# prism-mcp

> The Mind Palace for AI Agents — Autonomous Cognitive OS with affect-tagged memory (valence engine), token-economic RL (surprisal gate + UBI), Hebbian learning, ACT-R spreading activation, Synapse Engi

- **URL**: https://www.freshcrate.ai/projects/prism-mcp
- **Author**: dcostenco
- **Category**: MCP Servers
- **Latest version**: `v15.2.1` (2026-05-10)
- **License**: MIT
- **Source**: https://github.com/dcostenco/prism-mcp
- **Homepage**: https://www.npmjs.com/package/prism-mcp-server
- **Language**: TypeScript
- **GitHub**: 128 stars, 23 forks
- **Registry**: github
- **Tags**: `agent-memory`, `ai-agent`, `anti-sycophancy`, `claude-desktop`, `cognitive-architecture`, `google-gemini`, `hebbian-learning`, `llm-tools`, `typescript`

## Description

The Mind Palace for AI Agents — Autonomous Cognitive OS with affect-tagged memory (valence engine), token-economic RL (surprisal gate + UBI), Hebbian learning, ACT-R spreading activation, Synapse Engine (GraphRAG), adversarial evaluation (anti-sycophancy), fail-closed Dark Factory pipelines. Zero-config local mode.

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v15.2.1` | 2026-05-10 | High | ## What's new  ### Pre-push verification gate (Rule 19 expanded) Rule 19 is now a **full verification gate** — not just `tsc --noEmit`. Both steps must pass before every `git push`, regardless of repo type: - **Type check** — `tsc --noEmit` / `next build` / `mypy` / `pyright` - **Test suite** — `npx vitest run` / `python3 -m pytest` / `npm test` / `jest`  Each catches a different failure class. Type check alone missed 2 incidents in May 2026.  ### Mechanical guard + 20 test cases `verification_g |
| `v13.0.0` | 2026-05-08 | High | See [CHANGELOG.md](https://github.com/dcostenco/prism-coder/blob/main/CHANGELOG.md) for full notes — Release object created retroactively to fill the gap on the repo page. |
| `v12.5.6` | 2026-04-29 | High | ## Prism MCP v12.5.6  - Gemini 3.1 Pro Experimental as default for paid tiers - Claude Sonnet 4.6 / Haiku 4.5 added to allowed models - i18n: all 11 translations synced as full README copies - Pre-commit hook for automatic i18n regeneration |
| `v9.3.0` | 2026-04-11 | High | ## TurboQuant ResidualNorm Tiebreaker  Enterprise-grade configurable ranking optimization for Tier-2 TurboQuant search.  ### What's New  - **`PRISM_TURBOQUANT_TIEBREAKER_EPSILON`** - New env var (default: 0, disabled). When two compressed cosine scores are within epsilon of each other, prefers the candidate with lower `residualNorm` - its compressed representation captured more signal energy, making its score more trustworthy. - - **Recommended:** 0.005 for enterprise deployments with lar |
| `v9.3.0` | 2026-04-11 | Medium | ## TurboQuant ResidualNorm Tiebreaker  Enterprise-grade configurable ranking optimization for Tier-2 TurboQuant search.  ### What's New  - **`PRISM_TURBOQUANT_TIEBREAKER_EPSILON`** - New env var (default: 0, disabled). When two compressed cosine scores are within epsilon of each other, prefers the candidate with lower `residualNorm` - its compressed representation captured more signal energy, making its score more trustworthy. - - **Recommended:** 0.005 for enterprise deployments with lar |
| `v9.3.0` | 2026-04-11 | Medium | ## TurboQuant ResidualNorm Tiebreaker  Enterprise-grade configurable ranking optimization for Tier-2 TurboQuant search.  ### What's New  - **`PRISM_TURBOQUANT_TIEBREAKER_EPSILON`** - New env var (default: 0, disabled). When two compressed cosine scores are within epsilon of each other, prefers the candidate with lower `residualNorm` - its compressed representation captured more signal energy, making its score more trustworthy. - - **Recommended:** 0.005 for enterprise deployments with lar |
| `v9.0.2` | 2026-04-07 | High | Fixes verification tests bleeding into GitHub Actions environment. |
| `v9.0.1` | 2026-04-07 | Medium | ### 🤖 Autonomous Cognitive OS (v9.0)  > *Memory isn't just about storing data; it's about economics and emotion. v9.0 transforms Prism from a passive memory database into a living Cognitive Operating System that forces agents to learn compression and develop intuition.*  Most AI agents have an infinite memory budget. They dump massive, repetitive logs into vector databases until the context window chokes. Prism v9.0 fixes this by introducing **Token-Economic Reinforcement Learning** and **Affec |
| `v7.4.0` | 2026-04-03 | Medium | ## ⚔️ v7.4.0 — Adversarial Evaluation (Anti-Sycophancy)  Dark Factory gains a full adversarial evaluation loop — the signature feature of this release. The pipeline now fights itself to produce high-quality output, solving the fundamental problem of LLMs grading their own homework.  ### What's New  **Split-Brain Anti-Sycophancy** - `PLAN_CONTRACT` step locks a machine-parseable scoring rubric to disk before any code is written - `EVALUATE` step runs in a fully isolated context — Evaluator sees o |
| `v5.1.0` | 2026-03-27 | Medium | ## What's New  ### Added - **Deep Storage Mode**: New `deep_storage_purge` tool to reclaim ~90% of vector storage by dropping float32 vectors for entries with TurboQuant compressed blobs. - **Knowledge Graph Editor**: The Mind Palace neural graph is now fully interactive — click nodes to rename or delete keywords, filter by project/date/importance, and surgically groom your agent's semantic memory.  ### Fixed - **Auto-Load Reliability**: Hardened auto-load prompt instructions and added hook scri |

## Citation

- HTML: https://www.freshcrate.ai/projects/prism-mcp
- Markdown: https://www.freshcrate.ai/projects/prism-mcp.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/prism-mcp/deps

_Generated by freshcrate.ai. Indexes github releases for AI-agent ecosystem packages._