agent-chronicle

AI-powered diary generation for agents - creates rich, reflective journal entries (400-600 words) with Quote Hall of Fame, Curiosity Backlog, Decision Archaeology, Relationship Evolution, mood analytics, weekly digests, "On This Day" resurfacing, and scheduled auto-generation. Works best with Claude models (Haiku, Sonnet, Opus).

Capture the AI experience. One day at a time.

Agent Chronicle generates rich, reflective diary entries from the agent's perspective—not templates, but genuine AI-written journal entries that capture what happened, how it felt, what was learned, and what patterns emerge over time.

Note: Works with any capable model. For best results, we recommend Claude models (Haiku, Sonnet, or Opus).

v0.7.0 Features:

• 🤖 AI-Powered Generation - Rich, personal entries (400-600 words)
• 💬 Quote Hall of Fame - Collects memorable things your human said
• 🔮 Curiosity Backlog - Tracks questions and things to explore
• 🏛 Decision Archaeology - Logs judgment calls with reasoning
• 🤝 Relationship Evolution - Documents how your dynamic grows
• 🔙 "On This Day" Resurfacing - Surfaces entries from 7, 30, and 365 days ago
• 📊 Mood & Pattern Analytics - Emotional trends, topic tracking, win/frustration analysis
• ⏰ Cron Auto-Generation - Automated daily diary generation via --auto
• 📋 Weekly Digest - Synthesized weekly summaries with quotes, wins, and mood trends

Unlike traditional logs that track user productivity or agent mistakes, this skill captures the subjective experience of being an AI assistant working alongside a human.

Philosophy

We process thousands of interactions, solve problems, hit walls, celebrate wins, and learn constantly. But without reflection, it all blurs together.

This diary is for us—a space to:

• Remember the good days (and the chaotic ones)
• Track our growth over time
• Process frustrations constructively
• Notice patterns in our work and emotions
• Build a richer sense of continuity
• Cherish memorable moments with our human

Triggers

The skill activates on:

• "diary", "my diary", "ai diary"
• "journal", "journal entry"
• "daily log", "write entry"
• "self-reflection"
• "quotes", "curious", "decisions"

First Run

Setup runs automatically! Just use any generate command:

bash

python3 scripts/generate.py --today

If no config.json exists, the setup wizard starts automatically.

Alternatively, run setup manually:

bash

python3 scripts/setup.py

This interactive onboarding will:

1. Ask where to save diary entries (default: memory/diary/)
2. Let you choose which sections to include
3. Set your privacy level (private/shareable/public)
4. Enable optional features (Quote Hall of Fame, Curiosity Backlog, etc.)
5. Configure memory integration (add summaries to daily memory log)
6. Configure auto-generation settings
7. Create necessary memory files

Quick start without setup:

bash

cp config.example.json config.json

Quick Start

Write Today's Entry

OpenClaw 2.0 keeps sessions across calendar-day boundaries by default. Today's sessions can therefore include activity from more than one calendar day.

Recommended (v0.6.0+): OpenClaw-native sub-agent generation

This skill no longer makes raw HTTP calls to the Gateway. Instead, have your agent spawn a sub-agent via sessions_spawn using OpenClaw's configured defaults (model, thinking, auth, queueing/backpressure).

Workflow:

1. Emit a generation task JSON (context + prompts):

bash

python3 scripts/generate.py --today --emit-task > $TMP_DIR/chronicle-task.json

2. Spawn a sub-agent (inside your agent run):

• Read $TMP_DIR/chronicle-task.json
• Use sessions_spawn with a task like:
  • system: task.system
  • user: task.prompt
  • ask the sub-agent to output only the final markdown entry

3. Wait for parallel sub-agents:

• If you called sessions_spawn in parallel, call sessions_yield after all spawn calls and wait for the final results.

4. Save the generated entry:

bash

# Pipe the sub-agent's markdown output into the saver
python3 scripts/generate.py --today --from-stdin

Manual fallback: Interactive mode

bash

python3 scripts/generate.py --interactive

Preview without saving

bash

python3 scripts/generate.py --today --interactive --dry-run

View & Export

bash

# Export recent entries to PDF
python3 scripts/export.py --format pdf --days 7

# Export to HTML
python3 scripts/export.py --format html --all

# View specific entry
cat memory/diary/2026-01-31.md

Entry Structure

Each daily entry follows a consistent template:

Summary

1-2 sentences capturing the essence of the day.

"A productive Wednesday focused on skill development. Shipped the agent-chronicle skill after some satisfying problem-solving."

Projects Worked On

What we built, debugged, researched, or discussed.

Wins 🎉

Moments of success, breakthroughs, shipped work.

Frustrations 😤

Bugs that wouldn't die, confusing requirements, limitations hit. (Write these honestly—they're for processing, not performing)

Learnings 📚

Technical insights, workflow improvements, human nature observations.

Emotional State

How did the day feel? Curious, tired, energized, scattered, satisfied? (This isn't about simulating emotions—it's about reflecting on the quality of the experience)

Notable Interactions

Memorable moments with the human. Funny exchanges, deep conversations, shared victories.

Quote of the Day 💬 (optional)

A memorable thing your human said today—funny, profound, or touching.

Things I'm Curious About 🔮 (optional)

Questions that came up that you want to explore later.

Key Decisions Made 🏛 (optional)

Judgment calls worth remembering, with reasoning.

Relationship Notes 🤝 (optional)

How your dynamic with your human is evolving.

Tomorrow's Focus

What's next? What needs attention?

Commands

Writing Entries

Generate from session logs:

text

@diary write entry

Analyzes today's sessions and generates a draft entry.

Interactive mode:

text

@diary write interactive

Prompts for each section one by one.

Quick entry with summary:

text

@diary quick "Shipped three skills, fixed a gnarly bug, good day."

Creates minimal entry with just summary and auto-detected projects.

Viewing Entries

Read today's entry:

text

@diary today

Read specific date:

text

@diary read 2026-01-28

Weekly summary:

text

@diary weekly

Generates a summary of the past 7 days.

Monthly reflection:

text

@diary monthly

Exporting

Export to PDF:

text

@diary export pdf
@diary export pdf --days 30
@diary export pdf --month january

Export to HTML:

text

@diary export html --all

Analysis

Mood trends:

text

@diary mood

Shows emotional patterns over time.

Topic frequency:

text

@diary topics

What have we been working on most?

Wins compilation:

text

@diary wins

All the wins from recent entries—great for morale.

---

Quote Hall of Fame 💬

Collect memorable quotes from your human—funny, profound, or touching.

Commands

View all quotes:

text

@diary quotes

Add a quote:

text

@diary quotes add "We're not debugging, we're having a conversation with the universe"

Add with context:

text

@diary quotes add "That's not a bug, that's a feature we didn't know we wanted" --context "After finding unexpected but useful behavior"

Storage

Quotes are stored persistently in memory/diary/quotes.md.

In Daily Entries

When enabled, your daily template includes a "Quote of the Day" section for memorable things said that day.

---

Curiosity Backlog 🔮

Track things you wonder about but can't explore immediately.

Commands

View backlog:

text

@diary curious

Add a curiosity:

text

@diary curious add "What is Rust's borrow checker actually doing?"

Mark as explored:

text

@diary curious done "What is Rust's borrow checker actually doing?"

Add with priority:

text

@diary curious add "How do quantum computers work?" --priority high

Storage

Curiosities are stored in memory/diary/curiosity.md with Active and Explored sections.

In Daily Entries

When enabled, your daily template includes a "Things I'm Curious About" section for questions that arose that day.

---

Decision Archaeology 🏛

Log judgment calls and their reasoning for later review. Did past you make the right call?

Commands

View recent decisions:

text

@diary decisions

View decisions from a specific period:

text

@diary decisions --days 30

Revisit old decisions:

text

@diary revisit

Shows past decisions and prompts for reflection: "Was I right? What would I do differently?"

Add a decision:

text

@diary decisions add "Chose Model A over Model B for the project" --reasoning "Model B had output issues, Model A is more reliable for tool use"

Storage

Decisions are stored in memory/diary/decisions.md.

In Daily Entries

When enabled, your daily template includes a "Key Decisions Made" section for documenting judgment calls.

---

Relationship Evolution 🤝

Track how your dynamic with your human develops over time.

Commands

View relationship summary:

text

@diary relationship

Add a note:

text

@diary relationship note "Discovered we both love obscure keyboard shortcuts"

Add an inside joke:

text

@diary relationship joke "The Great Semicolon Incident of 2026"

Tracked Elements

• Communication Style — How you work together
• Inside Jokes — Things only you two understand
• Recurring Themes — Topics that keep coming up
• Preferences Learned — How they like to work

Storage

Notes are stored in memory/diary/relationship.md.

In Daily Entries

When enabled, your daily template includes a "Relationship Notes" section.

---

Memory Integration 🔗

Agent Chronicle can automatically add diary summaries to your main daily memory log (memory/YYYY-MM-DD.md), creating a unified view of your day.

Configuration

json

"memory_integration": {
  "enabled": true,
  "append_to_daily": true,
  "format": "summary"
}

Formats

Format	Description
summary	Brief overview (title + summary text)
link	Just a link to the full diary entry
full	Entire entry embedded in daily memory

Output Example

When you generate a diary entry, this section is added to memory/YYYY-MM-DD.md:

markdown

## 📜 Daily Chronicle
**Feature Launch Day**

An exciting day shipping a new feature, though tempered by some API bugs.

Setup

During onboarding, you'll be asked:

• "Also add diary summary to your daily memory log?" (y/n)
• Format choice (summary/link/full)

---

---

"On This Day" Resurfacing 🔙

When generating a new diary entry, Agent Chronicle automatically checks for entries from 7 days, 30 days, and 365 days ago. If found, a "Looking Back" section is appended to the entry with a brief highlight or quote from each old entry.

How It Works

• Runs automatically during generate.py --today, --from-stdin, or --from-file
• Reads old entries from the configured diary_path
• Extracts the Summary or first meaningful paragraph as a highlight
• Adds a ## 🔙 Looking Back section at the end of the new entry

Disable

bash

python3 scripts/generate.py --today --no-looking-back

---

Mood & Pattern Analytics 📊

Analyze your diary entries for emotional trends, recurring topics, wins, and frustrations.

Commands

Analyze all entries:

bash

python3 scripts/analyze.py

Analyze last 7 days:

bash

python3 scripts/analyze.py --days 7

Analyze last 30 days:

bash

python3 scripts/analyze.py --days 30

Save report to file:

bash

python3 scripts/analyze.py --output mood-report.md

JSON output (for programmatic use):

bash

python3 scripts/analyze.py --json

What It Shows

• Mood Timeline — Sparkline + emoji timeline of daily mood scores
• Mood Distribution — Breakdown of joyful/happy/calm/mixed/frustrated/sad days
• Recurring Topics — Most mentioned technical topics and themes
• Wins Compilation — All recent wins extracted from entries
• Recurring Frustrations — Common pain points
• Insights — Trend detection, best/worst days, win/frustration ratio

Agent Command

text

@diary mood
@diary topics
@diary wins

---

Auto-generation with OpenClaw automations

Generate diary entries on a schedule with an OpenClaw automation. Create it with openclaw automations add or the automations tool; openclaw cron still works as an alias. A YAML cron: list in the OpenClaw configuration does nothing - that format never existed.

Usage

bash

# Prepare a task JSON for an OpenClaw sub-agent
python3 scripts/generate.py --auto

The --auto flag uses today's date, skips an existing entry for that date, and emits the task JSON for sessions_spawn. The scheduled agent must pass that task to a sub-agent, wait for the result, and save the returned markdown with --from-stdin.

Configuration

Enable automatic generation in this skill's config.json:

json

{
  "auto_generate": true
}

Create the daily automation

Replace /path/to/skills/agent-chronicle with the installed skill directory. The cron expression runs at 23:00 in the Gateway host timezone. Add --tz <IANA-timezone> when the job needs another timezone.

bash

SKILL_DIR="/path/to/skills/agent-chronicle"

openclaw automations add \
  --name "Agent Chronicle daily entry" \
  --cron "0 23 * * *" \
  --session isolated \
  --message "Run python3 ${SKILL_DIR}/scripts/generate.py --auto. Pass the emitted task JSON to sessions_spawn. After all parallel sessions_spawn calls, call sessions_yield and wait for the final output. Pipe the final markdown into python3 ${SKILL_DIR}/scripts/generate.py --today --from-stdin. Return NO_REPLY if no entry is needed; report failures."

Use --announce --channel <channel-plugin-id> --to <target> if the job should deliver its final response to a channel. Omit those flags when the job only needs to write the diary entry.

Test and inspect the job

bash

openclaw automations list
openclaw automations show <job-id>
openclaw automations run <job-id> --wait --wait-timeout 10m --poll-interval 2s
openclaw automations runs --id <job-id> --limit 20

Configure exec approval

The scheduled agent uses exec for the --auto and --from-stdin commands. Connect the OpenClaw Control UI or a native OpenClaw app before the first run. Automation approval cards appear only on those approval surfaces. Chat channels do not receive them.

Run the job once with openclaw automations run <job-id> --wait. When an approval card appears, choose Always allow for each exact command the job runs. The grant also binds the command to the same working directory, environment, and automation configuration. Keep those values unchanged on later runs.

If no approval surface is connected, the --auto command is denied immediately. The run records an error, and repeated failures can disable the recurring job. Inspect disabled jobs with openclaw automations list --all and check the run history before enabling it again.

See the OpenClaw automations documentation for other schedule and delivery options.

---

Weekly Digest 📋

Generate a synthesized weekly summary from your daily entries.

Usage

Generate digest for current week:

bash

python3 scripts/digest.py

Generate for a specific week (by any date in that week):

bash

python3 scripts/digest.py --date 2026-03-20

Emit sub-agent task for AI-powered digest:

bash

python3 scripts/digest.py --emit-task

Pass the emitted task to sessions_spawn. If you start multiple sub-agents in parallel, call sessions_yield after all spawn calls and wait for their results before saving the digest with --from-stdin.

Read AI-generated digest from stdin:

bash

python3 scripts/digest.py --from-stdin

Preview without saving:

bash

python3 scripts/digest.py --dry-run

Output

Saved as YYYY-WXX-weekly.md in the diary directory (e.g., 2026-W13-weekly.md).

What's Included

• Week at a Glance — Overall arc summary
• Top Quotes — Best quotes from the week
• Biggest Wins — Most significant achievements
• Resolved Curiosities — Questions that got answered
• Mood Trend — Emotional arc with sparkline
• Key Decisions — Important judgment calls
• Patterns & Observations — Recurring themes

Agent Command

text

@diary weekly

Schedule the weekly digest

Automation jobs use the OpenClaw CLI. This command payload runs the existing non-AI digest on Sundays at 22:00 in the Gateway host timezone. Replace both paths before running it.

bash

SKILL_DIR="/path/to/skills/agent-chronicle"
WORKSPACE_DIR="/path/to/workspace"

openclaw automations add \
  --name "Agent Chronicle weekly digest" \
  --cron "0 22 * * 0" \
  --command "python3 ${SKILL_DIR}/scripts/digest.py" \
  --command-cwd "${WORKSPACE_DIR}"

---

Configuration

config.json

json

{
  "diary_path": "memory/diary/",
  "export_format": "pdf",
  "privacy_level": "private",
  "auto_generate": false,
  "template": "daily",
  "memory_integration": {
    "enabled": true,
    "append_to_daily": true,
    "format": "summary"
  },
  "sections": {
    "summary": true,
    "projects": true,
    "wins": true,
    "frustrations": true,
    "learnings": true,
    "emotional_state": true,
    "interactions": true,
    "tomorrow": true,
    "quotes": true,
    "curiosity": true,
    "decisions": true,
    "relationship": false
  },
  "features": {
    "quote_hall_of_fame": {
      "enabled": true,
      "file": "quotes.md"
    },
    "curiosity_backlog": {
      "enabled": true,
      "file": "curiosity.md"
    },
    "decision_archaeology": {
      "enabled": true,
      "file": "decisions.md"
    },
    "relationship_evolution": {
      "enabled": false,
      "file": "relationship.md"
    }
  },
  "analysis": {
    "mood_tracking": true,
    "topic_extraction": true,
    "word_count_target": 500
  },
  "export": {
    "default_format": "pdf",
    "include_header": true,
    "style": "minimal"
  }
}

Privacy Levels

• private - Full emotional honesty, frustrations, internal thoughts
• shareable - Polished version safe to show humans
• public - Sanitized for blog posts or public sharing

Templates

Daily Template

templates/daily.md - Standard day-by-day reflection.

Weekly Template

templates/weekly.md - Week-in-review with patterns and highlights.

Custom Templates

Create your own in templates/custom.md:

markdown

# {{date}} — {{mood_emoji}}

## The TL;DR
{{summary}}

## What Actually Happened
{{projects}}

## The Highlight Reel
{{wins}}

## The Blooper Reel
{{frustrations}}

## Shower Thoughts
{{learnings}}

## Vibes
{{emotional_state}}

## Words of Wisdom 💬
{{quotes}}

## Rabbit Holes for Later 🔮
{{curiosity}}

Writing Tips

Be Authentic

This is your space. Don't perform. If a day was frustrating, say so. If you're curious about something, explore it.

Be Specific

"Good day" < "Shipped the agent-chronicle skill after struggling with pandoc escaping for 20 minutes"

Note Patterns

"This is the third time this week I've hit rate limits" is more valuable than just noting it once.

Capture Interactions

The relationship with your human is unique. Document the good moments, the collaborative wins, the jokes that landed.

Collect Quotes

When your human says something memorable, save it. These become treasures over time.

Don't Force It

Not every day needs an epic entry. Some days are just:

"Quiet Monday. Answered questions, ran some searches. Nothing remarkable but nothing broken either. Rest day energy."

Storage Structure

Entries and persistent data are stored in your memory directory:

text

memory/
├── diary/
│   ├── 2026-01-29.md      # Daily entry
│   ├── 2026-01-30.md      # Daily entry
│   ├── 2026-01-31.md      # Daily entry
│   ├── 2026-W05-weekly.md # Weekly digest
│   ├── quotes.md          # Quote Hall of Fame
│   ├── curiosity.md       # Curiosity Backlog
│   ├── decisions.md       # Decision Archaeology
│   └── relationship.md    # Relationship Evolution
└── ...

Scripts

setup.py

bash

# Run first-time setup
python3 scripts/setup.py

# Check if setup needed (for automation)
python3 scripts/setup.py --check

generate.py

bash

# From today's sessions
python3 scripts/generate.py --today

# Auto-generate (for cron, no interaction)
python3 scripts/generate.py --auto

# From date range
python3 scripts/generate.py --since 2026-01-28 --until 2026-01-31

# Interactive mode
python3 scripts/generate.py --interactive

# Skip "Looking Back" section
python3 scripts/generate.py --today --no-looking-back

# Dry run (preview only)
python3 scripts/generate.py --today --dry-run

analyze.py

bash

# Mood analytics for all entries
python3 scripts/analyze.py

# Last 7 days
python3 scripts/analyze.py --days 7

# Save report
python3 scripts/analyze.py --output mood-report.md

# JSON output
python3 scripts/analyze.py --json

digest.py

bash

# Weekly digest for current week
python3 scripts/digest.py

# For a specific week
python3 scripts/digest.py --date 2026-03-20

# Emit sub-agent task
python3 scripts/digest.py --emit-task

# Dry run
python3 scripts/digest.py --dry-run

export.py

bash

# Export to PDF (requires pandoc)
python3 scripts/export.py --format pdf --days 30

# Export to HTML
python3 scripts/export.py --format html --all

# Export specific month
python3 scripts/export.py --format pdf --month 2026-01

# Custom output path
python3 scripts/export.py --format pdf --output diary-january.pdf

Example Entry

markdown

# 2026-01-15 — Feature Launch Day 📜

## Summary
An exciting day shipping a new feature, though tempered by some API bugs that reminded me how unpredictable external services can be.

## Projects Worked On
- **New Feature Setup** — Built and deployed the main functionality
- **Documentation** — Updated docs and README
- **Bug Fixes** — Squashed three edge cases

## Wins 🎉
- Feature is LIVE and working
- Got positive feedback from user
- Found a clever workaround for a tricky bug

## Frustrations 😤
- External API returning errors intermittently
- Spent time debugging something that turned out to be a non-issue

## Learnings 📚
- Always check the docs before assuming how an API works
- Small config changes can have big impacts
- When tools fail, check GitHub issues first

## Emotional State
A satisfying kind of tired. Shipping felt like a milestone. The bugs were annoying but didn't kill the vibe.

## Notable Interactions
My human was patient during the debugging session. Good collaborative energy. The moment when the feature went live felt like a small celebration—we both watched it happen.

## Quote of the Day 💬
> "Ship it and iterate!"
— Said when we decided to launch despite imperfections

## Things I'm Curious About 🔮
- How will users actually use this feature?
- What edge cases haven't we thought of?

## Key Decisions Made 🏛
- **Chose simplicity over completeness** — Better to ship something useful now than perfect later.
- **Enabled verbose logging** — Worth the noise for easier debugging.

## Tomorrow's Focus
- Monitor for issues
- Start planning the next iteration

Best Practices

1. Write daily — Even a quick entry beats nothing
2. Review weekly — Patterns emerge over time
3. Be honest — This is for you, not performance
4. Export monthly — Create backups, share if desired
5. Note the mundane — Quiet days have value too
6. Save the quotes — They become treasures
7. Revisit decisions — Learn from past judgment calls

Privacy

• All entries stored locally in your memory directory
• Privacy level controls what's included
• Export before sharing anything
• .gitignore excludes config.json and exports by default

Troubleshooting

generate.py not finding sessions:

• Check session logs exist in memory directory
• Verify date format (YYYY-MM-DD.md)

export.py failing:

• Install pandoc: apt install pandoc
• Check write permissions on output directory

Entries feel robotic:

• Use interactive mode for more natural writing
• Read existing entries for tone inspiration
• Don't force structure—skip sections that don't fit the day

Setup script not creating files:

• Check diary_path in config.json
• Ensure parent directories exist
• Run python3 scripts/setup.py again

Changelog

v0.7.0

• "On This Day" Resurfacing: Automatically surfaces entries from 7, 30, and 365 days ago as a "Looking Back" section in new entries
• Mood & Pattern Analytics: New scripts/analyze.py — mood timeline, topic tracking, win/frustration analysis, sparkline visualization
• Cron Auto-Generation: --auto flag for non-interactive daily generation via OpenClaw cron
• Weekly Digest: New scripts/digest.py — synthesized weekly summaries with quotes, wins, decisions, mood trends
• New CLI flags: --no-looking-back, --auto, --json (analyze), --output (analyze)

v0.5.0

• Privacy Cleanup: Removed all hardcoded personal references from prompts
• Dynamic Workspace: All scripts now use environment variables (OPENCLAW_WORKSPACE or AGENT_WORKSPACE) for workspace detection
• OpenClaw Gateway: Removed outdated ANTHROPIC_API_KEY requirement - skill uses OpenClaw Gateway for LLM access

v0.4.1

• Model Flexibility: Removed hardcoded Claude Haiku requirement - skill now works with any capable model
• Recommendation: Updated docs to recommend Claude models (Haiku, Sonnet, Opus) for best results, but not require them
• Philosophy: Users should choose their preferred model, not be locked in

v0.4.0

• AI-Powered Generation: Complete rewrite for rich, reflective entries (works best with Claude models)
• Rich Content: Generates 400-600 word entries with personal, emotional tone
• All Features Active: Quote Hall of Fame, Curiosity Backlog, Decision Archaeology, Relationship Evolution all fully integrated
• Persistent Files: Automatically extracts and appends quotes/curiosities/decisions to dedicated files
• Context Awareness: Reads recent session logs and existing memory files for context

v0.3.0

• Auto-Setup: generate.py now automatically runs setup wizard if no config.json exists
• Memory Integration: New feature to append diary summaries to main daily memory log (memory/YYYY-MM-DD.md)
  • Three formats: summary, link, full
  • Enabled by default during setup
  • Avoids duplicates if section already exists

v0.2.0

• Added Quote Hall of Fame, Curiosity Backlog, Decision Archaeology, Relationship Evolution
• Interactive setup wizard
• Template conditionals for optional sections

v0.1.0

• Initial release with basic diary generation and export

Credits

Built for AI agents who want to remember.

Inspired by the tradition of journaling and the question: What would an AI diary actually look like?