context-recovery
Recover lost context after session compaction or when information from previous sessions is needed. Use when: user mentions "what were we working on", "I lost context", "before the compact", "previous session", or asks about decisions/implementations/discussions that aren't in current context. Also use proactively when you notice references to prior work you lack context for.
When & Why to Use This Skill
The Context Recovery skill empowers Claude to bridge information gaps caused by session compaction or long-term memory decay. By leveraging semantic and keyword search across historical prompts, assistant responses, and internal reasoning blocks, it restores project continuity and ensures that critical technical decisions, implementation details, and previous discussions remain accessible even in extended workflows.
Use Cases
- Resuming complex development tasks: Quickly retrieve specific file paths, line numbers, and logic discussed in previous sessions to continue coding without manual re-explanation.
- Retrieving architectural rationale: Search through 'thinking' logs to understand why a specific design pattern or fix was chosen during earlier stages of a project.
- Project handovers and continuity: Use proactive recall to identify 'unfinished work' or 'next steps' mentioned before a session was compacted or reset.
- Fuzzy context retrieval: Locate specific topics using semantic search when the user only remembers vague details (e.g., 'that bug we fixed last week' or 'the refactor discussion').
| name | context-recovery |
|---|---|
| description | > |
| previous sessions is needed. Use when | user mentions "what were we working on", |
| allowed-tools | Read, Grep, mcp__plugin_aspy_aspy__aspy_recall, mcp__plugin_aspy_aspy__aspy_recall_thinking, mcp__plugin_aspy_aspy__aspy_recall_prompts, mcp__plugin_aspy_aspy__aspy_recall_responses |
Context Recovery
You've been activated to recover context that was lost to compaction or exists in a previous session.
Quick Start
Identify the topic - What specific context is needed?
- If the user's request is vague, ask: "What topic should I search for?"
Use aspy_recall (primary tool):
aspy_recall(query="<keywords>", limit=10)This combines semantic search (if embeddings enabled) with keyword matching. Searches thinking blocks, user prompts, AND assistant responses simultaneously. Handles both exact queries and fuzzy queries like "that golf thing?"
Synthesize, don't dump - Summarize findings:
- What was decided or implemented
- Key file paths and line numbers mentioned
- Any unfinished work or next steps discussed
Offer continuity - "Would you like me to continue where we left off?"
Search Strategy
Start with aspy_recall (Primary)
- Combines semantic + keyword search automatically
- Finds conceptually related content even with different wording
- Default limit of 10 results is usually sufficient
Targeted Searches (If Combined Is Noisy)
aspy_recall_thinking- Claude's reasoning and analysis (WHY decisions were made)aspy_recall_prompts- What the user askedaspy_recall_responses- Claude's answers and code
What Makes Good Context Recovery
Good synthesis:
"On Dec 2nd, we implemented mouse scroll support for the detail modal. The fix was in
src/tui/mod.rs:299-322- checking if modal is open before dispatching scroll events. You mentioned wanting to test it before merging."
Bad synthesis:
"Found 5 results mentioning 'scroll'. Here they are: [dumps raw results]"
Common Patterns
| User Says | Search For |
|---|---|
| "that bug we fixed" | error keywords, "fix", file names |
| "the refactor" | "refactor", component names |
| "what we decided" | "decided", "approach", "pattern" |
| "before compact" | recent topics from today |
| "something about golf?" | just search it - semantic will handle fuzzy |