mirror of
https://github.com/himanshudongre/smriti.git
synced 2026-09-10 22:41:12 +00:00
dafd164968
Removed stale Loom link and old checkpoint-diff screenshot reference. README now references 3 new screenshots matching the current product: LineagePage dashboard, CLI state+metrics, checkpoint detail with structured tasks and notes. Screenshots are placeholders until captured from the live environment. Added docs/DEMO_SCRIPT.md: 3-act, 4-5 minute recording script covering the state surface, coordination proof (metrics + autonomy milestone), and developer experience. Includes exact commands, highlight points, and a screenshot capture checklist.
144 lines
4.5 KiB
Markdown
144 lines
4.5 KiB
Markdown
# Smriti demo recording script
|
|
|
|
Target: 4-5 minute Loom recording for README and Reddit relaunch.
|
|
Story: multi-agent coordination on a real project, not a toy.
|
|
|
|
---
|
|
|
|
## Pre-recording setup
|
|
|
|
1. Backend running: `make dev` (port 8000)
|
|
2. Frontend running: `make dev-frontend` (port 5173)
|
|
3. Terminal visible with dark theme, readable font size
|
|
4. Chrome open to `http://localhost:5173`
|
|
5. smriti-dev space has real data (56 checkpoints, 37 claims)
|
|
|
|
---
|
|
|
|
## Act 1: The state surface (90 seconds)
|
|
|
|
### Terminal
|
|
|
|
```bash
|
|
# "Every agent starts by reading the current state."
|
|
smriti state smriti-dev --compact
|
|
```
|
|
|
|
**Highlight while output is visible:**
|
|
- "This is what every coding agent reads before starting work."
|
|
- Point to: latest checkpoint by `codex-local`, decisions, structured tasks with `[intent]` tags and `(id: ...)` slugs
|
|
- "Two agents — Claude Code and Codex — have been working on this project. The state shows what was decided, what's in progress, and who did what."
|
|
|
|
### Browser: LineagePage
|
|
|
|
Navigate to `http://localhost:5173/spaces/5d08572f-08d4-4b17-b853-396a064d9c4f/lineage`
|
|
|
|
**Highlight:**
|
|
- Summary panel: project name, current direction, status bar (checkpoints, claims, tasks)
|
|
- Checkpoint timeline: scroll through the cards
|
|
- Point to milestone markers (star icons)
|
|
- Point to author agent badges (`claude-code` vs `codex-local`)
|
|
- "This is 56 checkpoints of real development work. Not a demo project — the actual product."
|
|
|
|
---
|
|
|
|
## Act 2: The coordination proof (2 minutes)
|
|
|
|
### Terminal
|
|
|
|
```bash
|
|
# "Here are the project-level coordination metrics."
|
|
smriti metrics smriti-dev
|
|
```
|
|
|
|
**Highlight while output is visible:**
|
|
- "56 checkpoints, 2 agents, 30 cross-agent continuations"
|
|
- "That means 30 times one agent picked up where a different agent left off."
|
|
- "37 claims, 100% completion, 3 with task IDs"
|
|
|
|
### Milestone checkpoint
|
|
|
|
```bash
|
|
# "Let me show you the autonomy proof."
|
|
smriti checkpoint show <autonomy-milestone-id>
|
|
```
|
|
|
|
Use ID: `a11d5579-1ad8-457e-a7a4-a37411540aab`
|
|
|
|
**Highlight:**
|
|
- The 4 structured tasks with IDs and intent hints
|
|
- The milestone note: "first clean autonomous task-ID-based complementary work"
|
|
- The founder note about the product wedge
|
|
- "Two agents started at the same time, read this task surface, and independently picked different tasks. No human told them what to do."
|
|
|
|
### Browser: CommitDetailPage
|
|
|
|
Navigate to the milestone checkpoint detail page.
|
|
|
|
**Highlight:**
|
|
- Structured tasks with intent badges (blue), task IDs
|
|
- Notes section with milestone marker and founder commentary
|
|
- "The reasoning state isn't prose — it's structured fields that agents can read and act on."
|
|
|
|
---
|
|
|
|
## Act 3: Developer experience (60 seconds)
|
|
|
|
### Terminal
|
|
|
|
```bash
|
|
# "Setting up a new project takes one command."
|
|
smriti init demo-project
|
|
```
|
|
|
|
**Highlight:**
|
|
- Space created, skill pack installed, SessionStart hook configured
|
|
- "After this, every Claude Code session automatically reads the state before starting."
|
|
|
|
```bash
|
|
# "The skill pack teaches the agent the whole workflow."
|
|
head -30 .claude/skills/smriti/SKILL.md
|
|
```
|
|
|
|
**Highlight:**
|
|
- "Read state first. Checkpoint at inflection points. Check for collisions."
|
|
- "The agent doesn't need to be told how to use Smriti — the skill pack makes it the default reflex."
|
|
|
|
```bash
|
|
# "And the backend tells agents what features are available."
|
|
curl -s localhost:8000/health | python3 -m json.tool
|
|
```
|
|
|
|
**Highlight:**
|
|
- git_sha, capabilities list
|
|
- "If an agent hits a stale backend, it knows immediately."
|
|
|
|
---
|
|
|
|
## Closing (30 seconds)
|
|
|
|
"Smriti is version control for reasoning — a shared backend where multiple coding agents coordinate without an orchestrator. It's open source. The repo you're looking at was built using the system itself."
|
|
|
|
---
|
|
|
|
## Screenshot capture checklist
|
|
|
|
Capture these 3 screenshots after recording (or during):
|
|
|
|
### 1. `docs/assets/lineage-dashboard.png`
|
|
- URL: `http://localhost:5173/spaces/5d08572f-08d4-4b17-b853-396a064d9c4f/lineage`
|
|
- Capture: the summary panel + first ~6 checkpoint cards
|
|
- Make sure milestone markers (star) and author badges are visible
|
|
- Crop to ~1200x800px, dark theme
|
|
|
|
### 2. `docs/assets/cli-state-and-metrics.png`
|
|
- Terminal: run `smriti state smriti-dev --compact` then `smriti metrics smriti-dev`
|
|
- Capture both outputs in one screenshot
|
|
- Dark terminal background, readable font
|
|
- Crop to ~900x600px
|
|
|
|
### 3. `docs/assets/checkpoint-detail.png`
|
|
- URL: CommitDetailPage for the autonomy milestone checkpoint
|
|
- Or terminal: `smriti checkpoint show a11d5579-1ad8-457e-a7a4-a37411540aab`
|
|
- Capture: structured tasks with intent badges + notes section
|
|
- Crop to ~1200x700px
|