tide-watch_skill
- Python
2.6k
GitHub Stars
28
Bundled Files
2 months ago
Catalog Refreshed
3 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill tide-watch- _meta.json986 B
- async-gateway-proposal.md6.5 KB
- CHANGELOG.md15.5 KB
- CLAWHUB-SCANNER-RESPONSE.md10.5 KB
- INSTALLATION.md7.0 KB
- package.json1.2 KB
- PARSING.md8.7 KB
- README.md18.2 KB
- RESEARCH-ANYBINS-PATTERN.md4.2 KB
- SECURITY-ADVISORY-CVE-2026-001.md8.5 KB
- SECURITY-ANALYSIS.md8.1 KB
- SECURITY-ASSESSMENT-ISSUE-35.md7.0 KB
- SECURITY-ASSESSMENT-ISSUE-36.md8.4 KB
- SECURITY-ASSESSMENT-v1.1.0.md8.8 KB
- SECURITY-ASSESSMENT-v1.1.2.md7.9 KB
- SECURITY-ASSESSMENT-v1.1.3.md6.6 KB
- SECURITY-ASSESSMENT-v1.1.4.md8.1 KB
- SECURITY-ASSESSMENT-v1.1.5.md1.9 KB
- SECURITY-ASSESSMENT-v1.1.6.md9.1 KB
- SECURITY-ASSESSMENT-v1.2.0.md12.4 KB
- SECURITY-ASSESSMENT-v1.2.1.md8.4 KB
- SECURITY-ASSESSMENT-v1.3.0.md7.8 KB
- SECURITY-ASSESSMENT-v1.3.2.md6.3 KB
- SECURITY-ASSESSMENT-v1.3.3.md6.6 KB
- SECURITY-ASSESSMENT-v1.3.4.md9.4 KB
- SECURITY-ASSESSMENT-v1.3.6.md7.1 KB
- SECURITY-SCAN.md12.5 KB
- SKILL.md17.9 KB
Overview
This skill provides proactive session capacity monitoring and management for OpenClaw to prevent context window lockups. It warns at configurable thresholds (75%, 85%, 90%, 95%), automatically backs up sessions before resets, and manages session resumption prompts. It supports a directives-only safe mode and optional CLI tools for manual operations. Works with any model or provider.
How this skill works
Tide Watch parses agent configuration from AGENTS.md and runs scheduled checks via HEARTBEAT.md or on demand. It calls the session_status tool to compute token usage and compares the percentage against configured thresholds, issuing one-time warnings per threshold. When backup triggers are configured, it creates verified backups before resets, stores resumption prompts, and cleans up old backups per retention settings. Optional CLI tools expose status, dashboards, archive management, and resume-prompt editing for manual control.
When to use it
- Working on long-running projects that risk filling context windows
- Managing multiple conversation channels (Discord, Telegram, webchat) concurrently
- Preventing lost work when sessions approach full capacity
- When you need automated backups and resumption prompts before resets
- Performing manual audits or cross-session capacity checks with CLI tools
Best practices
- Use Directives-Only mode as the default to avoid installing code
- Set conservative thresholds and check frequency to match your workflow
- Enable auto-backup for high-severity thresholds (default: 90%, 95%)
- Keep retention reasonable (default: 7 days) and enable compression if disk space is limited
- If using CLI tools, verify version 1.0.1+ and inspect code before installing
Example use cases
- Automatically warn and backup a Discord channel before it hits 95% to avoid losing context
- Hourly heartbeat that monitors multiple sessions and only notifies when thresholds are crossed
- Use CLI dashboard to identify sessions across agents that are near capacity and archive old conversations
- Create and store session resumption prompts so collaborators can resume work cleanly after a reset
- Run manual capacity checks before a long debugging session or large batch task
FAQ
No. The recommended Directives-Only mode requires no code installation and runs via agent directives and heartbeats.
What should I do if I installed an early CLI version?
If you installed v1.0.0, update immediately to v1.0.1 or later; v1.0.0 had a patched shell-injection vulnerability.
Where are backups stored and how long are they kept?
Backups are stored under ~/.openclaw/agents/main/sessions/backups/ and retention defaults to 7 days, configurable in AGENTS.md.
