alive-capture

Featured

Capture external content into a walnut bundle -- emails, transcripts, screenshots, documents, research

AI & Automation 127 stars 8 forks Updated today MIT

Install

View on GitHub

Quality Score: 91/100

Stars 20%
70
Recency 20%
100
Frontmatter 20%
70
Documentation 15%
73
Issue Health 10%
50
License 10%
100
Description 5%
100

Skill Content

# Capture Context External content arrives -- email, transcript, file, research output. Route it into the system. ## Detect Content Type - Pasted text, forwarded message -> extract and store - File path mentioned -> read and classify - Research results from a cron or web search -> structure and route - Screenshot -> store in bundle raw/ ## Route Decision ``` This looks like [type]. Route it? 1. Add to [active bundle] raw/ 2. Create new bundle for it 3. Route to [suggested walnut] 4. Just stash it for now ``` ## Capture Process 1. **Name the file.** Date-prefixed, descriptive: `2026-04-07-witcheer-feedback.md` 2. **Store in bundle raw/.** Write the content to the active bundle's `raw/` directory. 3. **Update manifest.** Add to `sources:` in `context.manifest.yaml`: ```yaml sources: - path: raw/2026-04-07-witcheer-feedback.md description: Telegram feedback on ALIVE x Hermes spec type: document date: 2026-04-07 ``` 4. **Extract stash items.** Any decisions, tasks, insights, people mentions, or sharp quotes get stashed. 5. **Rename garbage filenames.** `CleanShot 2026-02-23...` becomes `2026-02-23-descriptive-name.ext`. Preserve original in manifest as `original_filename:`. ## If No Bundle Active Ask which walnut this belongs to, then offer to create a bundle or add to an existing one. ## If No Walnut Active Check 03_Inbox/ for unrouted files. Enter inbox scan mode -- present each file with routing suggestions.

Details

Author
alivecontext
Repository
alivecontext/alive
Created
6 months ago
Last Updated
today
Language
Python
License
MIT

Integrates with

Similar Skills

Semantically similar based on skill content — not just same category