chain-load
Load a chain to restore context from previous work. Use anytime when working on related topics - proactively search and suggest relevant chains.
When & Why to Use This Skill
The chain-load Claude skill is a powerful context restoration tool designed to maintain workflow continuity across multiple sessions. By retrieving historical 'chains' of work, it allows the AI to access previous summaries, key decisions, and code change histories, ensuring that users can pick up exactly where they left off without losing critical project intelligence.
Use Cases
- Resuming complex development tasks: Quickly restore the full context of a specific feature or bug fix by loading its dedicated work chain, including past rationales and next steps.
- Seamless context switching: Use fuzzy search to find and load relevant historical context when moving between different projects or topics within the same workspace.
- Proactive knowledge surfacing: Leverage the agent's ability to automatically detect related topics and suggest relevant memory chains to inform current decision-making.
- Project auditing and review: Access a structured history of a project's evolution, including previous summaries and outstanding issues, to ensure alignment with long-term goals.
| name | chain-load |
|---|---|
| description | Load a chain to restore context from previous work. Use anytime when working on related topics - proactively search and suggest relevant chains. |
Chain Load - Load Work Context
Load chain links to restore context. Can be used anytime, not just at session start.
Usage
/chain-load [chain-name]
/chain-load [search-query]
Proactive Behavior
When working on a task, proactively check for relevant context:
- Use
tinymem_chain_searchto find chains matching current work topic - Use
tinymem_searchto find individual memory segments - Suggest loading relevant chains or segments to the user
Segments vs Chains:
- Segments (memories): Individual insights via
tinymem_search+tinymem_get - Chains: Full workflow context via
tinymem_chain_load
Instructions
Find Chain:
- If exact chain-name provided → use
tinymem_chain_load - If query provided → use
tinymem_chain_searchfor fuzzy matching - If nothing provided → use
tinymem_chain_listto show all
- If exact chain-name provided → use
Load Chain Links - Call
tinymem_chain_load:chain_name: the chain to loadlimit: 5 (default, most recent links)
Parse and Present Context - Extract from XML:
- Previous summaries
- Decisions with rationale
- Code changes history
- Outstanding issues
- Next steps
Output Format
## Chain: [name] ([count] links)
### Latest: [slug] - [date]
<summary and next-steps from most recent link>
### Context
<key decisions and history from older links>
### Suggested Action
[Next step based on chain state]
Examples
User: /chain-load auth-feature
→ Loads auth-feature chain
User: /chain-load auth
→ Searches chains matching "auth"
User working on auth code (proactive): → Agent: "Found chain 'auth-feature' with 3 links. Load context?"