reforge-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| extensions | {
"io.modelcontextprotocol/ui": {}
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| scan_repo_toolC | Scan a repository and return a structured analysis report. |
| get_chunk_toolB | Retrieve a line-range slice from a file. |
| write_fix_toolB | Apply a unified diff to a file; run optional tests and auto-rollback on failure. |
| git_commit_toolB | Create an atomic git commit; enforces session budget and confirmation checkpoints. |
| read_memory_toolB | Read a value from reforge-state.json. Returns {key, value, found}. |
| write_memory_toolA | Write a value to reforge-state.json atomically. Returns {success, key, previous_value}. ttl_seconds is accepted but not enforced (stored as metadata only). |
| infer_architecture_toolA | Infer the architectural pattern of a repository. Runs scan_repo, detects pattern (REST API, CLI tool, library, web frontend), identifies entry points and logical modules, stores the result in reforge-state.json under architecture_hypothesis, and returns the inference. |
| score_risk_toolA | Score the risk of changing a symbol (0 = safe, 100 = very risky). Factors: inbound references, test coverage, file size, circular deps. Returns score, per-factor breakdown, and a recommendation string. |
| get_health_score_toolA | Compute a 0-100 health score from scan metrics and append to health_history. Scoring deductions: duplicate ratio (-30 max), dead code ratio (-30 max), god files (-20 max), circular deps (-20 max). |
| generate_changelog_toolC | Write REFORGE_CHANGES.md from current session state. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 10 tools
Tools are mostly distinct: read/write memory, apply diff, commit, and risk scoring have clear boundaries. Scanning-related tools (scan_repo, infer_architecture, get_health_score) overlap in inputs/outputs but serve different purposes, which descriptions clarify.
All tool names follow a uniform verb_noun_tool pattern (e.g., scan_repo_tool, write_fix_tool, read_memory_tool). The suffix is consistent and the verb-object structure is predictable, making it easy to guess tool behavior.
The server has 10 tools, which is well within the ideal 3-15 range. Each tool addresses a distinct need in the repository analysis and modification workflow, and none feel redundant or superfluous.
The tool set covers the core lifecycle: scanning, analysis (architecture, risk, health), editing (diff apply), committing, memory persistence, and changelog generation. Minor gaps exist, such as no explicit file listing or branch management, but these can be worked around using existing tools like get_chunk.