← Back to Skills Marketplace
jianglingling007

Session Relay

by jianglingling007 · GitHub ↗ · v1.2.0 · MIT-0
cross-platform ⚠ suspicious
120
Downloads
0
Stars
0
Active Installs
3
Versions
Install in OpenClaw
/install session-relay
Description
Automatic session continuity and task handoff across context window boundaries. Use when: (1) context token usage exceeds 70% of the model's context window,...
Usage Guidance
This skill does what it says (pulls recent sessions and saves snapshots) but it saves verbatim conversation text, commands, file paths and reproduction steps to disk. That makes it easy to accidentally capture API keys, passwords, private URLs, or other secrets even though the document says to 'strip' them — stripping is not enforced. Before installing, consider: (1) restrict or review what session content can be fetched (prefer summaries instead of verbatim tails), (2) require user confirmation before automatic startup fetches, (3) ensure snapshot files are stored in an encrypted/private workspace and not uploaded or backed up publicly, (4) audit the SKILL.md for the Unicode-control-chars finding and remove any suspicious characters, and (5) test in a sandboxed agent with no sensitive data to verify sanitization works. If you cannot guarantee secure storage and reliable secret-stripping, treat this skill as higher-risk and avoid enabling it for sessions containing credentials or private data.
Capability Analysis
Type: OpenClaw Skill Name: session-relay Version: 1.2.0 The 'session-relay' skill is a utility designed for session continuity and task handoff by monitoring context usage and saving state snapshots to the local workspace. It utilizes standard environment tools (sessions_list, sessions_history) to retrieve recent conversation history and includes explicit instructions and checklists in SKILL.md and snapshot-format.md to strip sensitive information like API keys or tokens before storage. The included shell script (archive-snapshot.sh) performs basic local file management, and there is no evidence of data exfiltration, unauthorized remote execution, or malicious prompt injection.
Capability Assessment
Purpose & Capability
The skill's name and description align with its actions: it pulls recent session history and writes snapshots for continuity. The included archive script and snapshot format reference are coherent with a relay/snapshot feature. No unrelated credentials, binaries, or install steps are requested.
Instruction Scope
The SKILL.md instructs the agent to fetch verbatim conversation tails (user+assistant) and to save reproduction commands, file paths, URLs, and error messages into workspace/memory/relay-snapshot.md. Although it states 'No secrets' and 'Strip any API keys/tokens that appear in conversation,' that sanitization is left to the agent and cannot be enforced. Writing verbatim history and commands to disk increases the risk of accidental secret capture and exfiltration. It also mandates performing the fetch 'before doing anything else' on startup, which broadens the action scope and could run automatically if the agent is allowed to invoke the skill.
Install Mechanism
No install spec (instruction-only) and a small, transparent shell script for archiving snapshots. Nothing is downloaded from external URLs and no archives are extracted. Low install risk.
Credentials
The skill requests no environment variables, credentials, or external config paths. That is proportionate to its purpose. However, its behavior of capturing verbatim messages and reproduction commands means it may inadvertently capture secrets present in conversation content despite claiming not to.
Persistence & Privilege
always:false (not force-included). The skill prescribes automatic behavior at session startup (fetch history before doing anything else) which implies it expects autonomous invocation; that is allowed by default but increases privacy exposure if the agent invokes it without explicit user consent. The skill writes snapshot files into {workspace}/memory and archives them — this is normal for persistence but means local disk storage must be trusted and protected.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install session-relay
  3. After installation, invoke the skill by name or use /session-relay
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.2.0
v1.2: Conversation tail inheritance — stores verbatim last 3 rounds from each of the 3 most recent sessions, not just summaries. New session can see exactly what was said, not a lossy digest.
v1.1.0
v1.1: Added Phase 0 — auto-inherit last 3 session histories on startup via sessions_list/sessions_history. Unified briefing merges live history + snapshot file. No more cold starts.
v1.0.0
v1.0: Rich snapshots with actionable detail, incremental saves from 60%, auto-briefing on restore, HEARTBEAT.md integration, daily log sync
Metadata
Slug session-relay
Version 1.2.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 3
Frequently Asked Questions

What is Session Relay?

Automatic session continuity and task handoff across context window boundaries. Use when: (1) context token usage exceeds 70% of the model's context window,... It is an AI Agent Skill for Claude Code / OpenClaw, with 120 downloads so far.

How do I install Session Relay?

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

Is Session Relay free?

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

Which platforms does Session Relay support?

Session Relay is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Session Relay?

It is built and maintained by jianglingling007 (@jianglingling007); the current version is v1.2.0.

💬 Comments