Home > MCP Servers > roampal-core

roampal-core

Outcome-based persistent memory MCP server for Claude Code and OpenCode. Good advice promoted, bad advice demoted. pip install roampal.

agent-memory ai-assistant ai-coding ai-coding-assistant ai-memory ai-tools chromadb claude-code python

Why this rank:Strong adoptionRecent releaseHealthy release cadence

Description

Outcome-based persistent memory MCP server for Claude Code and OpenCode. Good advice promoted, bad advice demoted. pip install roampal.

README

Roampal — Outcome-Based Persistent Memory MCP Server

Two commands. Your AI coding assistant gets outcome-based memory.
Works with Claude Code and OpenCode.

Why?

AI coding assistants forget everything between sessions. You explain your architecture, your preferences, your conventions — again. When they give bad advice, there's no mechanism to learn from it.

Roampal is an MCP server that gives your AI persistent, outcome-based memory across every session. Good advice gets promoted. Bad advice gets demoted. Your AI learns what works and what doesn't — automatically, with zero workflow changes.

Benchmarks

85.8% on LoCoMo (non-adversarial, end-to-end answer accuracy) — validated on 1,986 questions across 10 conversations with dual grading.

Result	Score
Conversational learning vs raw ingestion	+23 points (76.6% vs 53.0%, p<0.0001)
Architecture vs model effect	Architecture ~10x larger contributor
Poison resilience (1,135 adversarial memories)	-2.6 to -4.2 points only
TagCascade retrieval (tags-first + CE rerank)	+1.9 Hit@1 vs pure CE (p<0.0001)

Benchmark pipeline runs on a single GPU with no cloud dependencies. Roampal itself runs on CPU — no GPU required. Full methodology, data, and evaluation scripts: roampal-labs

Paper: "Beyond Ingestion: What Conversational Memory Learning Reveals on a Corrected LoCoMo Benchmark" (Logan Teague, April 2026)

Quick Start

pip install roampal
roampal init

Auto-detects installed tools. Restart your editor and start chatting.

Target a specific tool: roampal init --claude-code or roampal init --opencode

Platform Differences

The core loop is identical — both platforms inject context, capture exchanges, and score outcomes. The delivery mechanism differs:

	Claude Code	OpenCode
Context injection	Hooks (stdout)	Plugin (system prompt)
Exchange capture	Stop hook	Plugin `session.idle` event
Scoring	Main LLM via `score_memories` tool	Independent sidecar (your chosen model > Zen free)
Self-healing	Hooks auto-restart server on failure	Plugin auto-restarts server on failure

Claude Code prompts the main LLM to score each exchange via the score_memories tool. OpenCode never self-scores — an independent sidecar (a separate API call) reviews each exchange as a third party, removing self-assessment bias. The score_memories tool is not registered on OpenCode. During roampal init or roampal sidecar setup, Roampal detects local models (Ollama, LM Studio, etc.) and lets you choose a scoring model. If configured, these take priority (Zen is skipped for privacy). A cheap or local model works great — scoring doesn't need a powerful model. Defaults to Zen free models (remote, best-effort) if you skip setup.

How It Works

When you type a message, Roampal automatically injects relevant context before your AI sees it:

You type:

fix the auth bug

Your AI sees:

═══ KNOWN CONTEXT ═══
• JWT refresh pattern fixed auth loop [id:patterns_a1b2] (3d, 90% proven, patterns)
• User prefers: never stage git changes [id:mb_c3d4] (memory_bank)
═══ END CONTEXT ═══

fix the auth bug

No manual calls. No workflow changes. It just works.

The Loop

You type a message
Roampal injects relevant context automatically (hooks in Claude Code, plugin in OpenCode)
AI responds with full awareness of your history, preferences, and what worked before
Outcome scored — good advice gets promoted, bad advice gets demoted
Repeat — the system gets smarter every exchange

Five Memory Collections

Collection	Purpose	Lifetime
`working`	Current session context	24h — promotes if useful, deleted otherwise
`history`	Past conversations	30 days, outcome-scored
`patterns`	Proven solutions	Persistent while useful, promoted from history
`memory_bank`	Identity, preferences, goals	Permanent
`books`	Uploaded reference docs	Permanent

Commands

roampal init                # Auto-detect and configure installed tools
roampal init --claude-code  # Configure Claude Code explicitly
roampal init --opencode     # Configure OpenCode explicitly
roampal init --no-input     # Non-interactive setup (CI/scripts)
roampal start               # Start the HTTP server manually
roampal stop                # Stop the HTTP server
roampal status              # Check if server is running
roampal status --json       # Machine-readable status (for scripting)
roampal stats               # View memory statistics
roampal stats --json        # Machine-readable statistics (for scripting)
roampal doctor              # Diagnose installation issues
roampal summarize           # Summarize long memories (retroactive cleanup)
roampal score               # Score the last exchange (manual/testing)
roampal context             # Output recent exchange context
roampal ingest <file>       # Add documents to books collection
roampal books               # List all ingested books
roampal remove <title>      # Remove a book by title
roampal sidecar status      # Check scoring model configuration (OpenCode)
roampal sidecar setup       # Configure scoring model (OpenCode)
roampal sidecar test        # Test scoring model response format (OpenCode)
roampal sidecar disable     # Remove scoring model configuration (OpenCode)

MCP Tools

Your AI gets these memory tools:

Tool	Description	Platforms
`search_memory`	Deep search across all collections	Both
`add_to_memory_bank`	Store permanent facts (identity, preferences, goals)	Both
`update_memory`	Correct or update existing memories	Both
`delete_memory`	Remove outdated info	Both
`score_memories`	Score previous exchange outcomes	Claude Code
`record_response`	Store key takeaways from significant exchanges	Both

How scoring works: Claude Code's hooks prompt the main LLM to call score_memories every turn. OpenCode uses an independent sidecar that scores silently in the background — the model never sees a scoring prompt and score_memories is not registered as a tool. If the sidecar is unavailable, a warning prompts the user to run roampal sidecar setup. Choose your scoring model during roampal init or via roampal sidecar setup.

How Roampal Compares

Feature	Roampal Core	Claude Code built-in (CLAUDE.md / auto memory)	OpenCode built-in
Learns from outcomes	Yes — bad advice demoted, good advice promoted	No	No
Semantic retrieval	Yes — TagCascade + cross-encoder reranking	No — files loaded in full, no search	No memory system
Context injection	Automatic — relevant memories per query	Full CLAUDE.md every session, auto memory on demand	None
Atomic fact extraction	Yes — summaries + facts, two-lane retrieval	No — saves what Claude decides is useful	No
Works across projects	Yes — shared memory across all projects	Per-project only (per git repo)	No memory
Scales with history	Yes — 5 collections, promotion/demotion/decay	CLAUDE.md unbounded, auto memory first 200 lines	No memory
Fully local / private	Yes — ChromaDB on your machine	Yes	Yes

Architecture

┌─────────────────────────────────────────────────────────┐
│  pip install roampal && roampal init                    │
│    Claude Code: hooks + MCP → ~/.claude/                │
│    OpenCode:    plugin + MCP → ~/.config/opencode/      │
└─────────────────────────────────────────────────────────┘
                         │
                         ▼
┌─────────────────────────────────────────────────────────┐
│  HTTP Hook Server (port 27182)                          │
│    Auto-started on first use, self-heals on failure     │
│    Manual control: roampal start / roampal stop         │
└─────────────────────────────────────────────────────────┘
                         │
                         ▼
┌─────────────────────────────────────────────────────────┐
│  User types message                                     │
│    → Hook/plugin calls HTTP server for context          │
│    → AI sees relevant memories, responds                │
│    → Exchange stored, scored (hooks or sidecar)         │
└─────────────────────────────────────────────────────────┘
                         │
                         ▼
┌─────────────────────────────────────────────────────────┐
│  Single-Writer Backend                                  │
│    FastAPI → UnifiedMemorySystem → ChromaDB             │
│    All clients share one server, isolated by session    │
└─────────────────────────────────────────────────────────┘

See dev/docs/ for full technical details.

Requirements

Python 3.10+
One of: Claude Code or OpenCode
Platforms: Windows, macOS, Linux (primarily developed and tested on Windows)
RAM: ~800MB available (cross-encoder reranker + embeddings + ChromaDB)
Disk: ~500MB for models (multilingual embedding + reranker, downloaded automatically on first use)
CPU: Any modern x86-64 processor with AVX2 (Intel Haswell 2013+ / AMD Excavator 2015+)
GPU: Not required — all inference runs on CPU via ONNX Runtime

Troubleshooting

Hooks not working? (Claude Code)

Restart Claude Code (hooks load on startup)
Check HTTP server: curl http://127.0.0.1:27182/api/health

MCP not connecting? (Claude Code)

Verify ~/.claude.json has the roampal-core MCP entry with correct Python path
Check Claude Code output panel for MCP errors

Context not appearing? (OpenCode)

Make sure you ran roampal init --opencode
Check that the server auto-started: curl http://127.0.0.1:27182/api/health
If not, start it manually: roampal start

Server crashes and recovers?

This is expected. Roampal has self-healing -- if the HTTP server stops responding, it is automatically restarted and retried.

Still stuck? Ask your AI for help — it can read logs and debug Roampal issues directly.

Support

Roampal Core is completely free and open source.

Support development: roampal.gumroad.com
Feature ideas & feedback: Discord
Bug reports: GitHub Issues
Need help with AI memory? Reach out: roampal@protonmail.com | LinkedIn

License

Apache 2.0

Release History

Version	Changes	Urgency	Date
v0.5.7	Hardening release. No new user-visible features. Closes an unbounded-growth defect in the MCP hook's per-conversation lifecycle file (`_completion_state.json`) that surfaced during issue #8 repro. ## What's fixed `_completion_state.json` startup GC — the file accumulated one entry per `conversation_id` ever seen, with no cleanup. On long-running installs this drove: - I/O amplification (full-file rewrite on every state mutation) - Stuck `scored_this_turn=True` flags from long-dead session	High	5/12/2026
v0.5.6	Hardening release closing coverage gaps from the v0.5.5.x verification audit. No new user-visible features. ## Memory phantoms (#8 follow-up) — items 5-11, 32 - `_sweep_phantoms()` runs after `cleanup_archived()`, not just at startup - chromadb_adapter phantom filter strengthened (AND → OR mid-state catch) - Status backfill on startup for legacy entries - Auto-trigger `cleanup_archived()` under capacity pressure - Dedup observability log on every dedup-skip - Lane enforcement: `delete()` → `del	High	5/1/2026
v0.5.5	## Roampal Core v0.5.5 — 2026-04-27 Bugfix release. Fixes issue #8 where GUI deletion of memories blocks new memory generation. ### What's Fixed - Soft-delete for memory_bank — ChromaDB hard delete doesn't actually remove vectors from HNSW, causing phantom dedup matches that block new memories after GUI deletion. Replaced with status=archived metadata update plus status filter on all query/dedup paths. - Scoring mutex to async queue — Eliminates dropped scoring requests that caused	High	4/27/2026
v0.5.3.1	## Hotfix for v0.5.3 — facts and summary extraction broken on thinking models Three bugs in the OpenCode plugin's JSON extraction logic. All three only manifested with reasoning/thinking models on local sidecar setups. ### What was broken Bug 1 — Facts extraction silently failing on every exchange (latent since v0.4.8). The facts extraction path used a greedy regex that swallowed trailing markdown code fences. Models that wrap JSON in ` ```json ... ``` ` blocks broke `JSON.parse()` with `	High	4/23/2026
v0.5.2	# Roampal Core v0.5.2 Release Date: 2026-04-21 Type: Patch — chat-path performance, field-drift lifecycle fix, v0.5.1 profile-resolution bugfixes Coordination: Core ships first; Desktop v0.3.2 follows with the same latency trio and age-gate fix applied to the Tauri backend ## Summary Three bundled improvements: 1. Performance (Sections 1–3) — cut the per-message latency added by the memory pipeline on every MCP tool call. Beta feedback on Desktop v0.3.1 (2026-04-17) fla	High	4/21/2026
v0.5.1	# Roampal Core v0.5.1 Release Notes Release Date: TBD Type: Minor — named memory profiles + sidecar consistency patch ## Summary Two changes bundled: 1. Named memory profiles (feature) — support multiple isolated memory stores per user, switchable per command, per shell, per project (MCP config), or globally. 2. Sidecar scoring cap (patch) — cap exchange fields at 8K characters in the scoring prompt, matching the fact-extraction call. --- ## 1. Named Memory Profiles ### Mo	High	4/17/2026
v0.5.0	## Summary v0.5.0 addresses five issues in the OpenCode plugin: 1. Subagent filtering (Issue #4): Subagent exchanges no longer pollute the memory store 2. OpenAI sidecar fix: Removed `think: false` field that broke all non-Ollama sidecar targets 3. Sidecar input cap: Bumped fact extraction input from 800 chars to 16K (8K + 8K) 4. Summary output cap: Bumped scoring prompt summary limit from 300 chars to 2000 chars 5. Scoring status visibility: Replaced boolean `scoringBroken	High	4/16/2026
v0.4.9.2	## Summary Rewrite all 6 MCP tool descriptions for Glama TDQS (Tool Definition Quality Score) compliance. ### Changes Every tool description now includes: - BEHAVIOR section — explicit state changes, what gets created/modified/deleted - ERROR HANDLING section — what happens on invalid input, no matches, empty results - WHEN NOT TO USE section — clear disambiguation between tools - Parameter examples — concrete examples for every parameter ### Previous Glama Scores \| Tool \| G	High	4/14/2026
v0.4.9.1	## Hotfix Critical fix for server deadlock on startup and leftover regex tag fallback shipped in v0.4.9. ### Fixes - Server deadlock on startup — A v0.4.8 migration block ran on every boot inside the async `lifespan()` function, pulling up to 30,000 facts from ChromaDB via synchronous calls. With large datasets this blocked the event loop indefinitely — the server bound the port but never served requests. All hooks timed out silently. Removed entirely; use `roampal retag` instead. - **Reg	High	4/14/2026
v0.4.9	## Summary Fixes critical sidecar robustness issues and achieves full benchmark alignment for tag extraction. ### Key Changes - Wire TagService to sidecar LLM — v0.4.8 had `extract_tags()` in code but never called it. Tags were silently falling back to regex. Now properly wired for OpenCode sidecar scoring. - Remove regex tag extraction — LLM-only tags matching benchmark behavior. If LLM fails, returns `[]` (no fallback). Use `roampal retag` to clean up existing memories. - **Robust b	High	4/14/2026
v0.4.8	# v0.4.8 Release Notes Platforms: OpenCode (sidecar plugin), server (shared) Theme: Sidecar reliability + benchmark alignment Status: Complete Claude Code impact: None. Changes 1, 2, 4 are OpenCode-only (roampal.ts). Change 3 (fact tags) is in the shared server but is additive — puts tags back on facts. --- ## Overview Mirror the benchmark's proven sidecar architecture. Four changes: 1. Remove autoSummarize — eliminates Ollama contention that caused 20% scoring failures	Medium	4/14/2026
v0.4.7	## Changes - Compaction recovery: Generation counter ensures `session.idle` doesn't clear recovery flags prematurely - Cold start: New sessions pre-fetch 4 most recent exchange summaries for continuity - Sidecar resilience: autoSummarize gated (can't block scoring), circuit breaker 30min→2min, scoringBroken auto-reset - Recency metadata: `_search_all()` now adds recency strings; fallback to `created_at` when `timestamp` missing - Unified metadata format: Recent exchanges use	High	4/13/2026
v0.4.6.1	## v0.4.6.1 — MCP Tool Documentation Improves MCP tool descriptions for Glama quality evaluation. ### Changes - update_memory: Added memory lifecycle, matching behavior, return value docs - score_memories: Added Wilson scoring explanation, parameter examples, side effects - delete_memory: Added irreversibility warning, scope, matching behavior - search_memory: Added return value structure, sibling tool differentiation - add_to_memory_bank: Added record_response differentiat	Medium	4/13/2026
v0.4.6	## v0.4.6 — Context Quality Improves context injection quality by removing low-signal extractions and reframing auto-extracted facts. ### Changes 1. Reworded facts header in context injection - Facts section now indicates auto-extracted facts are directional, not authoritative - LLM treats them as leads to verify rather than ground truth 2. Removed regex tag extraction from book ingestion - `extract_tags_regex()` removed from `store_book()` — low signal on book chunks - Tags will com	Medium	4/13/2026
v0.4.5	## TagCascade Retrieval — Benchmark-Validated Architecture Replaces knowledge graph with TagCascade retrieval validated on LoCoMo benchmark. ### Highlights - 85.8% on LoCoMo (non-adversarial, end-to-end answer accuracy) - +23 points over raw ingestion (76.6% vs 53.0%, p<0.0001) - TagCascade retrieval: tags-first cascade + CE reranking, raw CE score (no Wilson blend) - Two-lane retrieval: 4 summaries + 4 facts = 8 memories per context injection - Atomic fact extraction: LLM	High	4/9/2026
v0.4.4	## v0.4.4 — Async Parallelization Performance release — all independent operations now run concurrently. ### Changes - Parallel collection searches via `asyncio.gather` - Parallel context injection (3 searches run concurrently) - Parallel collection adapter initialization and startup cleanup - Parallel per-memory scoring and background KG updates - Full memory metadata in KNOWN CONTEXT (`wilson:N%`, `used:Nx`, `last:outcome`) - Negative example added to `record_response` tool description - ARC	Low	3/16/2026
v0.4.3	## What's Changed Replaces PyTorch + sentence-transformers with direct ONNX Runtime inference. Install drops from ~2.5GB to ~200MB. Same model, same 768-dimension embeddings, same ChromaDB collections — zero user-facing change. ### Highlights - Zero new dependencies — `onnxruntime`, `tokenizers`, `huggingface-hub`, `numpy` are all already installed via `chromadb` - Identical embeddings — cosine similarity vs PyTorch: 1.0000000, multilingual verified across 8 languages - **Optimized mod	Low	3/13/2026
v0.4.2.2	## What's Changed Fixes a regression from v0.4.2.1 where `roampal init` would silently delete OpenCode sidecar scoring configuration. ### Fixes - OpenCode MCP environment merged instead of replaced — sidecar variables (`ROAMPAL_SIDECAR_URL`, `ROAMPAL_SIDECAR_MODEL`, `ROAMPAL_SIDECAR_KEY`) are now preserved during `roampal init` - Idempotency check ignores sidecar vars — config is no longer rewritten on every init when sidecar is configured ### Scope Claude Code and Cursor are not affe	Low	3/11/2026
v0.4.2.1	## What's Changed `roampal init` no longer wipes non-roampal hooks or blocks config updates behind `--force`. ### Fixes - Claude Code hooks merged instead of replaced — non-roampal hooks (linters, formatters, custom scripts) are now preserved - Cursor hooks merged instead of replaced — same fix for Cursor's `hooks.json` - MCP configs auto-update — no longer requires `--force` when config differs between versions - OpenCode plugin auto-updates — plugin is always updated when sou	Low	3/11/2026
v0.4.2	## Summary - Fix intermittent "UserPromptSubmit hook error" — embedding cache eliminates redundant encode calls (~12s → ~4s), hook timeout bumped 5s → 10s - OpenCode plugin scoring fixes — 1.5s debounce on `session.idle` prevents duplicate exchange storage from subagent completions, compacting hook and score detection registered correctly - Version check fix — `importlib.metadata` replaces `from roampal import __version__` which was shadowed by co-installed roampal-cli - **ONNX Runt	Low	3/11/2026
v0.4.1	## Highlights - Event loop unblocking: `embed_text()` wrapped in `asyncio.to_thread()` — fixes the root cause of "not responding" errors on Linux - Performance caps: `_search_all` capped at 1,000/collection, BM25 index rebuild at 2,000 docs — prevents OOM on memory-constrained systems - Sidecar-only scoring on OpenCode: `score_memories` tool hidden from MCP, no main LLM fallback — eliminates double-scoring - Timezone fix: `_humanize_age()` now shows correct memory ages on non-UT	Low	3/5/2026
v0.4.0	## Cross-platform compatibility audit and backend data integrity fixes ### Critical - Fix ChromaDB `hnsw:space` metadata mismatch between `query_vectors()` and `initialize()` ### High - Fix `search_books()` calling nonexistent adapter method (was broken since launch) - Write `original_id` metadata during all promotion paths so outcome scoring finds promoted docs - Add `_promotion_lock` to `handle_promotion` to prevent race with batch promotion - Use `split('_', 1)[1]` for promotion ID derivati	Low	3/4/2026
v0.3.9	### Fix scoring truncation, add memory storage safety cap Sidecar Scoring Truncation Fix: - Removed `.slice(0, 200)` in OpenCode sidecar scoring prompt (`roampal.ts`) — memories were evaluated with content beyond 200 characters removed since v0.3.5 Fallback Scoring Truncation Fix: - Removed 120-char truncation in `build_scoring_prompt_simple()` (`session_manager.py`) — fallback scoring path (when sidecar unavailable) was also discarding content **Memory Storage Safety Cap (2000 chars)	Low	3/2/2026
v0.3.8	### Docker support, memory_bank scoring transparency, version fix, thread-safety fix Docker Support: - Dockerfile + .dockerignore for Glama awesome-mcp-servers listing - `python:3.12-slim` base, installs from PyPI, health check on `/api/health` - `start-period=120s` for embedding model download on first boot Memory Bank Scoring Transparency: - `search_memory` results for memory_bank facts with 1+ uses now show Wilson score (`w:`), uses count, and outcome history - Fresh facts (0 uses)	Low	3/2/2026
v0.1.10	v0.1.10: Update notifications in Claude Code	Low	2/25/2026
v0.2.0	v0.2.0: Action KG sync, batch embedding, sentence chunking, date filters, auto-restart	Low	2/25/2026
v0.2.1	Fix MCP method not found error for fresh installs	Low	2/25/2026
v0.2.2	v0.2.2: Cursor support, always_inject, ghost registry, doctor command, schema migration	Low	2/25/2026
v0.2.3	Performance fix - deferred KG learning, faster scoring	Low	2/25/2026
v0.2.4	v0.2.4: Fix related parameter in MCP fallback, full doc_id display	Low	2/25/2026
v0.2.5	v0.2.5: Fix MCP config path (~/.claude.json), add idempotency, --force flag, migration, status command	Low	2/25/2026
v0.2.6	v0.2.6: Identity prompts, profile-only cold start, tool description improvements	Low	2/25/2026
v0.2.7	v0.2.7: Identity injection, lean cold start, scoring guidance	Low	2/25/2026
v0.2.8	v0.2.8: Wilson fix, FastAPI lifecycle, per-memory scoring	Low	2/25/2026
v0.2.8.1	Fix unknown outcome treated as partial	Low	2/25/2026
v0.2.9	v0.2.9: Natural Selection for Memory	Low	2/25/2026
v0.2.9.1	Fix memory_bank deletion bug	Low	2/25/2026
v0.2.9.2	Restore exchange scoring	Low	2/25/2026
v0.3.0	v0.3.0: Resilience - fix silent hook server crashes	Low	2/25/2026
v0.3.0.1	Fix version string in __init__.py	Low	2/25/2026
v0.3.1	Reserved working memory slot for context injection	Low	2/25/2026
v0.3.2	v0.3.2: Multi-client support, sidecar scoring, self-healing hooks, README overhaul	Low	2/25/2026
v0.3.3	v0.3.3: Fix OpenCode plugin packaging, add memory awareness preamble	Low	2/25/2026
v0.3.3.1	v0.3.3.1: Fix tool detection on fresh installs	Low	2/25/2026
v0.3.3.2	v0.3.3.2: Add relative timestamps to KNOWN CONTEXT injection	Low	2/25/2026
v0.3.4	Fix garbled UI, double-scoring, and scoring prompt for OpenCode	Low	2/25/2026
v0.3.4.1	Fix console.log leaking into OpenCode UI on Linux	Low	2/25/2026
v0.3.5	Token optimization, standardized metadata, tool rewrites, memory hygiene, security hardening	Low	2/25/2026
v0.3.5.1	Fix 4-slot context injection display	Low	2/25/2026
v0.3.5.2	Fix double-scoring in OpenCode plugin	Low	2/25/2026
v0.3.6	### Retrieval fairness, exchange summarization, sidecar scoring - Retrieval fairness: reserved working memory slot, diversity caps, unknown penalty - Exchange summarization via sidecar (platform-split: Claude Code main LLM, OpenCode sidecar) - Compaction recovery: recent exchange buffer via SessionStart hook - roampal summarize CLI for retroactive memory cleanup - roampal context --recent-exchanges for platform hooks - score_response renamed to score_memories with summary/outcome fields - Remov	Low	2/25/2026
v0.3.6.1	Add init --force to all upgrade notices	Low	2/25/2026
v0.3.7	### Sidecar-only scoring, cold start recovery, CLI sidecar setup - Sidecar is sole scorer for OpenCode (main LLM no longer prompted to score) - `roampal sidecar setup` CLI command for one-command scorer configuration - Init-time sidecar model selection with auto-detection - Wilson-only ranking for memory_bank facts with 3+ uses - CLI polish: NO_COLOR, --no-input, --json, exit codes, cached update checks - Security: bare excepts fixed, cache TTL eviction, collection prefix safety - Package weigh	Low	2/25/2026
v0.3.7.1	### Fix: Stale version detection in update check Bug: After upgrading via `pip install --upgrade roampal`, the server kept showing "update available" because the long-running process cached the old `__version__` from Python's module cache. Fix: - `_get_installed_version()` reads version from disk instead of stale `sys.modules` cache - Update check cache now expires after 30 minutes (was cached forever)	Low	2/25/2026

Dependencies & License Audit

Loading dependencies...

Similar Packages

sugarPersistent memory for AI coding agents. Cross-session context, global knowledge, and autonomous task execution.v3.9.1

AIDomesticCoreAIJ🛠️ Build a robust AI Kernel for stable, auditable, and sovereign AI systems, ensuring secure execution and compliance across various domains.main@2026-06-07

mcp-agent-framework🤖 Orchestrate AI agents at scale using the MCP framework, enabling seamless context sharing, communication, and integration for enhanced collaboration.master@2026-06-07

git-notes-memory🧠 Store and search your notes effectively with Git-native memory storage, enhancing productivity for Claude Code users.main@2026-06-06

a-mem-mcp-server🧠 Enhance LLM agents with an agentic memory system, featuring automatic note construction, dynamic memory updates, and intelligent semantic retrieval.main@2026-06-06

More in MCP Servers

agentroveYour own Claude Code UI, sandbox, in-browser VS Code, terminal, multi-provider support (Anthropic, OpenAI, GitHub Copilot, OpenRouter), custom skills, and MCP servers.

ProxmoxMCP-PlusEnhanced Proxmox MCP server with advanced virtualization management and full OpenAPI integration.

node9-proxyThe Execution Security Layer for the Agentic Era. Providing deterministic "Sudo" governance and audit logs for autonomous AI agents.

mcp-compressorAn MCP server wrapper for reducing tokens consumed by MCP tools.