```sh
352
307
  # Reprocess all sessions
353
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js reprocess
308
+ node .agents/myco-cli.cjs reprocess
354
309
 
355
310
  # Reprocess one session
356
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js reprocess --session ac5220
311
+ node .agents/myco-cli.cjs reprocess --session ac5220
357
312
 
358
313
  # Re-index without re-extracting
359
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js reprocess --index-only
314
+ node .agents/myco-cli.cjs reprocess --index-only
360
315
  ```
361
316
 
362
317
  ---
@@ -366,7 +321,7 @@ node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js reprocess --index-only
366
321
  ### `version` — Show plugin version
367
322
 
368
323
  ```sh
369
- node ${CLAUDE_PLUGIN_ROOT}/dist/src/cli.js version
324
+ node .agents/myco-cli.cjs version
370
325
  ```
371
326
 
372
327
  Also available as `--version` or `-v`.
package/skills/myco/references/reconfiguration.md CHANGED
@@ -1,92 +1,80 @@
1
1
  # Reconfiguration
2
2
 
3
- Workflows for changing LLM providers, models, and digest settings on an existing vault. **Use the AskUserQuestion tool** to ask which settings to change — do not guess.
3
+ Workflows for changing embedding settings and verifying an existing Myco project. Ask the user which setting they want changed; do not guess.
4
4
 
5
- ## Changing LLM or Embedding Provider/Model
5
+ ## Current Model
6
6
 
7
- Follow this exact order:
7
+ - LLM execution for the Myco agent is managed by the agent harness (currently Claude Agent SDK in dogfood paths), not by `setup-llm`.
8
+ - Embedding configuration is user-configurable and lives at the Grove tier.
9
+ - `setup-llm` configures embedding settings only. Legacy `--llm-*` flags are ignored with a notice.
10
+ - Digest and Cortex settings live in scoped config/UI surfaces; `setup-digest` is deprecated.
8
11
 
9
- ```bash
10
- # 1. Detect what's available
11
- node <plugin-root>/dist/src/cli.js detect-providers
12
-
13
- # 2. Apply the change (use the correct --llm- or --embedding- prefixed flags)
14
- node <plugin-root>/dist/src/cli.js setup-llm \
15
- --llm-provider <provider> --llm-model <model> \
16
- --embedding-provider <provider> --embedding-model <model>
17
-
18
- # 3. ALWAYS restart daemon after any config change
19
- node <plugin-root>/dist/src/cli.js restart
20
-
21
- # 4. Only rebuild if the EMBEDDING model changed (not needed for LLM-only changes)
22
- node <plugin-root>/dist/src/cli.js rebuild
23
-
24
- # 5. Verify connectivity
25
- node <plugin-root>/dist/src/cli.js verify
26
- ```
12
+ ## Changing Embedding Provider or Model
27
13
 
28
- ### Critical Flags
14
+ Follow this order:
29
15
 
30
- The `setup-llm` command uses `--llm-provider`, `--llm-model`, `--embedding-provider`, `--embedding-model` — NOT `--provider` or `--model`. Only pass flags for settings the user explicitly wants to change.
31
-
32
- ### Order Matters
16
+ ```bash
17
+ # 1. Detect local providers when choosing a local embedding backend
18
+ node .agents/myco-cli.cjs detect-providers
33
19
 
34
- 1. `setup-llm` writes config
35
- 2. `restart` loads the new config into the daemon
36
- 3. `rebuild` re-embeds with the new embedding model (skip if embedding didn't change)
37
- 4. `verify` confirms everything works
20
+ # 2. Show current Grove-tier embedding settings
21
+ node .agents/myco-cli.cjs setup-llm --show
38
22
 
39
- ### Embedding Model Warning
23
+ # 3. Apply only the requested embedding changes
24
+ node .agents/myco-cli.cjs setup-llm \
25
+ --embedding-provider <provider> \
26
+ --embedding-model <model>
40
27
 
41
- If the embedding model changed, tell the user: "Changing the embedding model requires a full vector index rebuild. This may take a few minutes."
28
+ # 4. Restart daemon so runtime reads the new config
29
+ node .agents/myco-cli.cjs restart
42
30
 
43
- ## Changing Digest Settings
31
+ # 5. Rebuild vectors if the embedding model changed
32
+ node .agents/myco-cli.cjs rebuild
44
33
 
45
- ```bash
46
- node <plugin-root>/dist/src/cli.js setup-digest \
47
- --context-window <number> --inject-tier <tier>
48
- node <plugin-root>/dist/src/cli.js restart
34
+ # 6. Verify embedding connectivity
35
+ node .agents/myco-cli.cjs verify
49
36
  ```
50
37
 
51
- For all available `setup-digest` flags (tiers, provider override, metabolism tuning, token budgets), see `cli-usage.md`.
38
+ If only the embedding URL changed and the provider/model stayed the same, restart and verify are enough. If the embedding model changed, tell the user: "Changing the embedding model requires a full vector index rebuild. This may take a few minutes."
52
39
 
53
40
  ## Viewing Current Settings
54
41
 
55
42
  ```bash
56
- node <plugin-root>/dist/src/cli.js setup-llm --show
57
- node <plugin-root>/dist/src/cli.js setup-digest --show
43
+ node .agents/myco-cli.cjs setup-llm --show
44
+ node .agents/myco-cli.cjs config get cortex.instructions.inject_on_session_start
45
+ node .agents/myco-cli.cjs config get cortex.digest.inject_on_session_start
58
46
  ```
59
47
 
48
+ Prefer `setup-llm --show` for embedding settings instead of raw config reads; it resolves the Grove tier correctly.
49
+
60
50
  ## Common Scenarios
61
51
 
62
- ### "Change my LLM model" (same provider)
52
+ ### "Change my embedding model"
63
53
 
64
54
  ```bash
65
- node <plugin-root>/dist/src/cli.js setup-llm --llm-model qwen3.5:35b
66
- node <plugin-root>/dist/src/cli.js restart
67
- node <plugin-root>/dist/src/cli.js verify
55
+ node .agents/myco-cli.cjs setup-llm --embedding-model bge-m3
56
+ node .agents/myco-cli.cjs restart
57
+ node .agents/myco-cli.cjs rebuild
58
+ node .agents/myco-cli.cjs verify
68
59
  ```
69
60
 
70
- No rebuild needed — embedding didn't change.
71
-
72
- ### "Switch from Ollama to LM Studio"
61
+ ### "Switch embedding provider"
73
62
 
74
63
  ```bash
75
- node <plugin-root>/dist/src/cli.js detect-providers
76
- node <plugin-root>/dist/src/cli.js setup-llm \
77
- --llm-provider lm-studio --llm-model "qwen/qwen3.5-35b-a3b"
78
- node <plugin-root>/dist/src/cli.js restart
79
- node <plugin-root>/dist/src/cli.js verify
64
+ node .agents/myco-cli.cjs detect-providers
65
+ node .agents/myco-cli.cjs setup-llm \
66
+ --embedding-provider ollama \
67
+ --embedding-model bge-m3
68
+ node .agents/myco-cli.cjs restart
69
+ node .agents/myco-cli.cjs rebuild
70
+ node .agents/myco-cli.cjs verify
80
71
  ```
81
72
 
82
- ### "Change everything" (provider, model, and embedding)
73
+ ### "Enable or disable Cortex instruction injection"
83
74
 
84
75
  ```bash
85
- node <plugin-root>/dist/src/cli.js detect-providers
86
- node <plugin-root>/dist/src/cli.js setup-llm \
87
- --llm-provider ollama --llm-model qwen3.5:35b \
88
- --embedding-provider ollama --embedding-model bge-m3
89
- node <plugin-root>/dist/src/cli.js restart
90
- node <plugin-root>/dist/src/cli.js rebuild
91
- node <plugin-root>/dist/src/cli.js verify
76
+ node .agents/myco-cli.cjs config set cortex.instructions.inject_on_session_start true
77
+ node .agents/myco-cli.cjs restart
92
78
  ```
79
+
80
+ Use the Settings UI when possible for scoped config edits; it shows whether a value is machine, Grove, project, or local.