geneseed 5.1.0 → 5.1.1
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.
- package/CHANGELOG.md +11 -0
- package/SHIPPED.md +2 -2
- package/harness.config.json +1 -1
- package/js/README.md +1 -1
- package/js/cli-table.json +3 -2
- package/js/hosts/memory-files.mjs +3 -0
- package/js/maintain/memory.mjs +27 -2
- package/package.json +1 -1
- package/src/agents/memory-curator.md +4 -4
- package/src/memory/README.md +1 -1
- package/src/skills/geneseed.md +1 -1
- package/src/skills/rule/consolidate.md +1 -1
- package/src/skills/rule.md +1 -1
- package/adapters/claude-code/skills/geneseed/SKILL.md +0 -76
package/CHANGELOG.md
CHANGED
|
@@ -8,6 +8,17 @@ label. For the capability ↔ spec map, see [SHIPPED.md](SHIPPED.md).
|
|
|
8
8
|
|
|
9
9
|
## [Unreleased]
|
|
10
10
|
|
|
11
|
+
## [5.1.1] — 2026-11-10
|
|
12
|
+
|
|
13
|
+
### Added
|
|
14
|
+
- **`geneseed memory consolidate`** rebuilds the memory index by hand: `MEMORY.md` and `topics/`, from the fact files on disk. It prints `+N indexed, -N pruned` and each pair of facts that share a description, and never merges them. It takes `--memory <dir>` like `list` and `rm`; a fact name, `--archived` or `--purge` is refused with exit 2.
|
|
15
|
+
|
|
16
|
+
### Fixed
|
|
17
|
+
- **The memory instructions name a command that runs.** The rule skill, the `memory-curator` agent and the memory README told the agent to run `geneseed learn --consolidate`, which the CLI refuses: `learn` is a hook verb. They now say `geneseed memory consolidate`. The `learn` hook still rebuilds the index after every fact it writes.
|
|
18
|
+
|
|
19
|
+
### Removed
|
|
20
|
+
- **`adapters/claude-code/skills/geneseed/SKILL.md`**, a stale hand-written copy of the `geneseed` skill that no install used. It told the agent to run `geneseed learn` and `geneseed context` by hand. The skill every install gets is rendered from `src/skills/geneseed.md`.
|
|
21
|
+
|
|
11
22
|
## [5.1.0] — 2026-11-09
|
|
12
23
|
|
|
13
24
|
### Added
|
package/SHIPPED.md
CHANGED
|
@@ -20,7 +20,7 @@ historical references, not files distributed with the repo.
|
|
|
20
20
|
> **On version numbers.** [`harness.config.json`](harness.config.json)'s `version` is a
|
|
21
21
|
> human-readable label only. The canonical identity of an installed harness is the
|
|
22
22
|
> **source fingerprint** stamped in `.geneseed-version` (see `geneseed version`), not the
|
|
23
|
-
> string. Current label: **5.1.
|
|
23
|
+
> string. Current label: **5.1.1**.
|
|
24
24
|
|
|
25
25
|
## Capabilities
|
|
26
26
|
|
|
@@ -39,7 +39,7 @@ Every row below is present in the tree today; "Spec" names the design record.
|
|
|
39
39
|
| **OpenCode adapter** | Native agents/skills mapping + plugins (context, learn, guard, workflow, notify, ponytail, activity); global-emit link integrity; JSONC config target | `2026-06-07-opencode-adapter-features` · `2026-06-07-global-emit-link-integrity` · `2026-06-12-opencode-jsonc-target` · `2026-06-16-ponytail-minimal-code-mode` |
|
|
40
40
|
| **Host hooks, per dialect** | Every host gets the hook surface its runtime offers, spoken in that runtime's own verdict dialect via `--host`: Claude Code asks (`permissionDecision`), OpenClaude asks exactly as Claude does (a Claude Code fork, under `~/.openclaude` / `.openclaude/`), Bob refuses by exit code 2 (five Claude-named events, nested `~/.bob/settings/settings.json`, no subagent event), OpenCode's guard throws. Laws I/IV are the hard tier everywhere; the consent rules ask where a host can ask and warn where it cannot. `tool-gate` fuses the two gates for hosts that run one command per event; `PreCompact` runs `learn` before Claude compacts; every block or ask lands in `notebook/gates.jsonl` on every host | [bob](adapters/bob/README.md) · [claude-code](adapters/claude-code/README.md) · [candidate hosts](docs/candidate-hosts.md) |
|
|
41
41
|
| **Workflow primitive** | Saved, code-driven subagent orchestration (`workflow` tool + runtime); shipped scripts include `dispatch` — decompose a goal, route each subtask to its owning agent, converge | `2026-06-09-opencode-workflow-primitive` |
|
|
42
|
-
| **Per-agent memory** | Each capability agent keeps durable lessons in `memory/agents/<name>.md`: read at dispatch, written back mechanically by `learn` on session/subagent end (OpenCode plugin + Claude `SubagentStop`; Bob has no subagent event); `learn
|
|
42
|
+
| **Per-agent memory** | Each capability agent keeps durable lessons in `memory/agents/<name>.md`: read at dispatch, written back mechanically by `learn` on session/subagent end (OpenCode plugin + Claude `SubagentStop`; Bob has no subagent event); the `learn` hook rebuilds the index after every write, and `geneseed memory consolidate` rebuilds it by hand | `2026-07-04-orchestration-memory-upgrade` |
|
|
43
43
|
| **Wiki integration** | Machine-wide knowledge base (`geneseed-wiki.jsonc`): eager/lazy loading, conventions, protected folders enforced at the tool boundary | `2026-06-11-wiki-knowledge-base` |
|
|
44
44
|
| **Collaboration layer** | Negotiated postures (peer/mentor/expert/assistant/artisan, chosen at setup, inlined in AGENT.md), the bidirectional Pact, epistemically-typed memory (`force:` + Bridge rule), and a seed-once `PROFILE.md` identity with a web editor tab and a `profile` skill that interviews the user to draft it | `2026-07-12-collaboration-layer` |
|
|
45
45
|
| **Ownership** | The developer keeps ownership of production code: a Credo section (the code belongs to whoever answers for it), the `ownership` skill (setup from the code and a short interview, `feature`/`file`/`function` granularity, interfaces and sensitive zones always a checkpoint, announce-don't-ask), a `## Ownership` section in the committed root `AGENTS.md` (`mode: on\|off`), and a first-write gate on every host that stops the session's first write once in an application repo without one. learn-mode always comes first; `develop`, the `developer` agent and Foreman hold the checkpoints. Off per repo (`mode: off`), per install (`--no-ownership-gate`), per session ("proceed without") | `2026-10-26-ownership` |
|
package/harness.config.json
CHANGED
|
@@ -2,5 +2,5 @@
|
|
|
2
2
|
"_comment": "version is a human-readable release label only; the canonical identity of an installed harness is the source fingerprint stamped in .geneseed-version (see `harness version`), not this string.",
|
|
3
3
|
"theme": "neutral",
|
|
4
4
|
"product": "Geneseed",
|
|
5
|
-
"version": "5.1.
|
|
5
|
+
"version": "5.1.1"
|
|
6
6
|
}
|
package/js/README.md
CHANGED
|
@@ -232,7 +232,7 @@ Verbs that change an install that already exists, as opposed to rendering one.
|
|
|
232
232
|
| `setup.mjs` | The install wizard. Its prompts go through `js/lib/prompt.mjs`'s stdin readers |
|
|
233
233
|
| `uninstall.mjs` | Removes or disables an install: manifest-driven reversal, plus stash deactivate/reactivate |
|
|
234
234
|
| `migrate.mjs` | `geneseed migrate` — all-or-nothing re-emit moving installs onto the npm install shape |
|
|
235
|
-
| `memory.mjs` | `geneseed memory list|rm` — CLI-side fact removal (archive by default, `--purge` deletes), twin of the web delete endpoint |
|
|
235
|
+
| `memory.mjs` | `geneseed memory list|rm|consolidate` — CLI-side fact removal (archive by default, `--purge` deletes), twin of the web delete endpoint; `consolidate` rebuilds the index by hand, twin of the hook's `learn --consolidate` |
|
|
236
236
|
|
|
237
237
|
**Before editing:** `update.mjs` is the one module that MUST spawn rather than import — the code on
|
|
238
238
|
disk is replaced mid-run, so an in-process call would validate the *old* source and then render it.
|
package/js/cli-table.json
CHANGED
|
@@ -1306,7 +1306,7 @@
|
|
|
1306
1306
|
},
|
|
1307
1307
|
{
|
|
1308
1308
|
"name": "memory",
|
|
1309
|
-
"help": "manage the memory store: list the facts the agent reads,
|
|
1309
|
+
"help": "manage the memory store: list the facts the agent reads, remove one by name (archived into memory/archive/; --purge deletes), or rebuild the index by hand with consolidate (MEMORY.md and topics/, from the fact files on disk). A global install's store is the shared tier, ~/.config/geneseed/memory; a project install keeps its own",
|
|
1310
1310
|
"description": "",
|
|
1311
1311
|
"positionals": [
|
|
1312
1312
|
{
|
|
@@ -1316,7 +1316,8 @@
|
|
|
1316
1316
|
"help": "",
|
|
1317
1317
|
"choices": [
|
|
1318
1318
|
"list",
|
|
1319
|
-
"rm"
|
|
1319
|
+
"rm",
|
|
1320
|
+
"consolidate"
|
|
1320
1321
|
],
|
|
1321
1322
|
"default": null,
|
|
1322
1323
|
"required": true,
|
|
@@ -466,6 +466,9 @@ export function archiveStore(store, move = renameSync,
|
|
|
466
466
|
* duplicate descriptions. It REPORTS them and never merges: merging is a judgement, made by
|
|
467
467
|
* the model at learn time (`writeMemories`'s `update`) or by the user, never by this rebuild. `added` and
|
|
468
468
|
* `pruned` compare against the pointers of BOTH levels, so a migration counts as neither.
|
|
469
|
+
*
|
|
470
|
+
* TWO DOORS, ONE REBUILD. The hook's `learn --consolidate` (`hooks-learn.mjs`) and the CLI's
|
|
471
|
+
* `geneseed memory consolidate` (`js/maintain/memory.mjs`) both land here, never on a copy.
|
|
469
472
|
*/
|
|
470
473
|
export function consolidateMemory(memDir) {
|
|
471
474
|
const { facts, old } = rebuildIndex(memDir, true);
|
package/js/maintain/memory.mjs
CHANGED
|
@@ -3,7 +3,7 @@
|
|
|
3
3
|
*
|
|
4
4
|
* WHY THIS VERB IS NEW RATHER THAN PORTED. The reference had no `memory` subcommand: a fact
|
|
5
5
|
* could be written by the `learn` hook and deleted from the web console, and there was no
|
|
6
|
-
* third way. So the semantics
|
|
6
|
+
* third way. So the semantics of `list` and `rm` are the DELETE ENDPOINT's, deliberately and field for
|
|
7
7
|
* field — the same reserved names, the same "a bare slug, never a path" rule, the same
|
|
8
8
|
* index line dropped from MEMORY.md after the file goes — because two ways to remove a fact
|
|
9
9
|
* that disagree about what a fact is would be worse than one way.
|
|
@@ -32,11 +32,23 @@
|
|
|
32
32
|
* learn uses), so a slip of the keyboard or a click in the console is reversible; only
|
|
33
33
|
* `--purge` unlinks. `--archived` points `list` and `rm` at `archive/` itself, and a file
|
|
34
34
|
* there can only be purged — it has nowhere further to go.
|
|
35
|
+
*
|
|
36
|
+
* `consolidate` IS THE BY-HAND DOOR TO THE INDEX REBUILD, AND IT HOLDS NO REBUILD OF ITS OWN.
|
|
37
|
+
* The other door is `geneseed-hook learn --consolidate`, which the specs once told a reader
|
|
38
|
+
* to type as `geneseed learn --consolidate` — a command this binary refuses by design: `learn`
|
|
39
|
+
* spawns the model CLI, so the whole verb lives in the hook binary (ADR 0002), and hook verbs
|
|
40
|
+
* are the host's to run, never a person's. The rebuild itself is pure filesystem work, so it
|
|
41
|
+
* is legal here. Both doors call `consolidateMemory` inside `recordOwnWrites` — the
|
|
42
|
+
* `memoryDelete` precedent: one implementation, two entries. What differs is the SURFACE, in
|
|
43
|
+
* three places: the report line carries the `[memory]` prefix on stdout where the hook prints
|
|
44
|
+
* `[learn]` on stderr; with no store this exits 1 with the verb's shared refusal where the
|
|
45
|
+
* hook exits 0; and a fact name, `--archived` or `--purge` is a usage refusal (exit 2),
|
|
46
|
+
* because a rebuild takes none of them and silently ignoring `--purge` would be a lie.
|
|
35
47
|
*/
|
|
36
48
|
import { readdirSync, unlinkSync } from 'node:fs';
|
|
37
49
|
import path from 'node:path';
|
|
38
50
|
|
|
39
|
-
import { RESERVED_STEMS, archiveFact, memoryDropIndex } from '../hosts/memory-files.mjs';
|
|
51
|
+
import { RESERVED_STEMS, archiveFact, consolidateMemory, memoryDropIndex } from '../hosts/memory-files.mjs';
|
|
40
52
|
import { recordOwnWrites } from '../hosts/store-watch.mjs';
|
|
41
53
|
import { resolveMemoryDir } from '../hosts/hosts.mjs';
|
|
42
54
|
import { printOut, printErr, isFile } from '../lib/fs.mjs';
|
|
@@ -98,6 +110,19 @@ export function cmdMemory(args) {
|
|
|
98
110
|
+ 'install.\n');
|
|
99
111
|
return 1;
|
|
100
112
|
}
|
|
113
|
+
if (args.action === 'consolidate') {
|
|
114
|
+
if (args.name || args.archived || args.purge) {
|
|
115
|
+
printErr('[memory] consolidate rebuilds the whole index: it takes no fact name, '
|
|
116
|
+
+ '--archived or --purge.\n');
|
|
117
|
+
return 2;
|
|
118
|
+
}
|
|
119
|
+
const report = recordOwnWrites(path.dirname(dir), () => consolidateMemory(dir));
|
|
120
|
+
printOut(`[memory] consolidated ${dir}: +${report.added.length} indexed, `
|
|
121
|
+
+ `-${report.pruned.length} pruned, `
|
|
122
|
+
+ `${report.duplicates.length} duplicate description(s)\n`);
|
|
123
|
+
for (const [a, b] of report.duplicates) printOut(` duplicate: ${a} <-> ${b}\n`);
|
|
124
|
+
return 0;
|
|
125
|
+
}
|
|
101
126
|
const where = args.archived ? path.join(dir, 'archive') : dir;
|
|
102
127
|
if (args.action === 'list') {
|
|
103
128
|
const facts = memoryFacts(where);
|
package/package.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"//": "Every field here is argued and gated in tests/unit/package_manifest.test.mjs. `files`, `bin`, `exports` and `engines` each have a test that fails when reality drifts from this file; `version` mirrors harness.config.json and is gated too. Read that docblock before editing this one.",
|
|
3
3
|
"name": "geneseed",
|
|
4
|
-
"version": "5.1.
|
|
4
|
+
"version": "5.1.1",
|
|
5
5
|
"description": "A portable, theme-able harness you implant once and use everywhere to grow a disciplined AI coding agent.",
|
|
6
6
|
"keywords": [
|
|
7
7
|
"ai",
|
|
@@ -6,7 +6,7 @@
|
|
|
6
6
|
- The {{MEMORY}} store needs tidying in bulk. The session context says so in one
|
|
7
7
|
line when a topic holds over 40 facts or a `misc` topic exists; or the user asks
|
|
8
8
|
to tidy, dedupe or re-topic their memory.
|
|
9
|
-
- `geneseed
|
|
9
|
+
- `geneseed memory consolidate` reported duplicate descriptions that need a merge.
|
|
10
10
|
|
|
11
11
|
## When NOT to dispatch
|
|
12
12
|
- A handful of facts to review — the rule {{SKILL}}'s consolidation flow does
|
|
@@ -19,8 +19,8 @@
|
|
|
19
19
|
scope: the whole store, or named topics only.
|
|
20
20
|
|
|
21
21
|
## Allowed tools
|
|
22
|
-
- **Read + write to the {{MEMORY}} store only.** May run `geneseed memory`
|
|
23
|
-
`
|
|
22
|
+
- **Read + write to the {{MEMORY}} store only.** May run `geneseed memory`
|
|
23
|
+
(`list`, `rm`, `consolidate`). A write into the store meets no prompt at the tool boundary
|
|
24
24
|
(memory is written freely), but bulk curation reworks facts the user already keeps, so it still
|
|
25
25
|
shows each topic's plan and applies it with the user's consent.
|
|
26
26
|
<!-- bash: allow -->
|
|
@@ -51,7 +51,7 @@
|
|
|
51
51
|
and run `geneseed memory rm <name> --archived --purge` only for the codes the user
|
|
52
52
|
approves. A fact holding a secret, live or archived, is proposed for `--purge` on sight
|
|
53
53
|
({{LAW:sealed-secrets}}).
|
|
54
|
-
7. Finish with `geneseed
|
|
54
|
+
7. Finish with `geneseed memory consolidate --memory <the store's path>`, so both
|
|
55
55
|
index levels match the files.
|
|
56
56
|
|
|
57
57
|
## Output contract
|
package/src/memory/README.md
CHANGED
|
@@ -111,7 +111,7 @@ facts the store holds:
|
|
|
111
111
|
|
|
112
112
|
A fact's topic is its `topic:`, else its `type:`, else `misc`, normalised to
|
|
113
113
|
lowercase `[a-z0-9-]`. Both levels are generated from the fact files:
|
|
114
|
-
`geneseed
|
|
114
|
+
`geneseed memory consolidate --memory <this dir>` rebuilds them (refreshing each
|
|
115
115
|
hook from its `description`), and the `learn` hook rebuilds them after every fact it
|
|
116
116
|
writes, updates or archives. Only top-level fact files are indexed; `archive/` and
|
|
117
117
|
`agents/` never are. A build moves a flat, one-line-per-fact `MEMORY.md` from an older install to
|
package/src/skills/geneseed.md
CHANGED
|
@@ -7,7 +7,7 @@
|
|
|
7
7
|
|
|
8
8
|
## Procedure
|
|
9
9
|
1. **Locate.** Run `geneseed status --json` first: `installs` lists every install on this machine — host, scope, root, state, theme, footprint, posture, mode, packs, excluded rules — and `rebuild`, the command that reproduces each one exactly. `geneseed` not on PATH → ask the user before using `npx geneseed` (it downloads the package), and prefix every command below the same way. Still nothing → look for a host root file: `~/.config/opencode/AGENT.md` or `.opencode/AGENT.md` (OpenCode), `~/.claude/CLAUDE.md` or `.claude/` (Claude Code), `.openclaude/` (OpenClaude), `AGENTS.md` beside `.bob/` (Bob). None → say so; never install to "fix" it.
|
|
10
|
-
2. **Read state** with the read-only verbs, cheapest first ({{DOCTRINE:context-economy}}): `geneseed status --json`, `geneseed version`, `geneseed catalog` (optionally `agents|skills|laws|ontology|doctrines`), `geneseed doctor`, `geneseed diff`, `geneseed mcp`, `geneseed memory list`, `geneseed exclude list`.
|
|
10
|
+
2. **Read state** with the read-only verbs, cheapest first ({{DOCTRINE:context-economy}}): `geneseed status --json`, `geneseed version`, `geneseed catalog` (optionally `agents|skills|laws|ontology|doctrines`), `geneseed doctor`, `geneseed diff`, `geneseed mcp`, `geneseed memory list`, `geneseed memory consolidate` (rebuilds only the generated index, from the fact files), `geneseed exclude list`.
|
|
11
11
|
3. **Read a piece** from disk, not from recall ({{LAW:verify-before-asserting}}): `<root>/agents/<name>.md`, `<root>/skills/<name>/SKILL.md`, `<root>/memory/`, `<root>/notebook/`. The {{ONTOLOGY}}, the {{LAWS}} and the active {{DOCTRINES}} are numbered `## N. …` sections of the install's root file.
|
|
12
12
|
4. **Change an install only when the user asked for that change** ({{LAW:one-intent-one-act}}). Take that install's `rebuild` command from step 1, edit only the flag the user named (add it if the command omits it — `--posture`, `--mode` and `--trust` are left out at their defaults) — `--theme`, `--footprint lean|full`, `--posture`, `--mode direct|foreman`, `--trust prudent|balanced|aggressive`, `--doctrines craft,rigor|none`, `--exclude-rules "process 7"|none` — dry-run it first by running the same flags through `geneseed validate`, show the command, and run it once the user agrees. Never rebuild from a bare `geneseed build`: every omitted flag resets that setting to the generator default. Confirm with `geneseed status`.
|
|
13
13
|
5. **Other writes also need the user's word:** `geneseed memory rm <name>` (archives; `--purge` deletes for good), `geneseed exclude add|remove <path>`, `geneseed web start|stop`, `geneseed uninstall --target <root> --yes`. Personal rules and the profile belong to their own {{SKILLS}}; do not edit `user-rules.md` or `PROFILE.md` from here.
|
|
@@ -25,7 +25,7 @@ instead: it runs this flow topic by topic in its own context and also assigns to
|
|
|
25
25
|
archiving would keep it on disk, so propose deleting it outright, on sight
|
|
26
26
|
({{LAW:sealed-secrets}}).
|
|
27
27
|
5. **Rebuild the index** with
|
|
28
|
-
`geneseed
|
|
28
|
+
`geneseed memory consolidate --memory <the {{PATH_MEMORY}}/ dir's path>`: one pointer line per
|
|
29
29
|
surviving file in its `topics/<topic>.md`, and the `MEMORY.md` map recounted. Each
|
|
30
30
|
hook is the fact's `description`, so sharpen a weak hook there — it should say why a
|
|
31
31
|
future session would open the fact.
|
package/src/skills/rule.md
CHANGED
|
@@ -44,7 +44,7 @@ Then take exactly one branch.
|
|
|
44
44
|
3. Infer the `type` (`user` | `feedback` | `project` | `reference`) and draft the file
|
|
45
45
|
per the {{MEMORY}} convention.
|
|
46
46
|
4. Write the file, with no prompt, and rebuild the index
|
|
47
|
-
(`geneseed
|
|
47
|
+
(`geneseed memory consolidate --memory <the {{PATH_MEMORY}}/ dir's path>`) so its line
|
|
48
48
|
lands in `topics/<topic>.md` and the `MEMORY.md` map recounts. Then say in one line
|
|
49
49
|
what you wrote and where, so a wrong inference is visible at once; the web console
|
|
50
50
|
shows it to the user later too.
|
|
@@ -1,76 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
name: geneseed
|
|
3
|
-
description: "Interact with a deployed Geneseed harness from any repo — locate the install, read its ontology/rules/doctrines/agents/skills/memory/notebook by name, run status/doctor/diff, distill memory with learn, and manage the local web UI. Use when the user mentions Geneseed, AGENT.md, the harness, an agent operating system, or asks you to inspect, refresh, or read from the deployed bundle."
|
|
4
|
-
---
|
|
5
|
-
|
|
6
|
-
# geneseed — agent skill
|
|
7
|
-
|
|
8
|
-
Geneseed is a portable AI-agent harness deployed once and used everywhere. This skill lets you talk to whichever deployment is in scope from any repo: read the canonical files (the three constitution tiers, agents, skills, memory, notebook), probe state (status, version, doctor, diff), distill durable memory, and drive the local web UI. The harness is theme-independent in structure — directory and section names are always plain English — so the paths below are stable across themes.
|
|
9
|
-
|
|
10
|
-
## Find the deployed harness before doing anything
|
|
11
|
-
|
|
12
|
-
Pick the first that resolves; remember the path for the rest of the session.
|
|
13
|
-
|
|
14
|
-
1. `command -v geneseed` — if the launcher is on PATH, use it. The CLI knows its own install and is the source of truth for `status`, `version`, `doctor`, etc.
|
|
15
|
-
2. `$GENESEED_HARNESS` — if set, that directory is the deployment root.
|
|
16
|
-
3. `~/.config/opencode/AGENT.md` — the default OpenCode global install.
|
|
17
|
-
4. `./.opencode/AGENT.md` (walking up from `pwd`) — a per-repo OpenCode install.
|
|
18
|
-
5. Any `AGENT.md` in the repo root — a plain bundle install.
|
|
19
|
-
|
|
20
|
-
If none resolve, the user does not have Geneseed deployed here — say so and stop. Do not run `setup`, `build`, or `upgrade` on your own to "fix" this; that is a user decision.
|
|
21
|
-
|
|
22
|
-
A live harness root contains: `AGENT.md`, `ontology/universal.md`, `laws/universal.md`, `doctrines/<pack>.md`, `agents/<name>.md`, `skills/<name>/SKILL.md`, `memory/MEMORY.md`, `notebook/NOTEBOOK.md`, `themes/`, and (on OpenCode) `plugins/`, `workflows/`, `opencode.jsonc`, `geneseed-wiki.jsonc`.
|
|
23
|
-
|
|
24
|
-
The constitution is **three tiers**: an always-on **Ontology** (five prose sections — Telos, Evidence, Decisions, Ownership, Conduct), nine always-on **Rules** numbered I–IX, and the **Doctrines** — practice packs (`craft`, `rigor`, `ops`, `process`, `comms`) chosen per install at build time. `AGENT.md` carries all three inline (whole at the `full` footprint, heading-plus-first-line at `lean`), *and* the complete text ships as files beside it. All five pack files ship even when a pack was not built in, so a citation into an inactive pack still resolves on disk — which is why a question about a rule is answered from the tier file, never only from `AGENT.md`.
|
|
25
|
-
|
|
26
|
-
## Read-only verbs (safe, no confirmation needed)
|
|
27
|
-
|
|
28
|
-
- **Inventory** — `ls <harness>/agents`, `ls <harness>/skills`, `ls <harness>/memory`, `ls <harness>/notebook`. Counts also appear in `geneseed status`.
|
|
29
|
-
- **Status** — `geneseed status`. One-screen dashboard: theme, install mode, component counts, memory path & fact count, version fingerprint, whether `AGENT.md` is present, and whether the install matches its source.
|
|
30
|
-
- **Version** — `geneseed version`. Source vs installed fingerprint; `up to date` / `differs` / `unknown source`.
|
|
31
|
-
- **Doctor** — `geneseed doctor`. Health check on every theme + parity + links + drift. Exit 0 means `ok`.
|
|
32
|
-
- **Diff** — `geneseed diff` (or `geneseed diff --out FILE`). Shows local edits the agent has made in the deployed bundle vs the canonical source; the same content the TUI shows under *Review local edits*. Useful to find drift worth back-porting into `src/` of the Geneseed repo.
|
|
33
|
-
- **Prompt** — `geneseed prompt`. Prints the install prompt Geneseed uses to ask an external agent to recreate the bundle by hand. Safe to read; do not pipe it into another agent unless the user asks.
|
|
34
|
-
- **Read a specific piece** — read the file directly:
|
|
35
|
-
- Agent by name: `<harness>/agents/<name>.md`
|
|
36
|
-
- Skill by name: `<harness>/skills/<name>/SKILL.md`
|
|
37
|
-
- Ontology section by name: `<harness>/ontology/universal.md` — `#### Telos` / `Evidence` / `Decisions` / `Ownership` / `Conduct` (theme-independent names)
|
|
38
|
-
- Rule by numeral I–IX: `<harness>/laws/universal.md` — heading `### <Rule-noun> <ROMAN> — <title>`
|
|
39
|
-
- Doctrine rule by pack and number: `<harness>/doctrines/<pack>.md` — heading `### <Doctrine-noun> <pack> <n> — <title>`
|
|
40
|
-
- Memory fact: `<harness>/memory/<file>.md`; index at `MEMORY.md`
|
|
41
|
-
- Notebook entry: `<harness>/notebook/<file>.md`; index at `NOTEBOOK.md`
|
|
42
|
-
|
|
43
|
-
When the user asks "what does the X skill say", "what's Rule VI" or "what does Doctrine process 5 say", read the file directly — do not summarise from memory.
|
|
44
|
-
|
|
45
|
-
## Side-effecting verbs (confirm with the user first)
|
|
46
|
-
|
|
47
|
-
- **`geneseed learn`** — distills the current session into memory under `<harness>/memory/`. Writes new files and may update `MEMORY.md`. Confirm before running, and only at end-of-session or when the user explicitly asks.
|
|
48
|
-
- **`geneseed context`** — refreshes the project-docs context manifest. Touches `context.json` in the current repo. Confirm before running.
|
|
49
|
-
- **`geneseed web start`** / **`stop`** / **`status`** — manages the local web UI daemon on `127.0.0.1:4747` (offline, per-session token). `status` is read-only; `start`/`stop` need user assent. Use `start` only if the user asks to "open the web UI" or similar.
|
|
50
|
-
|
|
51
|
-
## Hard no without explicit user ask
|
|
52
|
-
|
|
53
|
-
These are destructive or scope-changing and must be initiated by the user:
|
|
54
|
-
|
|
55
|
-
- `geneseed setup` — interactive install wizard.
|
|
56
|
-
- `geneseed build` — re-renders the bundle from `src/` (only meaningful inside the Geneseed source repo anyway).
|
|
57
|
-
- `geneseed upgrade` / `bootstrap` / `update` — pull-and-rebuild from the published source; overwrites the deployed bundle. Local edits are auto-exported to an `improvements/` markdown beside the install before being overwritten, but still: user decision.
|
|
58
|
-
- `geneseed link` / `unlink` — modifies PATH symlinks.
|
|
59
|
-
- `geneseed uninstall` — removes a global or per-repo (project-scoped) install.
|
|
60
|
-
|
|
61
|
-
If the user asks for any of these, run them as requested. Do not run them as a side effect of a more general request.
|
|
62
|
-
|
|
63
|
-
## Common patterns
|
|
64
|
-
|
|
65
|
-
- **"What harness do I have here?"** — `geneseed status` (if launcher on PATH); else resolve via the fallback list and report theme, install mode, path, and version.
|
|
66
|
-
- **"What does the `<X>` skill / agent / rule say?"** — read the file directly. A Rule is `### <Rule-noun> <ROMAN> —` in `laws/universal.md`; a doctrine rule is `### <Doctrine-noun> <pack> <n> —` in `doctrines/<pack>.md`; an ontology section is `#### <Name>` in `ontology/universal.md`. The tier noun is themed, the address never is.
|
|
67
|
-
- **"Show me what's drifted in my install."** — `geneseed diff`. Offer to write the output to a file (`--out`) if it's long.
|
|
68
|
-
- **"Distill what we did today into memory."** — confirm scope, then `geneseed learn`. Afterwards, `git status` inside the harness dir to show what landed.
|
|
69
|
-
- **"Open the dashboard / web UI."** — `geneseed web start`. Tell the user the URL it printed and that it binds to localhost only.
|
|
70
|
-
- **End-of-session check-in** — if the user is wrapping up and there's something memorable, offer `geneseed learn`; if there are edits the agent made to the deployed bundle, offer `geneseed diff` so they can back-port them.
|
|
71
|
-
|
|
72
|
-
## Notes
|
|
73
|
-
|
|
74
|
-
- The launcher is `./geneseed` in the Geneseed source repo and plain `geneseed` once linked. Inside a non-Geneseed repo, only the linked form is available — if it's not on PATH, you cannot drive the CLI and must fall back to reading the deployed files directly.
|
|
75
|
-
- The deployed bundle's directory and section names are theme-independent; only voice (descriptions, prose, sigils) changes per theme. Don't pattern-match on flavor words.
|
|
76
|
-
- `memory/` and `notebook/` are personal and git-ignored. Anything you write there with `learn` does not leave the user's machine.
|