@limina-labs/momentum 0.31.0 → 0.31.2
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/README.md +49 -9
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -1,9 +1,14 @@
|
|
|
1
1
|
<div align="center">
|
|
2
|
-
<
|
|
2
|
+
<picture>
|
|
3
|
+
<source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/LiminaLabsAI/momentum/main/site/public/brand/momentum-lockup-dark.svg">
|
|
4
|
+
<img src="https://raw.githubusercontent.com/LiminaLabsAI/momentum/main/site/public/brand/momentum-lockup.svg" width="280" alt="momentum" />
|
|
5
|
+
</picture>
|
|
3
6
|
|
|
4
|
-
###
|
|
7
|
+
### The right context for your coding agent
|
|
5
8
|
|
|
6
|
-
|
|
9
|
+
momentum keeps your project's memory and state as Markdown files in the repo —
|
|
10
|
+
so every session your agent retrieves the right context, hallucinates less,
|
|
11
|
+
and turns a vague request into a well-grounded plan.
|
|
7
12
|
|
|
8
13
|
[](https://www.npmjs.com/package/@limina-labs/momentum)
|
|
9
14
|
[](LICENSE)
|
|
@@ -14,11 +19,24 @@
|
|
|
14
19
|
|
|
15
20
|
---
|
|
16
21
|
|
|
17
|
-
##
|
|
22
|
+
## Your agent doesn't need more context. It needs the right context.
|
|
18
23
|
|
|
19
|
-
**
|
|
24
|
+
**Without it, your agent guesses.** Ask an agent to fix a bug and it greps
|
|
25
|
+
around, fills the gaps with assumptions, and drafts a plan on shaky ground —
|
|
26
|
+
burning tokens rediscovering what your project already knew. And whatever it
|
|
27
|
+
does learn evaporates at the context limit or the next fresh session.
|
|
20
28
|
|
|
21
|
-
momentum
|
|
29
|
+
momentum hands the agent your project's real state up front. Phases, decisions,
|
|
30
|
+
history, and backlog live as plain Markdown files in the repo — read at the
|
|
31
|
+
start of every session, updated as the agent works — so it reasons from
|
|
32
|
+
recorded facts instead of guesses, and the *project* persists, not just
|
|
33
|
+
whatever one session shipped.
|
|
34
|
+
|
|
35
|
+
| Without momentum | With momentum |
|
|
36
|
+
| --- | --- |
|
|
37
|
+
| Greps the codebase, hopes it found the right files | Reads project state first — oriented in seconds |
|
|
38
|
+
| Fills the gaps with assumptions — hallucinates | Reasons from recorded decisions, not guesses |
|
|
39
|
+
| Plans on shaky ground, burns tokens rediscovering | Produces a well-grounded, actionable plan |
|
|
22
40
|
|
|
23
41
|
## Quick install
|
|
24
42
|
|
|
@@ -44,6 +62,31 @@ npx @limina-labs/momentum@latest init --agent antigravity # Antigravity
|
|
|
44
62
|
npx @limina-labs/momentum@latest init --ecosystem my-eco
|
|
45
63
|
```
|
|
46
64
|
|
|
65
|
+
## Better context in. Better work out.
|
|
66
|
+
|
|
67
|
+
- **Less hallucination** — the agent reasons from recorded facts (status,
|
|
68
|
+
decisions, history) instead of confident guesses. Fewer wrong turns, fewer
|
|
69
|
+
retries.
|
|
70
|
+
- **Higher-signal context** — the relevant history and prior decisions surface
|
|
71
|
+
first, not whatever a blind search returned. Less noise in the window, less
|
|
72
|
+
token waste.
|
|
73
|
+
- **Better-grounded plans** — a vague ask becomes a concrete plan: what to
|
|
74
|
+
change, where in the codebase, and how to verify it.
|
|
75
|
+
|
|
76
|
+
Based on hands-on use — momentum reduces guesswork and re-derivation. Your
|
|
77
|
+
mileage will vary by project and agent.
|
|
78
|
+
|
|
79
|
+
## The phase loop
|
|
80
|
+
|
|
81
|
+
Work happens in **phases** — one repeatable unit of work, five stages:
|
|
82
|
+
|
|
83
|
+
**Brainstorm → Plan → Execute → Verify → Release**
|
|
84
|
+
|
|
85
|
+
Your agent drives all of it: scopes through dialogue (disk writes gated until
|
|
86
|
+
you approve), plans groups + tasks + acceptance criteria, commits per group
|
|
87
|
+
with conventional commits, and marks tasks done only with passing evidence.
|
|
88
|
+
It stops at exactly one hard gate — the merge and release are yours to approve.
|
|
89
|
+
|
|
47
90
|
## How it works — no daemon, no lock-in
|
|
48
91
|
|
|
49
92
|
momentum isn't a framework or a background service. It's plain files and conventions your agent already knows how to read:
|
|
@@ -154,9 +197,6 @@ flowchart LR
|
|
|
154
197
|
B2 --> B4[project-b/]
|
|
155
198
|
B2 --> B5[project-c/]
|
|
156
199
|
end
|
|
157
|
-
|
|
158
|
-
style single fill:#F8FAFC,stroke:#94A3B8,stroke-width:1px
|
|
159
|
-
style multi fill:#EEF2FF,stroke:#4F46E5,stroke-width:1px
|
|
160
200
|
```
|
|
161
201
|
|
|
162
202
|
**Hard invariant**: a project running `momentum init` without `--ecosystem` sees zero difference from before ecosystem mode existed. Ecosystem mode is purely additive.
|
package/package.json
CHANGED