Skip to main content
Glama
tzervas

tero-mcp-lite

by tzervas
OverviewSchemaSourceRelated ServersScoreDiscussionsAdmin
Rust
Local

tero-mcp-lite

CI Security

A lightweight, portable MCP (Model Context Protocol) server over a Tero corpus index.json — the Python-only counterpart to the Rust tero-mcp binary in tero-rs (src/bin/tero-mcp.rs in the tero-rs crate, DN-87 / E39-1). It answers cited, provenance-carrying queries about a project's corpus (docs, decisions, issues, changelog, skills) over stdio JSON-RPC 2.0, with the same never-silent-refusal contract: an answer without a resolvable citation is a typed refusal, never a silent empty result (DN-87 §6.2).

This package is deliberately "lite": it has no runtime dependencies (stdlib json/argparse only), reads a committed index.json rather than building one, and implements only the Layer-1 (deterministic-index) query surface — no VSA/Layer-2 semantic memory. It is meant to drop into any repo that has (or generates) a Tero-shaped index, not just Mycelium's own.

What it is / isn't

  • Is: a thin, honest query engine + MCP stdio front over a pre-built index.json. Five query kinds (query_by_id, query_by_status, query_by_kind, cross_ref, text_search) plus cite/explain/identify/refresh — nine tools total, matching the Rust server's tool surface and JSON envelope shapes exactly (see "Matching the Rust server" below).

  • Isn't: an index builder. Regenerating index.json for your repo is a separate concern — see GENERATING-AN-INDEX.md.

  • Isn't: Layer-2 (VSA semantic search). identify always reports layer2_enabled: false. If you need that, use the full Rust tero-rs tero-mcp binary this package delegates to when present. Memory tools (memory_store / memory_retrieve / memory_consolidate) are tero-rs-only — build with Cargo feature memory, scopes memory-read / memory-write, runtime TERO_MEMORY_ENABLED, TERO_MEMORY_DB, optional TERO_MEMORY_MODEL — see docs/MEMORY_TOOLS.md; lite parses the same scopes but refuses memory tools/call honestly.

Related MCP server: Atlas

Install

Requires Python >= 3.11 and uv.

cd packages/tero-mcp-lite
uv sync

This creates .venv and resolves the (dev-only) dependency group — pytest for the test suite. The runtime server itself has zero third-party dependencies; uv sync --no-dev installs nothing but the package itself.

Run directly:

TERO_TOKENS='devtoken:read' uv run tero-mcp-lite --index /path/to/index.json

Or via the console-script entry point once installed (uv tool install . / pip install .):

TERO_TOKENS='devtoken:read' tero-mcp-lite --index /path/to/index.json

--index defaults to docs/tero-index/index.json (relative to the process's working directory), and can also be set via TERO_INDEX_PATH.

Register in .mcp.json (persistent use in this repo)

The repo-root .mcp.json already registers a tero server (see the top-level file). The entry:

{
  "mcpServers": {
    "tero": {
      "command": "uv",
      "args": [
        "run", "--project", "packages/tero-mcp-lite",
        "tero-mcp-lite", "--index", "docs/tero-index/index.json"
      ],
      "env": {
        "TERO_TOKENS": "local-dev:refresh"
      }
    }
  }
}

Claude Code (and any other MCP-aware client) picks this up automatically for sessions rooted at the repo. Rotate TERO_TOKENS for anything beyond local/dev use — never commit a real secret token; the placeholder above is intentionally a non-secret local-dev value, matching the Rust server's own "refuses to start without tokens, but the token value itself is just an opaque bearer string" model.

Auth

Exactly like the Rust server: set TERO_TOKENS (or TERO_TOKENS_FILE, a path to the same grammar) — a whitespace/comma-separated token:scope list, e.g. s3cr3t:read other:refresh. refresh implies read. The server refuses to start with no tokens configured — there is no anonymous default. Every tools/call carries its own token argument (checked against the operation's required scope before dispatch) — auth is per-call, not per-connection, matching the Rust server's model exactly.

Generating an index for any repo

See GENERATING-AN-INDEX.md for the index.json schema and how to produce one — either with Mycelium's own Rust tero-index binary, or a from-scratch tool in your own repo that emits the same shape.

Matching the Rust server

This package was built by reading the tero-rs tero-mcp front (src/bin/tero-mcp.rs and src/front/{core,mcp,auth}.rs, plus src/model.rs / src/query.rs) and mirroring:

  • the same nine tools, same inputSchema shapes, same required arguments;

  • the same JSON-RPC transport: newline-delimited JSON-RPC 2.0 over stdio, initialize → tools/list → tools/call, MethodNotFound (-32601) for anything else;

  • the same envelope shapes (answer/citations/explain/refusal/error);

  • the same refusal semantics: no_match / unknown_anchor / no_text_match, each carrying candidates_scanned and a human-readable message — DN-87 §6.2's contract enforced the same way (an Answer-shaped dataclass simply cannot be constructed with zero items; every query function raises a typed Refusal instead); the same cross_ref BFS over depends_on/doc_refs edges (issue-only depends_on targets, corpus:DOC[#anchor]-only resolvable doc_refs, same dedup-suffix anchor-matching rule, same MAX_CROSSREF_DEPTH=6 clamp reported in Explain.query, never silently);

  • the same text-search scoring (id match x4 + title match x3 + summary match x1 per matched term, ties broken by canonical (family, file, line, anchor) order, capped to 20 results);

  • the same token-scoped auth model (per-call token argument, read/refresh scopes, refuse to start with no tokens).

The wire-visible shapes checked by tests/test_rust_parity.py — tool descriptor JSON (values and key order), the JSON-RPC error code mapping, auth-error message wording, refusal variant tags, and the process exit codes — are transcribed verbatim from the Rust source and asserted byte-for-byte equal; a diff there means real drift, not a documentation gap. Where behavior could plausibly diverge structurally (the cross_ref clamp-reporting rule, the is_dedup_suffix_of anchor-matching grammar, the Family sort-rank order used for the canonical key, the text-search scoring weights) this package copies the Rust logic structurally, not just by description, specifically to avoid silent semantic drift.

The one deliberate, documented divergence is the identify tool's payload (not its tool descriptor, which does match): name/engine/summary describe this Python server's own identity ("tero-mcp-lite", Layer-1-only) rather than claiming to be the Rust binary — an honest identity beats a byte-identical lie. tests/test_rust_parity.py does not assert this field, and the serverInfo.name in initialize is likewise "tero-mcp-lite", by design.

The parity test suite is a transcription-derived check, not a live differential (no checked-out Rust source is available at Python test/CI time here) — see "Framework — remaining tasks" below for the live-differential follow-up.

Why a minimal implementation instead of the official mcp Python SDK

The official mcp SDK (PyPI mcp) does install cleanly via uv with no version conflicts — it was evaluated. It was not used here for three concrete reasons:

  1. Weight vs. the "lite"/portable goal. mcp pulls in ~30 transitive packages (pydantic, pydantic-core, cryptography, starlette, uvicorn, sse-starlette, ...) — mostly HTTP/SSE transport machinery this package doesn't use (stdio only). A package meant to be zipped and dropped into an arbitrary repo is better served staying small.

  2. Exact semantic control. The Rust server's auth model is unusual for MCP: the bearer token is a per-tools/call argument, not a transport-level header, and an auth/bad-request failure is a top-level JSON-RPC error, not an isError:true tool result. The SDK's high-level @server.call_tool() decorator catches all exceptions (including a raised McpError) and turns them into isError:true tool results by default — matching the Rust behavior exactly would mean bypassing that decorator and registering a raw low-level request handler anyway, which erodes most of the SDK's convenience value for this specific shape of server.

  3. Zero dependency-conflict risk, trivially auditable. ~700 lines of pure-stdlib Python across 5 files is easy to read start to finish and carries no supply-chain surface beyond the interpreter.

uv is still used as a real project/dependency manager (uv.lock, [dependency-groups] dev carrying pytest) — this isn't a bare script; it's satisfied at the project-management layer rather than by adding runtime weight the package doesn't need. If a future maintainer wants full MCP-spec coverage (resources, prompts, sampling, elicitation, streamable-HTTP transport, ...), switching to the mcp SDK is a reasonable evolution — see "Framework — remaining tasks" below.

Adding a new tool (the registry pattern)

src/tero_mcp_lite/mcp_server.py derives both tools/list's descriptors and tools/call's dispatch + auth-scope check from one declarative TOOL_REGISTRY: dict[str, ToolSpec]. Adding a tool means adding one ToolSpec — nothing else in the file changes.

from tero_mcp_lite.mcp_server import ToolSpec, TOKEN_ARG

def _handle_my_tool(state: McpState, args: dict) -> dict:
    ...  # read args, touch state.report, return a JSON-able dict

my_tool = ToolSpec(
    name="my_tool",
    description="One line: what it does.",
    properties={"some_arg": {"type": "string"}, "token": TOKEN_ARG},
    required=("some_arg", "token"),
    handler=_handle_my_tool,
    # scope=Scope.REFRESH,  # omit for the default (read-only)
)

then add my_tool to the specs list in _build_registry(). That's it:

  • tools/list advertises it automatically (ToolSpec.descriptor(), derived — see _tool_descriptors()).

  • tools/call authorizes against scope (default Scope.READ) and dispatches to handler automatically (_handle_tools_call()).

  • A new query kind (as opposed to a new top-level tool) is a smaller change: add the kind to tero_mcp_lite.query.Query.parse + a _<kind>() function in query.py (mirroring the existing _by_id/_by_status/_cross_ref/_text shape — refuse on an empty match set, always attach an Explain trace), then wire a ToolSpec (or extend cite/explain's kind argument, since those already forward whatever kind string they're given) exactly as above.

  • Extend tests/test_rust_parity.py's RUST_TOOL_DESCRIPTORS (and, if the tool has a Rust-side twin, transcribe its exact wording from source) so the new tool's shape stays pinned too.

Tests

uv run pytest

Covers: a JSON-RPC round-trip (initialize → tools/list → query_by_id returning a cited answer), a refusal test (an uncited query returns a typed refusal, never an empty result), unit coverage for auth/query/model, and tests/test_rust_parity.py — the Rust-source-transcribed byte-level checks described above. All fast and fully offline (an in-memory synthetic index — no network, no real repo required).

Framework — remaining tasks

A checklist for whoever picks this up next (in this repo or an extracted one):

  • Byte-level parity harness — transcription version. tests/test_rust_parity.py pins the tool descriptor JSON (values + key order), the JSON-RPC error code mapping, auth-error wording, refusal variant tags, and exit codes as verbatim transcriptions of the Rust source. This caught two real wording bugs on introduction (an identify tool description with an extra clause, ... where the Rust source uses …) — evidence the check has teeth.

  • Byte-level parity harness — live differential. The stronger version: a test that actually runs both the Rust tero-mcp binary and this package over the same index.json and diffs their JSON-RPC responses field-by-field, so a future Rust-side wording change is caught automatically instead of requiring a human to notice and re-transcribe. Needs a Rust toolchain + a built tero-mcp binary available at test time, which this repo's own CI does not provide (this package is meant to be extracted/dropped into other repos, most of which won't have the mycelium Rust crate either) — plausibly a mycelium-repo-side CI job instead of a tero-mcp-lite-side one.

  • HTTP front. The Rust crate also ships tero-http (a plain HTTP/JSON front sharing the same core). This package only implements the MCP/stdio front; an HTTP front (e.g. http.server or a minimal ASGI app) is a natural, still-lightweight follow-up if a non-MCP client needs it.

  • Layer-2 / VSA. Deliberately out of scope (layer2_enabled is hardcoded false). If the VSA semantic layer (DN-87 §2 fork 1) ever needs a Python-native front, that's new work, not a gap in this package.

  • refresh hot-reload race. _refresh swaps state.report between requests; this server is single-threaded/single-client over stdio (matching the Rust server's own single-threaded stdio model), so there's no concurrency hazard today — flag if this is ever adapted to a multi-client transport.

  • Consider the mcp SDK if/when this needs full MCP-spec surface (resources, prompts, sampling) beyond tools — see the tradeoff write-up above; the SDK does install cleanly with uv.

  • Packaging polish. Currently zipped as source (uv sync on first use in the target repo). A uv build-produced wheel could be attached to a release instead, if the target repo prefers not to keep a pyproject.toml + src/ tree around.

  • Security scans + hardening. This package has not been run through a dedicated supply-chain/security scan in this environment (api.x.ai and most external scan tooling are unreachable from this repo-scoped session) — see packages/GROK-HANDOFF.md at the repo root for the runbook to do that on infrastructure that can reach it.

Contact

Maintainer contact for this package: tz-dev@vectorweight.com · github.com/tzervas. (This is a swap-able project email/handle — update pyproject.toml's [project.authors]/[project.urls] and this section if ownership moves.)

License

MIT — see the repository root LICENSE (or add one in an extracted repo; ADR-022 §7 / CONTRIBUTING §Licensing require MIT-only for first-party Mycelium artifacts, and this package inherits that posture as a Mycelium-repo artifact).

Status & roadmap

Semver + Releases

Current release: v0.2.0 (2026-07-21).

Source

Version

pyproject.toml / tero_mcp_lite.__version__

0.2.0

rust/Cargo.toml (CARGO_PKG_VERSION / MCP serverInfo)

0.2.0

GitHub Release / tag

v0.2.0

Prior: v0.1.0 baseline, v0.1.1 first package release. PyPI not published (install from git/uv path). Process + cites: docs/ROADMAP.md semver section.

Available Tools

9 tools
citeD

Citations only for a query (kind + its args, as query_*).

ParametersJSON Schema
NameRequiredDescriptionDefault
kindYesid|status|kind|cross_ref|text
depthNo
startNo
tokenYesbearer token (from TERO_TOKENS)
valueNo

TDQS

D1.8/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations provided, and the description does not disclose behavioral traits such as side effects, authentication requirements (beyond the token parameter), rate limits, or outcome format.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is extremely concise (one sentence), but it sacrifices essential information. It is under-specified rather than effectively concise.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness1/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given five parameters, no output schema, and no annotations, the description fails to explain how to use the parameters or what the tool returns. It is completely inadequate for an agent to invoke correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 40% (only 'kind' and 'token' have descriptions). The description adds no additional meaning beyond the schema, e.g., no clarification for 'depth', 'start', or 'value'.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose2/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description is vague: 'Citations only for a query (kind + its args, as query_*).' It does not clearly state what the tool does (e.g., retrieve citations, cite something). The phrase 'as query_*' is ambiguous and does not specify the verb or resource.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines1/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance on when to use this tool versus its siblings (cross_ref, explain, identify, query_by_kind, etc.). The description offers no context for selection.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

cross_refC

Breadth-first walk of depends_on/doc_refs edges from a start id/anchor.

ParametersJSON Schema
NameRequiredDescriptionDefault
depthNohop count (default 1)
startYes
tokenYesbearer token (from TERO_TOKENS)

TDQS

C2.6/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations provided, and the description only states the traversal algorithm. It omits side effects, authentication requirements (token noted in schema but not description), rate limits, or output format, which is insufficient for a no-annotation tool.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single sentence that is concise and front-loaded, but it could include more detail without becoming verbose. It is minimal but not overly long.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given three parameters, no output schema, and no annotations, the description is incomplete. It fails to explain the return type, depth format, or any behavioral nuances of the graph walk.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 67%, but the tool description adds no parameter information beyond what the schema provides. The meaning of 'depth' as a string vs integer or the role of 'token' is not clarified.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description specifies a breadth-first walk over specific edge types from a starting point, clearly indicating the tool's function and distinguishing it from sibling tools like text_search or cite.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance on when to use this tool versus alternatives. The description does not mention context, prerequisites, or exclusions, leaving the agent to infer usage.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

explainC

EXPLAIN trace only for a query (kind + its args, as query_*).

ParametersJSON Schema
NameRequiredDescriptionDefault
kindYes
depthNo
startNo
tokenYesbearer token (from TERO_TOKENS)
valueNo

TDQS

C2.5/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description lacks behavioral details such as whether the tool is read-only, requires special permissions, or has side effects. The term 'trace' hints at analysis, but no explicit safety or mutation information is given. With no annotations, the description fails to provide necessary transparency.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single, concise sentence, but it is under-specified. While brevity is good, it sacrifices clarity and completeness on crucial details.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness1/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Without an output schema or annotations, and with low schema coverage, the description is severely incomplete. It does not explain what an 'EXPLAIN trace' returns, how to use parameters, or any contextual prerequisites for the agent.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is low (20%) with only token described. The description adds minimal parameter meaning: it hints that 'kind' relates to query kinds and mentions args, but does not explain other parameters like depth, start, or value. This does not compensate for the schema gaps.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description identifies the tool as performing an 'EXPLAIN trace' for a query, mentioning 'kind + its args, as query_*'. This clearly states the verb and resource, and the mention of 'query_*' suggests a specific scope that differentiates from sibling tools like query_by_id or query_by_kind.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance is provided on when to use this tool versus alternatives. The sibling tools include various query retrieval and search tools, but the description does not specify contexts where explain is appropriate or when other tools should be used instead.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

identifyB

Server identity, version, and whether the Layer-2 gate is open.

ParametersJSON Schema
NameRequiredDescriptionDefault
tokenYesbearer token (from TERO_TOKENS)

TDQS

B3.2/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Without annotations, the description only hints at a read operation but doesn't disclose effects, permissions, or side effects.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Single sentence, efficient and to the point, though slightly terse.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Adequately describes output but lacks detail on return format or structure.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema covers token with description; tool description adds no new parameter insight but is baseline adequate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states it returns server identity, version, and gate status, which is specific and distinct from sibling tools like cite, query, etc.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance on when to use this tool versus alternatives, or prerequisites such as authentication.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

query_by_idA

Exact lookup by corpus id (RFC-0034, M-1015, DN-87, an issue id).

ParametersJSON Schema
NameRequiredDescriptionDefault
tokenYesbearer token (from TERO_TOKENS)
valueYesthe id to match

TDQS

A3.7/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations provided, so description must disclose behavioral traits. Only states 'lookup', implying read operation, but does not confirm idempotency, error behavior on missing ID, or whether it's read-only. Lacks detail beyond basic function.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Single concise sentence with no unnecessary words. Front-loaded with key action and resource.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a simple lookup tool with no output schema or annotations, description is adequate but missing expected output format or failure behavior. Could mention that it returns a single document or nothing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Despite 100% schema coverage, description adds value by providing concrete ID examples (RFC-0034, etc.) that clarify expected format beyond schema descriptions.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Description clearly states 'Exact lookup by corpus id' with specific examples (RFC-0034, M-1015, DN-87), distinguishing it from sibling tools like text_search or cross_ref which are not exact id lookups.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Implies use for exact ID matching, but no explicit when-to-use or when-not-to-use guidance. Does not mention alternatives like query_by_kind for type-based queries.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

query_by_kindC

All rows of a given kind (rfc, adr, note, issue, section, …).

ParametersJSON Schema
NameRequiredDescriptionDefault
tokenYesbearer token (from TERO_TOKENS)
valueYes

TDQS

C2.9/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description must disclose behavior. It states it returns all rows, but does not mention read-only nature, pagination, error conditions, or any side effects. Minimal disclosure.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single clear sentence with no fluff. It is appropriately sized for a simple query tool, though it could be expanded slightly without losing conciseness.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a tool with two required parameters and no output schema, the description is incomplete. It does not explain the return format, possible values for the kind parameter, or how it differs from sibling tools. Given the context of multiple sibling tools, more guidance is needed.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 50% (only token has a description). The description adds meaning to the value parameter by listing example kinds, partially compensating. However, it does not specify allowed values precisely or explain the token parameter beyond what schema provides.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states that the tool returns all rows for a given kind, and provides examples of kinds (rfc, adr, etc.). However, it does not distinguish itself from sibling tools like query_by_status or query_by_id, missing a chance to differentiate.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance is given on when to use this tool versus alternatives. The description only states what it does, but does not explain context or exclusions, leaving the agent to infer usage.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

query_by_statusB

All rows with a given status (Accepted, todo, done, …).

ParametersJSON Schema
NameRequiredDescriptionDefault
tokenYesbearer token (from TERO_TOKENS)
valueYes

TDQS

B3.2/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden. It fails to disclose whether the operation is read-only, what side effects exist, or any behavioral traits beyond returning rows. The agent cannot infer safety or mutability from this description.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single, efficient sentence with no unnecessary words. It front-loads the key purpose and uses ellipsis to indicate a non-exhaustive list, maintaining conciseness.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given no output schema, the description should ideally mention the return format (e.g., list of rows, row structure). It also doesn't repeat the token requirement from the schema, but the schema already documents that. Overall, adequate but missing output details.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The description adds significant meaning to the 'value' parameter by providing examples of statuses (Accepted, todo, done), which the input schema lacks. This helps the agent understand expected values beyond the bare type string.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly indicates the tool returns all rows filtered by a status value, distinguishing it from siblings like query_by_id and query_by_kind. However, it does not explicitly state the verb (e.g., 'query' or 'list'), making the purpose slightly less crisp.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides no guidance on when to use this tool versus alternatives such as query_by_id or text_search. There are no exclusion criteria, prerequisites, or explicit usage contexts.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

refreshB

Reload the served index from disk (requires the refresh scope).

ParametersJSON Schema
NameRequiredDescriptionDefault
tokenYesbearer token (from TERO_TOKENS)

TDQS

B3.1/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description must bear full behavioral context. It states the tool reloads from disk and requires a scope, but does not disclose side effects (e.g., temporary unavailability, idempotency, performance impact) or response behavior. The description is insufficient for an agent to fully understand the consequences of invocation.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single sentence that front-loads the core action ('Reload the served index from disk') and includes a necessary requirement in parentheses. No extraneous words, making it highly concise and efficient for an agent.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool has a simple signature (1 required parameter, no output schema, no nesting), the description covers the basic action and a prerequisite. However, it does not explain what 'served index' means or what the tool returns after reload, which could leave gaps for an agent unfamiliar with the system.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, with the single parameter 'token' described as 'bearer token (from TERO_TOKENS)' in the schema. The description adds no additional semantics beyond the schema, so it does not exceed the baseline score of 3.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description 'Reload the served index from disk' uses a specific verb and resource, clearly indicating a reload operation. However, it lacks context on what the 'served index' refers to, which could cause ambiguity without domain knowledge. It does not differentiate from siblings, but sibling tools are distinct in function.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description mentions a required scope ('refresh' scope) but offers no guidance on when to use this tool versus alternatives. There is no mention of prerequisites, appropriate contexts, or situations to avoid. Sibling tools like 'query_by_id' or 'text_search' have different purposes, but no explicit comparison is made.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

text_searchA

Ranked free-text search over id/title/summary.

ParametersJSON Schema
NameRequiredDescriptionDefault
tokenYesbearer token (from TERO_TOKENS)
valueYesthe query text

TDQS

A3.8/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description adds value by specifying 'ranked' and the fields searched, but with no annotations, more behavioral details (e.g., result limits, case sensitivity) would be helpful.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

A single, front-loaded sentence with no wasted words. Every word contributes meaning.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's low complexity (2 simple params, no output schema), the description covers the core purpose adequately. Additional details like pagination or special characters would be nice but not essential.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema covers both parameters fully (100%), so baseline is 3. The description adds no extra param semantics beyond what the schema provides.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the verb 'search' and the resource 'free-text over id/title/summary', distinguishing it from sibling tools like query_by_id and query_by_status which are structured queries.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description implies usage for free-text search but provides no explicit guidance on when to use vs. alternatives or when not to use. The list of sibling tools exists but is not referenced.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.

  1. 9 tool updatesv0.1.0
    • First observedcite
    • First observedcross_ref
    • First observedexplain
    • First observedidentify
    • First observedquery_by_id
    • First observedquery_by_kind
    • First observedquery_by_status
    • First observedrefresh
    • First observedtext_search
Install Server
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

TDQS

B3/5.0
Disambiguation5/5

Each tool serves a clearly distinct purpose: citation, graph traversal, explain, identity, various queries, refresh, and text search. There is no functional overlap that would confuse an agent.

Naming Consistency3/5

Tool names are a mix of single-word verbs (cite, explain, refresh) and multi-word patterns (query_by_id, query_by_kind). While all are readable, there is no consistent verb_noun or pattern across the set, causing moderate inconsistency.

Tool Count5/5

Nine tools is an appropriate count for a knowledge base server. Each tool covers a needed operation without being excessive or sparse.

Completeness4/5

The tool set covers querying by ID, kind, status, text search, graph traversal, citation, and explain. For a read-only documentation system, this is comprehensive. Minor gaps like aggregation are acceptable.

Maintenance

ActivitySlowing
ResponsivenessUnresponsive

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Connectors

Related MCP Servers

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/tzervas/tero-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server