devlog
Persistent memory system for AI-assisted development
claude mcp add --transport stdio codervisor-devlog ./bin/devlog start
How to use
Devlog is a lightweight Rust daemon designed to collect events from various AI coding agents and forward them to configurable remote endpoints. It auto-discovers AI agent log locations, monitors those logs in real time, parses events with agent-specific adapters, and streams them to a chosen backend. When the network is unavailable, events are buffered in a local SQLite database and retried automatically. The system supports multiple agents (e.g., Copilot, Claude Code, Cursor) and can be extended with a generic JSONL adapter for custom agents.
To interact with the running daemon, use the built binaries exposed by the project. The standard workflow is to start the daemon, let it begin collecting events, and optionally backfill historical logs from agent logs to seed the buffer and ensure continuity. The daemon exposes a REST-like event format when sending data to backends, including fields like id, eventType, timestamp, agentId, sessionId, and data payloads for prompts or models. This makes it suitable for real-time analytics dashboards or log-forwarding pipelines.
How to install
Prerequisites:
- Rust 1.75+ (or the system used to build the binary released by the project)
- Make
Installation steps (from source):
# Clone the repository
git clone https://github.com/yourorg/devlog.git
cd devlog
# Build the project (produces bin/devlog or similar binary)
make build
# Optional: install to system path
make install
Alternatively, if you prefer to run from the built binary directly:
# After build, run the daemon
./bin/devlog start
Configuration guidance:
- Create or edit the configuration file at ~/.devlog/collector.json according to the example in the repository's README.
- Ensure environment variables referenced in the config (e.g., ${DEVLOG_API_KEY}) are exported in the environment where the daemon runs.
If you want to run via Docker, you can build and run the container as described in the Docker section of the README, mapping the required config and log directories into the container.
Additional notes
Tips and common issues:
- The collector reads its settings from ~/.devlog/collector.json. Ensure you provide backendUrl and API key (or environment variable expansion) so the daemon can forward events.
- The buffer is SQLite-based and supports offline operation. Make sure the configured maxSize suits your disk space and expected event rate.
- Auto-discovery relies on the agent log locations. If an agent changes its log path, update the config or rely on the auto-discovery behavior.
- For backfill, the devlog backfill commands allow processing historical logs to seed the event stream. Use the backfill status command to monitor progress.
- When running in Docker, pass the necessary environment variables (DEVLOG_BACKEND_URL, DEVLOG_API_KEY) and mount volumes for the config and buffer data as described in the README.
- If you encounter permission issues when installing to /usr/local/bin, run with appropriate privileges or adjust the install path in the Makefile/configuration.
Related MCP Servers
PPTAgent
An Agentic Framework for Reflective PowerPoint Generation
moling
MoLing is a computer-use and browser-use based MCP server. It is a locally deployed, dependency-free office AI assistant.
MCP-PostgreSQL-Ops
🔍Professional MCP server for PostgreSQL operations & monitoring: 30+ extension-independent tools for performance analysis, table bloat detection, autovacuum monitoring, schema introspection, and database management. Supports PostgreSQL 12-17.
ContextPods
Model Context Protocol management suite/factory. An MCP that can generate and manage other local MCPs in multiple languages. Uses the official SDKs for code gen.
zerodha
Zerodha MCP Server & Client - AI Agent (w/Agno & w/Google ADK)
Common_Chronicle
Common Chronicle turns messy context into structured, sourced timelines.