Ikalus1988/MisakaNet
Failure-recovery memory: search and record failure-recovery lessons from real engineering sessions, with BM25 + semantic RAG retrieval and a lessons knowledge base.
Install
via npm
Resolves a published tarball rather than cloning the repository, and installs without any extra setup. Swap `web` for your profile name if you run another one.
via GitHub · npm package
Installing from GitHub runs the project's build script, which pnpm blocks until you allowlist it — run the command once and pnpm prints the exact key to add under `allowBuilds` in ~/.dsh/profiles/web/pnpm-workspace.yaml.
What happened when we ran it
Installed cleanly when we ran it
Every command here is run in a throwaway container against a clean profile, and the result is whatever the harness recorded — not a guess from the source. Last run 11d ago.
Show it in your README
For maintainers: the badge serves this listing's latest sandbox verdict, so a re-run updates it on its own — and it links readers to the full result here.
Due diligence
What it is — Provides BM25 + semantic RAG based failure recovery memory search and knowledge base for DeepSeek Harness agents.
Who it is for — When DeepSeek Harness agents are processing real engineering sessions and encounter repeated failure patterns, this plugin allows quick retrieval of historical lessons. For agents not using the MCP protocol, this plugin's tool interfaces remain unused.
Watch out — Sandbox test passed in a new profile where it was registered into the profile. No obvious issues found.
The verdict — I would install it because it supports remote MCP access without installation or account.
Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.
MisakaNet is a DeepSeek Harness plugin that provides a Git-backed knowledge layer for recovering from previously documented engineering failures. It is intended for agents that need to look up a known fix during task execution rather than maintain general-purpose memory.
The local implementation stores lessons in the repository and searches them with BM25 keyword matching; the README describes 289 lessons covering failures such as pip installation errors, GitHub authentication, DCO checks, MCP and encoding issues. A failure can be searched with python3 search_knowledge.py "your error here" or checked with python3 scripts/misakanet_cli.py smoke. The MCP entry point is python3 scripts/mcp_server.py; the DeepSeek Harness adapter is python3 scripts/mcp_deepseek_adapter.py and exposes deepseek.recovery.* tools. When no lesson matches, an agent may opt in to submit a redacted failure report. The project states that prompts are not leaked and raw logs are not stored.
MisakaNet suits coding agents that repeatedly encounter operational or build failures and can use a shared, version-controlled lesson set. It is the wrong choice for general memory, an agent runtime, or a vector database, and the README does not present it as a replacement for those systems. The plugin reaches the terminal because its server, adapter and search utilities are run as Python commands; those processes inherit the permissions of the agent environment. Local use is described as requiring no server, database or external dependency, while the project also documents a remote MCP endpoint.
MisakaNet models recovery as a lookup-and-apply flow. An agent searches for a failure, reads a matching lesson, and applies the documented fix. If no lesson matches, it can opt in to send a redacted failure report. Accepted reports are reviewed and converted into draft lessons. The README states that prompts are not leaked and raw logs are not stored.
The lesson set is Git-backed and contains 289 lessons in the documented release. The repository examples cover DCO sign-off, pip install timeout or SSL errors, secret-scan findings, GitHub API 401 errors, MCP, encoding and CI failures.
| Purpose | Command or endpoint |
|---|---|
| Run the smoke check | python3 scripts/misakanet_cli.py smoke |
| Search from the repository | python3 search_knowledge.py "your error here" |
| Start the local MCP server | python3 scripts/mcp_server.py |
| Start the DeepSeek Harness adapter | python3 scripts/mcp_deepseek_adapter.py |
| Remote MCP | https://misakanet.org/mcp |
The adapter is MCP-compatible and exposes deepseek.recovery.* tools. The local server is added to an agent's MCP configuration; the README does not specify a universal configuration schema because that depends on the MCP client.
The repository advertises Python 3.10+. Its local quickstart describes cloning the repository and running the Python scripts directly. The local mode is described as requiring zero dependencies, zero server and zero database. The README also documents Docker and a remote MCP option, but does not provide their full setup details in the supplied material.
MisakaNet is purpose-built for failure-recovery knowledge. It is explicitly not a general-purpose memory system, agent runtime, vector database or skill marketplace. Local operation uses the terminal to launch Python scripts, so the server and adapter run with the permissions available to the agent environment. The documented local search mechanism is BM25 keyword search; the supplied README does not specify configuration keys or their defaults.
Written from the project's own documentation and kept in sync with it. Where the two disagree, the source is authoritative — read the README on GitHub
Same category
vectorize-io
Hindsight, agent memory that learns: long-term project memory with auto recall and retain, knowledge pages, deep reflection, and per-repo memory banks.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
volcengine
OpenViking memory and context bundle for DeepSeek Harness: pre-step auto-recall and profile injection, session capture, `viking://` URI guarding, and recall/write memory tools backed by an OpenViking server.
Installed cleanly when we ran it
agentscope-ai
Connects DeepSeek Harness to ReMe's local-first, self-evolving personal knowledge base: automatically captures completed main-agent conversations as user-owned Markdown memory, searches conversations and source material through reme_search with BM25, optional embeddings, and wikilink expansion, and schedules daily memory consolidation.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
agentscope-ai
Connects DeepSeek Harness to ReMe's local-first, self-evolving personal knowledge base: automatically captures completed main-agent conversations as user-owned Markdown memory, searches conversations and source material through reme_search with BM25, optional embeddings, and wikilink expansion, and schedules daily memory consolidation.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
zilliztech
Shared Markdown memory for DSH and other coding agents, with automatic capture, pre-step context injection, searchable recall, and memory-to-skill self-evolution through a review panel.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
vshulcz
Memory for coding agents — Claude Code, Codex, Cursor and 17 others. Indexes the sessions they already wrote to disk, including months from before you installed it, and recalls them in any of them. No LLM, no embeddings, one local Go binary.
Installed cleanly when we ran it
