← Back to Skills Marketplace
3184
Downloads
0
Stars
1
Active Installs
18
Versions
Install in OpenClaw
/install phoenixclaw
Description
Passive journaling skill that scans daily conversations from ALL session paths (main, agents, cron) via cron to generate markdown journals using semantic und...
Usage Guidance
Before installing, consider the following: 1) This skill will scan all OpenClaw session logs and media (including agent/cron sessions) and copy images into a journal assets folder — that may include sensitive messages, receipts, or GPS metadata. 2) If you want journaling but with tighter privacy, change the default journal path to a safe location, limit which session directories are scanned, and disable cron registration initially. 3) Review the two included scripts (rolling-journal.js and session-day-audit.js) for the shell execSync calls and ensure inputs (paths, filenames, log contents) cannot be attacker-controlled; consider removing or sandboxing shell fallbacks. 4) Plugins: PhoenixClaw will discover and run plugins that declare depends: phoenixclaw; only enable trusted plugins (finance/ledger plugins will see payment screenshots). 5) If you proceed, test in an isolated account or VM, validate config.yaml before the skill runs, and consider stripping EXIF/GPS from images or disabling image extraction if you do not want location metadata captured. 6) If you need a cleaner security posture, ask the author to (a) explicitly declare env vars and required paths, (b) avoid executing shell fallbacks, and (c) document exact plugin discovery and permission model.
Capability Analysis
Type: OpenClaw Skill
Name: phoenixclaw
Version: 0.0.19
PhoenixClaw is a comprehensive journaling skill designed to aggregate user activity and media from various OpenClaw session logs. It utilizes Node.js and shell scripts (e.g., `scripts/rolling-journal.js` and `references/session-day-audit.js`) to scan local directories like `~/.openclaw/sessions` and `~/.agent/sessions`. While the skill employs high-privilege capabilities—including recursive file system scanning, shell execution for file management, and Python one-liners for timezone math—these actions are transparently documented and strictly aligned with the stated purpose of 'passive journaling.' There are no indicators of data exfiltration, obfuscation, or unauthorized remote communication.
Capability Assessment
Purpose & Capability
The skill's stated purpose (passive journaling from session logs + memory) matches the code and SKILL.md: it enumerates OpenClaw session and memory locations, extracts messages and images, and writes Markdown journals. The required read/write paths in _meta.json align with journaling. Minor oddity: default journalPath in code points to /mnt/synology/... (an external mount) which is a surprising default but can be overridden via config.
Instruction Scope
The runtime instructions explicitly require scanning ALL session directories (main, agents, cron) and per-message timestamps, extracting and copying image files, preserving EXIF/GPS, and executing plugins at multiple hook points. This broad scan will read other agents' session logs and media, and the mandatory 9-step 'NEVER skip' workflow + automatic plugin execution increases the amount of user data processed and shared with plugins. The instructions also recommend registering a cron job that runs nightly and must have read/write access to many user paths.
Install Mechanism
No install spec is provided (instruction-only plus included scripts), so nothing is being downloaded from external URLs during installation. The code files are included in the skill bundle, which reduces supply-chain download risk. However, the included node scripts execute shell fallbacks (execSync) which has runtime risk if inputs are not carefully sanitized.
Credentials
The skill declares no required env vars, but scripts read environment variables (PHOENIXCLAW_JOURNAL_PATH, OPENCLAW_SESSIONS_PATH, TARGET_TZ) and a config at ~/.phoenixclaw/config.yaml — these are not listed as required in metadata. The manifest writes to many user directories and the code's default journal path points to an external mount. The skill also preserves EXIF/GPS from images and routes finance screenshots to a Ledger plugin: both are sensitive and justify explicit, clear consent and narrower scope.
Persistence & Privilege
The skill explicitly instructs registering a nightly cron job (openclaw cron add) that will run automatically and scan all sessions and media. While autonomous invocation is the platform default, combining scheduled runs + broad read/write access + plugin execution increases the blast radius if misconfigured or if a plugin is malicious. The skill does not set always:true, but cron registration implies persistent background activity and system-wide file access.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install phoenixclaw - After installation, invoke the skill by name or use
/phoenixclaw - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.0.19
- Improved session log scanning: now recursively scans all session directories (main, agents, cron) to capture multi-agent data.
- Enhanced session file handling: includes all session files regardless of naming or directory depth.
- Updated documentation to reflect multi-agent architecture support and ensure no conversational data is missed.
- No code changes detected; changes are documentation-only for clarity and completeness.
v0.0.18
- Version bump from 0.0.15 to 0.0.18.
- No code or documentation changes detected except for the updated version number in SKILL.md.
- All workflows and descriptions remain unchanged.
v0.0.17
No functional or documentation changes detected in this release.
- Version bump only; no updates to the code or SKILL.md content.
- All workflows, instructions, and feature descriptions remain unchanged.
v0.0.16
No changes detected in this version.
- Version bumped from 0.0.15 to 0.0.16, but no file or documentation changes are present.
v0.0.15
- Expanded session log scanning to include additional OpenClaw directories for more comprehensive data coverage.
- Updated log filtering instructions to read from all known session, agent, and cron directories.
- No code changes were detected in this version.
- Documentation improved for greater clarity around context retrieval and session file location handling.
v0.0.14
**PhoenixClaw v0.0.14 Changelog**
- Added `_meta.json` metadata file and a new script at `scripts/rolling-journal.js`.
- Updated the workflow to explicitly filter out system heartbeat and cron messages from session logs to avoid non-user activity in journals.
- Introduced strict handling for finance/payment screenshots: these are now detected, excluded from the main journal, and routed directly to the Ledger plugin if enabled.
- Clarified the mandatory exclusion of payment-related images from regular journal image sections, unless explicitly described as part of a life moment.
- Improved documentation of session log filtering, including example `jq` filters and detection/handling criteria for finance screenshots.
- Incremented SKILL.md version metadata to 0.0.15.
v0.0.13
Version 0.0.13
- Added support for reading `memory/YYYY-MM-DD.md` and `memory/YYYY-MM-DD-*.md` files to capture manually recorded daily reflections.
- Daily context retrieval now merges explicit user memories from memory files with conversational and media context from session logs, ensuring more complete journal entries.
- Memory files are now treated as a critical source for user thoughts and emotions, providing richer and more accurate journal content.
- Clarified that explicit memory files must not be skipped, as they may contain important details missing from session logs.
v0.0.12
**Changelog for phoenixclaw v0.0.12**
- Improved image linking in journals: image paths are now dynamically generated relative to the journal file, avoiding hardcoded path depths and always matching the actual copied asset filenames.
- Updated daily journal generation instructions to ensure links use the copied filename as the authoritative source.
- Clarified that absolute paths are never output; all image links compute their path from the daily journal location and asset paths.
- No code changes detected; documentation updates only.
v0.0.11
Version 0.0.11
- Added a new reference file: `references/session-day-audit.js`.
- Updated session log scanning: workflow now aggregates logs from both `~/.openclaw/sessions` and `~/.agent/sessions`, with timezone-aware filtering to avoid UTC/local day mismatches.
- Improved filtering of session log entries for better accuracy across filename and timestamp boundaries.
- Clarified that heartbeat/cron noise is ignored; only user/assistant content and real media are considered for journal moments.
v0.0.10
- Improved context retrieval by filtering session log messages and images by message timestamp, instead of relying on session file modification time.
- Enhanced edge case handling for late-night activity by expanding timestamp filtering to include spillover into adjacent days.
- Ensured only images with matching timestamps are included in daily journals, increasing accuracy and relevance.
- Updated documentation to clarify new log scanning and image filtering procedures.
- Bumped version to 0.0.10.
v0.0.8
- Improved session log discovery: now scans files by modification time to ensure all relevant conversations are captured, regardless of filename.
- Added specific edge case handling for late-night activity spanning midnight, ensuring completeness across date boundaries.
- Clarified that daily context reconstruction updates the day's memory, not just general memory.
- Minor language clarifications to emphasize that ALL matching session files modified today must be read for accurate journaling.
v0.0.7
**Enhanced image handling and journal embedding workflow.**
- Now extracts all image entries from session logs, saves them in date-based asset folders, and generates descriptive names when possible.
- Adds critical image processing: generates alt-text via Vision Analysis, categorizes images, and links them to specific journal moments.
- Journal generation now curates embedded images (not all), prioritizing highlights and moments; finance screenshots are routed into Ledger sections.
- Documentation for context retrieval and journal assembly expanded for clarity on image integration.
v0.0.6
- Version updated to 0.0.6 in metadata.
- No code or workflow changes; documentation now defines version solely under metadata.
- All core features, workflow, triggers, and documentation references remain unchanged.
v0.0.5
Major update: Strict enforcement of the complete workflow and plugin system for journal generation.
- Added a critical 9-step workflow that must always be fully executed, both for cron and manual invocation.
- Mandatory scanning of all session logs alongside memory to ensure photos and media are always included.
- Introduced plugin execution at specific hook points, with explicit plugin lifecycle requirements.
- Added detailed instructions and warnings to prevent partial or incorrect pipeline execution.
- Extended documentation to cover the new plugin protocol and reinforced the requirement for complete pipeline runs on all journal generation requests.
v0.0.4
- Improved context retrieval by always scanning raw session logs to capture in-progress activity, even if daily memory exists.
- Context reconstruction now uses session logs when memory is missing or sparse, ensuring more comprehensive journal entries.
- Adjusted fallback logic: session logs supplement daily memory rather than only activating when memory is missing.
- Minor workflow description clarifications for increased transparency.
v0.0.3
Version 0.0.3
- Added documentation: Introduced references/plugin-protocol.md to provide new protocol reference material.
- Updated metadata: SKILL.md now includes an explicit version field ("version: 0.0.3").
v0.0.2
- Adds logic to gracefully handle missing memory embeddings provider: if unavailable, skips cross-day pattern search and relies only on current daily memory and session-log fallback.
- No other changes detected.
v0.0.1
Initial release of PhoenixClaw, a passive journaling skill that creates semantic daily journals from conversations.
- Automatically generates daily Markdown journals by scanning daily conversations at 10:00 PM (local time) via cron.
- Identifies journal-worthy moments, patterns, mood changes, milestones, and growth opportunities.
- Summarizes activities without user intervention, but supports explicit user requests for journal review, summaries, pattern analysis, and goal tracking.
- Integrates structured templates and maintains a timeline, growth map, and evolving user profile.
- Ensures compatibility with personal knowledge management tools like Obsidian.
Metadata
Frequently Asked Questions
What is phoenixclaw?
Passive journaling skill that scans daily conversations from ALL session paths (main, agents, cron) via cron to generate markdown journals using semantic und... It is an AI Agent Skill for Claude Code / OpenClaw, with 3184 downloads so far.
How do I install phoenixclaw?
Run "/install phoenixclaw" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is phoenixclaw free?
Yes, phoenixclaw is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does phoenixclaw support?
phoenixclaw is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created phoenixclaw?
It is built and maintained by betterest (@goforu); the current version is v0.0.19.
More Skills