MCP 服务器

cognos-session-memory

Enables persistent session memory for LLMs by trust-scoring and injecting verified context between conversations.

README

CognOS Session Memory

mcp-name: io.github.base76-research-lab/cognos-session-memory

Verified context injection via epistemic trust scoring for LLMs.

Solves session fragmentation by maintaining verified, high-confidence session context between conversations.

Problem

Large language models suffer from session fragmentation: each new conversation starts without verified context of previous work. This forces repeated explanations, loses decision history, and breaks long-running workflows.

Existing solutions (persistent memory systems, vector retrieval) either:

Lack trust scores before injection → hallucinations propagate
Don't audit which context was injected → compliance gaps
Treat all past information equally → noise overwhelms signal

Solution

A plan-mode gateway that:

Extracts structured context from 3-5 recent traces
Scores context quality via CognOS epistemic formula: C = p · (1 − Ue − Ua)
Injects as system prompt only if C > threshold
Flags for manual review if C < threshold
Audits every context injection with trace IDs → EU AI Act compliance

Architecture

recent_traces (n=5)
    ↓
extract_context() → ContextField + coverage
    ↓
compute_trust_score(p, ue, ua) → C, R, decision
    ↓
if C > threshold:
    system_prompt ← inject
else:
    flagged_reason ← manual review

Core Formula

C = p · (1 − Ue − Ua)
R = 1 − C

where:
  p   = prediction confidence (coverage of required fields)
  Ue  = epistemic uncertainty (divergence between traces)
  Ua  = aleatoric uncertainty (mean risk in traces)

Action Gate

R < 0.25       → PASS      (inject without review)
0.25 ≤ R < 0.60 → REFINE   (inject with caution)
R ≥ 0.60       → ESCALATE  (flag for manual review)

API

POST /v1/plan

Extract and score context.

Request:

{
  "n": 5,
  "trust_threshold": 0.75,
  "mode": "auto"
}

Response (if injected):

{
  "status": "injected",
  "trust_score": 0.82,
  "confidence": 0.82,
  "risk": 0.18,
  "decision": "PASS",
  "context": {
    "active_project": "CognOS mHC research",
    "last_decision": "Verify P1 hypothesis",
    "open_questions": ["How does routing entropy scale?"],
    "current_output": "exp_008 complete",
    "recent_models": ["gpt-4", "claude-3", "mistral"]
  },
  "system_prompt": "## CognOS Context...",
  "trace_ids": ["uuid-1", "uuid-2", ...]
}

Response (if flagged):

{
  "status": "flagged",
  "trust_score": 0.45,
  "decision": "REFINE",
  "flagged_reason": "Trust score 0.45 below threshold 0.75. Manual review recommended.",
  "trace_ids": [...]
}

Modes

auto (default) — inject if trust_score ≥ threshold, else flag
force — always inject (for testing)
dry_run — compute score but never inject

Claude Code Integration

As a /compact replacement

# In any Claude Code session:
/save

Claude writes a structured summary, trust-scores it, and persists it to SQLite. Next session: automatically injected as SESSION_CONTEXT before your first prompt.

See docs/COMPACT_ALTERNATIVE.md for a full comparison.

As an MCP server

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "cognos-session-memory": {
      "command": "python3",
      "args": ["/path/to/cognos-session-memory/mcp_server.py"]
    }
  }
}

Tools exposed:

Tool	Description
`save_session(summary, project?)`	Trust-score and persist a session summary
`load_session(threshold?)`	Retrieve last verified context (default threshold: 0.45)

Quick Start

Installation

git clone https://github.com/base76-research-lab/cognos-session-memory
cd cognos-session-memory
pip install -e .

Run Gateway

python3 -m uvicorn --app-dir src main:app --port 8788

Test /v1/plan (dry_run)

curl -X POST http://127.0.0.1:8788/v1/plan \
  -H 'Content-Type: application/json' \
  -d '{"n": 5, "mode": "dry_run"}'

Test /v1/plan (auto)

curl -X POST http://127.0.0.1:8788/v1/plan \
  -H 'Content-Type: application/json' \
  -d '{"n": 5, "trust_threshold": 0.75, "mode": "auto"}'

Modules

trust.py — CognOS confidence formula, action gate, signal extractors
trace_store.py — SQLite persistence (write/read/purge)
plan.py — Context extraction, trust scoring, system prompt building
main.py — FastAPI gateway + middleware
mcp_server.py — MCP stdio server (save_session, load_session)

Testing

pytest tests/ -v --cov=src

Documentation

COMPACT_ALTERNATIVE.md — Why this beats /compact
PAPER.md — Research paper

Research Paper

See docs/PAPER.md — "Verified Context Injection: Epistemically Scored Session Memory for Large Language Models"

Status: Independent research — Base76 Research Lab, 2026 Authors: Björn André Wikström (Base76)

Citation

@software{wikstrom2026cognos,
  author = {Wikström, Björn André},
  title = {{CognOS Session Memory}: Verified Context Injection via Epistemic Trust Scoring},
  year = {2026},
  url = {https://github.com/base76-research-lab/cognos-session-memory}
}