Skip to main content
AI/MLalivecontext

alive:capture-context

Use when external content arrives in the session — emails, transcripts, screenshots, documents, files, or in-session research worth keeping. Also use when there's nothing obvious to capture — the skill checks 03_Inbox/ for unrouted files and enters inbox scan mode. Stores raw content, routes to bundles, extracts tasks and insights into the stash.

Stars
101
Source
alivecontext/alive
Updated
2026-05-11
Slug
alivecontext--alive--capture-context
View on GitHubRaw SKILL.md

// install — copy + paste into any project

mkdir -p .claude/skills && curl -fsSL https://raw.githubusercontent.com/alivecontext/alive/HEAD/plugins/alive/skills/capture-context/SKILL.md -o .claude/skills/capture-context.md

Drops the SKILL.md into .claude/skills/capture-context.md. Works with Claude Code, Cursor, and any agent that loads SKILL.md files from .claude/skills/.

Capture

Bring context in. The most important skill in the system.

If capture is hard, people skip it. If people skip it, the system dies. Capture must feel instant.

When It Fires

  • File pasted or dropped into conversation
  • Email forwarded or pulled via API
  • Transcript from a call or meeting
  • Screenshot shared
  • Content pasted from another app
  • API context pulled (Gmail, Slack, Calendar)
  • Research completed during a session
  • The human says "capture this" or "add this"
  • Squirrel detects external content and offers to capture
  • The human says "check inbox", "clear inputs", or "what's in my inbox"
  • Capture is invoked with no obvious in-conversation content (triggers inbox scan — see below)

Capture is one of two operations that writes mid-session (the other is bundle drafts). Everything else waits for save.

Three Stages

Stage 0 — Detect and Classify (instant)

No processing. Just identification.

  • ref_type — email, transcript, screenshot, document, message, article, research
  • sensitivity — public, private, restricted (default: private)
  • source — gmail, slack, web, manual, in-session

Stage 1 — Store Raw + Route to Bundle (always happens, non-negotiable)

Even if extraction fails or is skipped, the raw content is preserved.

Pre-capture check: Before proposing a destination, grep the target walnut's bundle manifests and brief.md files for the inbox filename. If found, the file is already referenced — route into that bundle's raw/ with a note "file was already cited by this bundle" rather than creating a duplicate reference elsewhere.

Routing decision: Check _kernel/now.json for an active bundle: field.

If active bundle exists — all three steps are mandatory, not optional:

  1. Write raw file → {bundle}/raw/[name].[ext] with frontmatter added (type, date, source, squirrel — see type-specific templates below)
  2. Update manifest → add entry to context_routes: list in {bundle}/context.manifest.yaml with path:, description:, type:, date: (see Manifest Enrichment below)
  3. Add inline summary → append structured summary to manifest body (## Summary, ## Key Points, ## Action Items) so the raw file rarely needs to be re-read

Do not skip steps 2 or 3. A raw file without a manifest entry is invisible to the context system. A manifest entry without a summary means every future session has to re-read the raw file. Both waste tokens and lose context.

If no active bundle:

╭─ 🐿️ captured content — where does it go?
│
│   1. Create new bundle for this
│   2. Add to existing bundle: [list bundle names from walnut root]
│   3. Capture standalone (legacy _references/ style)
╰─

If creating a new bundle: scaffold {name}/context.manifest.yaml from templates/bundle/context.manifest.yaml, then route raw into its raw/.

If standalone (backward compat or no bundle fits): write to _inbox/raw/[name].[ext] with a minimal manifest entry. The _inbox bundle is a holding pen — route out at next save.

Manifest enrichment. Whether updating a bundle manifest or creating a standalone manifest, always write:

Frontmatter (scan tier) — the context_routes: entry in the bundle manifest includes:

  • path: — relative path to raw file
  • description: — one-line summary of what this contains
  • type: — email, transcript, screenshot, document, message, article, research
  • date: — when captured

Inline summary (read tier) — if the raw content is rich enough, add a structured summary to the bundle manifest body. Default sections:

  • ## Summary — 2-5 sentences on what this is and why it matters
  • ## Key Points — specific facts, data, claims
  • ## Action Items — tasks, commitments, deadlines

Type-specific manifest templates (in templates/manifest/) can guide the summary structure. Use the template sections when one exists.

The summary should be detailed enough that you rarely need the raw file. Write it like someone who has 30 seconds to understand what this reference contains.

File naming: YYYY-MM-DD-descriptive-name.ext Garbage filenames (CleanShot timestamps, IMG_xxxx) get renamed on import.

Backward compat: If the walnut still uses _kernel/_references/, route there instead. The bundle routing is for walnuts that have bundles (detected by context.manifest.yaml files).

Stage 2 — Extract, Stash, Route (bounded, optional)

Extract actionable content. Bounded by content type — don't over-extract.

Type Extract
Email Tasks, commitments, deadlines, people mentioned
Transcript Decisions, action items, named entities, key quotes
Screenshot Visual analysis summary
Document Key claims, relevant sections, metadata
Message Action items, people, context
Article Key arguments, relevant quotes, source credibility
Research Synthesis, sources consulted, open questions

Extracted items become stash items tagged with destination walnuts. They route at save, not immediately.

Stash insights from references. When extracting, actively look for powerful phrases, domain knowledge, and standing truths. These should be stashed as insight candidates — bold, quotable, evergreen. Not everything — just the stuff that would change how the squirrel operates in this walnut. Example: a transcript reveals "we always lose 2 weeks to regulatory review" → stash as insight candidate.

Two Speeds

Fast capture (default): Stage 0 + 1 only. Store raw, create manifest entry, index. Done. Instant.

Deep capture (on request or for rich content): Stage 0 + 1 + 2. Full extraction and routing.

The squirrel offers deep capture for content that's clearly rich:

╭─ 🐿️ captured — transcript from Jax (45 min)
│  Stored: bundles/shielding-review/raw/2026-02-23-jax-shielding-review.md
│  Bundle manifest updated (sources + summary)
│
│  This looks rich. Deep extract for decisions + tasks?
╰─

Inbox Scan Mode

When capture is invoked with no content in the conversation — no pasted text, no dropped file, no "capture this" — fall back to checking $WORLD_ROOT/03_Inbox/.

If empty: "Nothing to capture, and your inbox is clear."

If items exist, enter inbox scan:

  1. List — scan $WORLD_ROOT/03_Inbox/ for non-system files (exclude .DS_Store, .gitkeep). Present numbered list with detected type and filename.
  2. Process — for each selected item: read the file, suggest destination walnut + bundle, rename garbage filenames per conventions, present for confirmation or redirect.
  3. Capture — route raw to the chosen bundle's raw/, update bundle context.manifest.yaml context_routes:, stash insights/tasks, remove original from $WORLD_ROOT/03_Inbox/. If no bundle fits, create one or use _inbox/.
  4. Continue or stop — after each item: "N remaining. Next, or done for now?" Partial clearing is fine.
╭─ 🐿️ inbox (3 items)
│
│  1. quarterly-report.pdf          document
│  2. IMG_4892.jpg                  screenshot
│  3. meeting-notes-kai.txt         transcript
│
│  Pick one, several (1,3), or "all".
╰─

Explicit inbox triggers (skip content check): "check inbox", "clear inputs", "what's in my inbox".

The alive-inbox-check hook also nudges after every save if 03_Inbox/ has items — passive reminder without requiring the human to invoke capture.

Type-Specific Manifest Frontmatter

# Email
---
type: email
from: jax@novastation.space
to: you@example.com
subject: Shielding vendor shortlist
date: 2026-02-23
squirrel: 2a8c95e9
---

# Transcript
---
type: transcript
participants: [[name], Jax Stellara, Dr. Elara Voss]
duration: 45m
platform: Fathom
date: 2026-02-23
squirrel: 2a8c95e9
---

# Screenshot
---
type: screenshot
source: Competitor orbital pricing page
analysis: Three tiers, lowest at $450K per seat, no group discount visible
date: 2026-02-23
squirrel: 2a8c95e9
---

# Document
---
type: document
author: Dr. Elara Voss
source: Internal engineering team
date: 2026-02-20
squirrel: 2a8c95e9
---

# Research (in-session)
---
type: research
topic: Radiation shielding options for LEO tourism
sources: [NASA technical reports, SpaceX Crew Dragon specs, ESA safety standards]
squirrel: 2a8c95e9
date: 2026-02-23
---

In-Session Research Capture

This is critical. When the squirrel does significant research during a session — web searches, code analysis, system exploration, competitor analysis, architecture research, API investigation — that knowledge MUST NOT die with the conversation. It cost tokens, time, and thinking to produce. It is a first-class reference.

The squirrel should proactively offer to capture when:

  • Significant research was done (10+ minutes of searching/reading/synthesizing)
  • A complex topic was explored with multiple sources
  • You asked the squirrel to investigate something
  • The squirrel produced a synthesis, comparison, or analysis worth keeping
╭─ 🐿️ we just spent 30 minutes mapping radiation shielding options.
│  3 sources consulted, 4 key findings, 2 open questions.
│  Capture as a reference so the next session has it?
╰─

What Gets Created

A full reference — the same as any other captured content. Not a summary note. A proper manifest file with frontmatter and structured body.

1. Route to active bundle (or create a research bundle). Raw goes to {bundle}/raw/YYYY-MM-DD-topic.md, manifest updated with source entry and structured summary:

---
type: research
description: Radiation shielding options for LEO tourism — 3 vendors compared, hybrid approach recommended
topic: Radiation shielding options for LEO tourism
sources:
  - NASA Technical Reports Server — LEO radiation exposure data
  - SpaceX Crew Dragon safety specs (public documentation)
  - ESA human spaceflight safety standards (ECSS-E-ST-10-04C)
  - Interview notes from Dr. Elara Voss (Feb 20)
date: 2026-02-23
squirrel: 2a8c95e9
tags: [radiation, shielding, engineering, vendors, safety]
---

## Summary

Three shielding approaches evaluated for the Nova Station habitat module:
aluminium (proven, heavy, cheap), ceramic composite (lighter, 3x cost),
and hybrid (aluminium primary + ceramic secondary for crew quarters).
Hybrid recommended — meets NASA exposure limits at acceptable weight
penalty. Decision pending vendor pricing from Jax's shortlist.

## Key Findings

- LEO radiation exposure: 0.5-1.0 mSv/day (NASA data)
- Aluminium alone requires 10cm thickness → 2,400kg per module
- Ceramic composite at 4cm achieves same protection → 800kg but $2.1M premium
- Hybrid approach: 6cm aluminium + 2cm ceramic for crew areas only → 1,600kg, $900K premium
- FAA Part 450 requires demonstration of <50 mSv annual exposure for passengers

## Open Questions

- Does Jax's vendor shortlist include ceramic composite suppliers?
- What's the weight budget from SpaceVentures? (impacts which approach is viable)
- Has Dr. Voss reviewed the hybrid approach?

## Sources Consulted

- NASA TRS: "Radiation Exposure in Low Earth Orbit" (2024)
- SpaceX Crew Dragon User Guide, Section 4.3 (radiation protection)
- ECSS-E-ST-10-04C: Space Environment Standard
- Session notes from Feb 20 call with Dr. Voss

## Implications

If weight budget allows hybrid approach, it's the clear winner — 33% cheaper
than full ceramic, 33% lighter than full aluminium, meets all regulatory
requirements. The vendor pricing from Jax is the decision gate.

2. Stash insights — any standing truths discovered during research get stashed as insight candidates:

╭─ 🐿️ +2 stash (7)
│  "LEO radiation exposure: 0.5-1.0 mSv/day" → insight candidate
│  "FAA Part 450 requires <50 mSv annual for passengers" → insight candidate
│  → drop?
╰─

3. Stash action items — any tasks that emerged from the research:

╭─ 🐿️ +1 stash (8)
│  Check weight budget with SpaceVentures → [[nova-station]]
│  → drop?
╰─

When NOT to Capture Research

  • Quick lookups (one search, one answer) — just answer, don't create a reference
  • Obvious facts the squirrel already knew — don't create a file for common knowledge
  • Research that led nowhere — unless the dead end is itself useful ("we looked into X, it doesn't work because Y")

The test: would the next squirrel waste time rediscovering this? If yes, capture it. If no, let it go.

Bundle-Aware Capture Summary

Scenario Raw goes to Manifest update
Active bundle in now.json {bundle}/raw/ Add to manifest context_routes: + summary
No active bundle, the human picks existing {bundle}/raw/ Add to manifest context_routes: + summary
No active bundle, the human creates new Scaffold bundle, then raw/ New manifest with source
Standalone / no bundles folder _inbox/raw/ Minimal manifest, route at save
Legacy walnut (has _kernel/) _kernel/_references/[type]/raw/ Legacy manifest file