sigil-mcp-server

sigil-mcp-server is a Python Model Context Protocol (MCP) server that indexes local repositories and exposes code navigation/search tools (symbol lookup, trigram-based substring search, and optional semantic search via vector embeddings) along with health/admin endpoints.

Evaluated Apr 04, 2026 (27d ago)
Homepage ↗ Repo ↗ Search mcp code-search repository-indexing symbol-search ctags trigram-index rocksdb lancedb embeddings oauth2 python
⚙ Agent Friendliness
55
/ 100
Can an agent use this?
🔒 Security
67
/ 100
Is it safe for agents?
⚡ Reliability
28
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
86
Documentation
70
Error Messages
0
Auth Simplicity
65
Rate Limits
10

🔒 Security

TLS Enforcement
85
Auth Strength
80
Scope Granularity
40
Dep. Hygiene
55
Secret Handling
70

Security documentation mentions path traversal protection, OAuth 2.0 with PKCE, localhost bypass, API key fallback, and optional IP whitelisting. It also explicitly states DNS rebinding protection and certain validations are disabled for ChatGPT compatibility (host/header and content-type validation). OAuth token expiration is enforced (~1 hour) and OAuth credentials are stored with 0600 permissions.

⚡ Reliability

Uptime/SLA
0
Version Stability
35
Breaking Changes
20
Error Recovery
55
AF Security Reliability

Best When

You want an AI assistant connected through MCP to explore codebases on your own machine or a controlled network, with optional OAuth-based remote access and indexing that can be maintained over time.

Avoid When

You need a minimal/fully documented REST/OpenAPI service for third-party programmatic consumption (this is primarily an MCP server) or you cannot accept the stated tradeoffs in ChatGPT connector compatibility (DNS rebinding/header/content-type validation behavior).

Use Cases

  • Index local source code for IDE-like symbol navigation and code search in an AI assistant
  • Fast substring search across multiple repositories using trigram indexing
  • Go-to-definition and symbol listing for functions/classes/methods/variables
  • Semantic/natural-language code search using embeddings (e.g., local llama.cpp or other embedding providers)
  • Operational management of indexing (rebuilds, stats, logs) via an admin API
  • Provide an MCP tool surface to other systems that can connect via MCP + OAuth

Not For

  • Publicly exposed, unauthenticated code search endpoints
  • Environments that forbid disabling DNS rebinding and strict header/content-type validation for ChatGPT compatibility
  • Use cases requiring strict multi-tenant data isolation without additional deployment controls

Interface

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

Authentication

Methods: OAuth 2.0 (with PKCE) Local connection bypass (localhost) API key fallback IP whitelisting (optional)
OAuth: Yes Scopes: No

Docs mention OAuth 2.0 with PKCE and token expiry (~1 hour) plus refresh support, and also mention a localhost bypass and API key/IP whitelist fallbacks.

Pricing

Free tier: No
Requires CC: No

No hosted pricing indicated; appears self-hosted (AGPL/commercial licensing). Costs depend on indexing and embedding runtime (e.g., local models or paid providers like OpenAI if used).

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • ChatGPT compatibility notes indicate some security validations are disabled (e.g., DNS rebinding protection and certain header/content-type validation), so agents should assume the server is tuned for ChatGPT connector behavior.
  • Indexing operations may be compute/storage heavy; agents should avoid repeatedly rebuilding indexes without checking index stats/logs.
  • Embedding/vector indexing is optional but may require additional local services/models; agent workflows should branch accordingly.

Alternatives

Full Evaluation Report

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

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 2026-04-04.

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