Skip to content

igor/eplog

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 

Repository files navigation

eplog

A Claude Code skill for logging file operations to folder-level activity indexes.

What It Does

/eplog maintains _index.md files in your project folders, creating a navigable activity log that Claude can read to understand what's been happening without scanning every file.

The insight: In AI-assisted workflows, the index file becomes the functional navigation layer. Folders are emotional infrastructure (human peace of mind). The index is what Claude actually uses.

Installation

Claude Code (recommended)

Copy eplog.md to your project's .claude/commands/ folder:

mkdir -p .claude/commands
cp eplog.md .claude/commands/

Or for global availability, copy to ~/.claude/commands/.

Cowork

The skill works in Cowork as well — add the skill file to your project context.

Usage

After creating or modifying files:

/eplog "description of what you did"

Examples:

/eplog "Created initial architecture doc"
/eplog "Updated pricing models with new tier"
/eplog "Resolved entity structure decision"

What It Generates

Each folder gets an _index.md with timestamped entries:

# Folder Index

## Recent Activity

### 2026-02-02
- `17:45` **Created** [`design-doc.md`](./design-doc.md) (draft) — Initial architecture
- `14:32` **Updated** [`synthesis.md`](./synthesis.md) (active) — Added open questions

### 2026-02-01
- `09:15` **Created** [`notes.md`](./notes.md) (exploration) — Early thinking on approach

Entry Format

- `HH:MM` **Action** [`filename.md`](./filename.md) (status) — description

Actions: Created, Updated, Deleted

Status indicators:

  • (exploration) — Early thinking, may not go anywhere
  • (draft) — In progress, not ready for review
  • (nearly final) — Ready for partner review
  • (active) — Living document, updated regularly
  • (reference) — Background material, not actively edited

Reading Depth Configuration

Add to your .claude/CLAUDE.md to control how much history Claude reads:

## Index Reading Rules

**Default:** Read the 7 most recent entries from each `_index.md`

**Per-folder overrides:**
| Folder | Depth | Reason |
|--------|-------|--------|
| `decisions/` | all | Decisions are few and important |
| `working/comms/` | 10 | Coordination needs more context |

**On-demand:** If context seems incomplete, read more.

Design Principles

  1. Index files are the functional layer — Claude navigates via indexes, not folder structures
  2. Folders are emotional infrastructure — They exist for human peace of mind and legibility
  3. Every folder gets its own index — Subfolders are domains with their own activity logs
  4. Timestamps enable precision — Within a day, order matters
  5. Status indicators make working files navigable — You can tell draft from final at a glance
  6. Links make entries actionable — Click to open the actual file

Use Cases

  • Session start: "What happened since I last worked?"
  • Partner sync: "What has my collaborator been doing?"
  • Context reconstruction: "What led to this decision?"
  • Onboarding: New team member can scan indexes to understand project state

Part of Explicit Protocol

This skill embodies the Explicit Protocol philosophy: making implicit work visible, creating infrastructure that compounds, and enabling AI-assisted collaboration.

Developed by Explicit Protocol Advisory.

License

MIT

About

A Claude Code skill for logging file operations to folder indexes

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors