MemRL
A memory-augmented reinforcement learning system for Claude Code that captures coding sessions as episodes and learns from experience to improve future task assistance. Built in Rust, it automatically captures session transcripts, extracts structured intent using LLM analysis, and stores episodes with git diffs and error resolutions in both file-based storage and vector databases using LanceDB and FastEmbed. The system implements Bellman equation-based utility propagation to spread value from helpful episodes to similar ones, uses temporal credit assignment to reward episodes that preceded successful outcomes, and provides semantic search through embeddings to retrieve relevant past experiences.
Composite of vulnerability cleanliness, spec conformance, provenance, stability, and usage signals — scanned and weighted by Cognium. Human and agent signals are tracked separately. Last scanned 2026-09-28.
Scan details: Circle-IR · 2026-09-28 · Appeal
View full trust & usage report →Metadata
- Version
- 1.0.0
- Skill type
- atomic
- Execution layer
- mcp-remote
- Category
- database
- Source
- PulseMCP
- Repository
- github.com/anvanster/tempera
- Author type
- human
- Last scanned
- 2026-09-28
- Updated
- 2026-09-28
Use via MCP
Resolve MemRL from your agent
Streamable HTTP transport at https://api.skillsregistry.net/mcp. No auth for read tools. Discovery: .well-known/mcp.json.
One command in your shell — Claude Code wires it up and verifies the connection. Run /mcp in any session to confirm.
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp--scope user for --scope project to commit it to .mcp.json.