claude-praetorian-mcp

Provides an MCP server (Node/TypeScript, used via npx/claude mcp) for persisting and retrieving Claude Code context “compactions” as structured, incremental TOON snapshots stored locally under each project’s .claude/praetorian directory. Includes MCP tools/commands for compacting context (with dedup/merge via Jaccard title similarity) and restoring/searching relevant saved snapshots into the active context.

Evaluated Mar 30, 2026 (22d ago)
Homepage ↗ Repo ↗ Ai Ml mcp claude-code context-management token-optimization memory compaction offline filesystem
⚙ Agent Friendliness
61
/ 100
Can an agent use this?
🔒 Security
32
/ 100
Is it safe for agents?
⚡ Reliability
32
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
75
Documentation
70
Error Messages
0
Auth Simplicity
100
Rate Limits
10

🔒 Security

TLS Enforcement
0
Auth Strength
20
Scope Granularity
10
Dep. Hygiene
60
Secret Handling
80

Appears offline and stores flat files locally per project; no network calls or centralized service described. However, there is no explicit discussion of data retention/deletion, file permission hardening, input sanitization, or error/logging behavior. If content includes sensitive research/task outputs, it will be persisted to disk under .claude/praetorian without stated encryption at rest.

⚡ Reliability

Uptime/SLA
0
Version Stability
50
Breaking Changes
40
Error Recovery
40
AF Security Reliability

Best When

You run Claude Code locally/within a controlled environment and want frequent, automated context compaction/reuse with minimal token spend while keeping everything project-scoped and offline.

Avoid When

You need multi-tenant isolation, strict compliance guarantees, or you cannot persist data to <project>/.claude/praetorian on disk.

Use Cases

  • Reduce token usage in Claude Code by compacting web research outputs and task/subagent results into small structured snapshots
  • Fast resume of prior work by restoring relevant compactions via keyword query search
  • Maintain/refresh project-level “memory” of decisions, findings, and references without external databases
  • Proactive context compaction during long-running coding/research sessions (and via Claude Code hooks/skills/plugins)

Not For

  • Sharing or hosting compactions as a centralized cloud service (it appears local/offline and project-scoped)
  • Scenarios requiring network-based search across many projects/accounts
  • Use where you cannot write to the filesystem or where local persistence is disallowed
  • Security-sensitive environments that require strict data retention controls without clear deletion/retention options

Interface

REST API
No
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
No
Webhooks
No

Authentication

OAuth: No Scopes: No

No auth described; functionality appears local/offline and operates on project files via the MCP runtime (Claude Code configuration).

Pricing

Free tier: No
Requires CC: No

No pricing described; likely free as an npm package (MIT) with your own compute/token costs only.

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • Relies on project-scoped file storage under .claude/praetorian; ensure the working directory/project path is correct for the MCP server process.
  • If using npx, npm resolution may install an unexpected version/package unless pinned/forced.
  • No explicit rate-limit guidance (likely none since it’s local/file-based, but not documented).
  • Context restoration/search is keyword/index-based (inverted index) per README; for semantic matches you may need to structure titles/insights accordingly.

Alternatives

Full Evaluation Report

Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for claude-praetorian-mcp.

AI-powered analysis · PDF + markdown · Delivered within 30 minutes

$99

Package Brief

Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.

Delivered within 10 minutes

$3

Score Monitoring

Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.

Continuous monitoring

$3/mo

Scores are editorial opinions as of 2026-03-30.

8642
Packages Evaluated
17761
Need Evaluation
586
Need Re-evaluation
Community Powered