← Back to Skills Marketplace
q012315

context-manager

by zone · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ suspicious
136
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install q012315-context-manager
Description
智能上下文管理系统 - 支持多模型自适应、分层记忆、动态注入、SQLite 数据库
Usage Guidance
This skill's documentation promises many features (search, compression, import/export, tagging, advanced DB schema, dynamic model injection) that the bundled script does not implement — the script only supports init, add, stats, and a simple auto-manage calculation. Before installing or running: (1) review the script yourself — it writes an SQLite DB to ~/.openclaw/workspace-telegram-bot1/memory/memories.db; (2) run it in a sandbox or isolated account if you want to test; (3) if you expect the advertised features, ask the author for a version that implements them or for clear release notes; (4) be cautious trusting SKILL.md claims (env vars and config paths are documented but unused). There are no signs of network exfiltration or credential access in the provided code, but the documentation/code mismatch is a red flag — treat as suspicious and verify behavior before giving it production access.
Capability Analysis
Type: OpenClaw Skill Name: q012315-context-manager Version: 1.0.1 The skill bundle provides a basic SQLite-based context management system for AI agents. The Python script (scripts/context-manager.py) implements simple database initialization, memory storage, and statistics reporting without any network access, sensitive data exfiltration, or dangerous execution patterns. While the documentation (SKILL.md) describes more advanced features like compression and search that are not yet fully implemented in the provided code, there is no evidence of malicious intent or prompt injection attacks.
Capability Assessment
Purpose & Capability
Name/description promise a full-featured context manager (multi-model adaptation, layered memory, dynamic injection, compression, import/export, search, indexing, management history). The included Python script implements only basic persistence and simple auto-management (init, add, stats, auto-manage). Required binary (python3) is appropriate, but many claimed capabilities are absent from the code and some DB fields in SKILL.md are not present in the actual schema. The workspace default path (~/.openclaw/workspace-telegram-bot1) is oddly specific and not explained.
Instruction Scope
SKILL.md documents many CLI commands and behaviors (--search, --export, --import, --compress, --load-for-task, --list-all, --delete, --update, --history, --reset) that the script does not implement. SKILL.md also suggests environment variables and a config file path (~/.openclaw/config/context-manager.json) that the script never reads. The instructions therefore give the agent broad expectations it cannot actually follow; this mismatch increases risk of surprising behavior or user confusion.
Install Mechanism
No install spec; skill is instruction + a Python script. That is low-risk compared with arbitrary remote downloads. The script writes an SQLite DB into a subdirectory in the user's home (~/.openclaw/...); creating files under the user's home is expected for this kind of tool but worth noting.
Credentials
SKILL.md suggests optional environment variables (OPENCLAW_MODEL, CONTEXT_LIMIT, AUTO_MANAGEMENT) and a config file, but the Python script does not read these env vars or config file. The metadata lists sqlite3 as a package (sqlite3 is part of Python stdlib). No credentials or secrets are requested. The inconsistency between advertised env/config usage and actual code is notable.
Persistence & Privilege
The skill is not 'always: true' and does not request elevated privileges. It creates a workspace and writes an SQLite DB under the user's home directory (~/.openclaw/workspace-telegram-bot1/memory/memories.db). That is expected for a local memory tool but the hard-coded workspace name (contains 'telegram-bot1') is unusual and may be a leftover; consider this suspicious but not privileged.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install q012315-context-manager
  3. After installation, invoke the skill by name or use /q012315-context-manager
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
v1.0.1 - No functional changes; documentation only. - SKILL.md updated, but with no substantive content difference from the previous version.
v1.0.0
- Initial release of the Context Manager skill. - Introduces multi-model adaptive context management with automatic context limit handling. - Implements a layered memory system (hot/warm/cold) stored in SQLite with fast search and indexing. - Supports dynamic memory injection and auto management strategies based on model context size. - Includes memory compression, merging, and archiving based on usage thresholds. - CLI operations for adding, searching, exporting, importing, and managing memories.
Metadata
Slug q012315-context-manager
Version 1.0.1
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 2
Frequently Asked Questions

What is context-manager?

智能上下文管理系统 - 支持多模型自适应、分层记忆、动态注入、SQLite 数据库. It is an AI Agent Skill for Claude Code / OpenClaw, with 136 downloads so far.

How do I install context-manager?

Run "/install q012315-context-manager" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is context-manager free?

Yes, context-manager is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does context-manager support?

context-manager is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created context-manager?

It is built and maintained by zone (@q012315); the current version is v1.0.1.

💬 Comments