readscout
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": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| local_askA | Read files or directories with a local model so you don't have to; get back a short answer. Use instead of Read/Grep when you need facts from large or many files but not the files themselves ("which routes lack auth", "where is X handled", "what does this module do"). Each fact comes with a quote verified verbatim in the source and its real path:line; facts the model could not quote are dropped. mode='summary' adds a short unverified summary. quality: leave unset (auto: the bigger model for negative questions or when the fast one finds nothing), True to force it. Do not use for decisions (security, correctness) without opening the cited lines. |
| local_digestA | Triage a long log or command output file (CI log, test output, server log) locally. Returns the root-cause error and distinct failures with verified path:line quotes, instead of loading the whole file into context. Tip: redirect a noisy command to a file, then digest it. |
| local_editA | Apply a small, fully specified mechanical change with a local model (docstrings, comments, type hints, renames, boilerplate, test scaffolding, applying a decided pattern). Pass paths, not contents: do not Read the files first. gate = commands (argv lists, at most 5, no shell, run with the user's rights: only the project's own tests or linters) that must pass, e.g. [["python","-m","pytest","-q"]]; if they fail after one retry the files are restored and the result is ESCALATE. Returns APPLIED with the diff to review, or ESCALATE. Never for design, debugging or security-sensitive code. |
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 3 tools
local_ask and local_digest both use a local model to extract info from files, which creates mild overlap, but their descriptions clearly separate general code Q&A from log/command-output triage. local_edit is unmistakably distinct as a mutation tool with its own gating semantics.
All three tools share the same local_ prefix followed by a single verb/noun (ask, digest, edit), giving a fully predictable pattern. No mixing of conventions.
Three tools is a tight, well-scoped set for a focused helper: query files, triage logs, and apply mechanical edits. Each tool is substantial and earns its place with no redundancy.
The surface covers the core read/query, triage, and edit lifecycle for local model assistance, which matches the stated purpose. There is no explicit multi-file batch or non-mechanical edit path, but that exclusion appears deliberate and documented rather than a gap.