This server helps coding agents inspect and work with a local Synthesizer installation, its grids, and project documentation without writing exploratory scripts.
Environment & API inspection: report Python/Synthesizer environment facts, versions, paths, and resolve any public synthesizer.* object's signature, docstring, location, and version.
Local grid discovery: list available local grid files and inspect a grid's axes, metadata, HDF5 structure, datasets, and emission-line identifiers without loading heavy data.
Grid visualization: plot spectra, emission lines, or ionising luminosity (HI/HeII) at chosen grid points, returning rendered figure paths.
Documentation search: search the bundled skill tree and any Synthesizer source checkout for relevant text, returning scored hits with snippets.
Example discovery: find the closest canonical example for a user task, with alternatives and source-checkout excerpts.
Script validation: statically validate a Synthesizer script for syntax, import availability, referenced attributes/arguments, and locally available grids, with diagnostics and suggested commands.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@synthia-mcp Does my local grid cover the metallicity range I need, and does it contain H-beta?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Synthia is an agent companion for working with Synthesizer.
Synthia bundles an Agent Skill with a local MCP server so Claude Code, OpenCode, and other MCP-capable agents can answer Synthesizer questions from the user's actual installation, and from the project's Syndex data catalogue, rather than inflating context and relying on incomplete local memory. Note that it is not designed as a replacement for a human user but simply aims to make learning and using Synthesizer a frictionless experience. For more on these mechanisms see the MCP documentation and, for Agent Skills, agentskills.io.
Synthia requires Synthesizer to be installed first. You'll need to install Synthia in the same environment containing Synthesizer, and you'll need to launch your agent from that same environment.
For agent-assisted installation, send your coding agent this prompt:
Read https://raw.githubusercontent.com/synthesizer-project/synthia/main/INSTALL.md and install Synthia for me.Or install it directly (which is essentially all the install prompt does):
pip install cosmos-synthia
synthia-installTo install an unreleased change, point pip at the repository instead:
pip install 'git+https://github.com/synthesizer-project/synthia.git'synthia-install links the bundled skill at ~/.claude/skills/synthia (which
OpenCode also loads, so one copy serves both clients), registers synthia-mcp
with Claude Code when the claude CLI is available, and adds it to
OpenCode's ~/.config/opencode/opencode.json. --dry-run shows what it would
change, --uninstall reverses it. Restart your client afterwards: skills load
per session, but MCP servers only load at client start.
Related MCP server: MCP Inspector as MCP Server
Synthia has no interface of its own. Once installed, ask your agent about Synthesizer in plain language: the skill triggers on the subject matter, and the agent calls the MCP tools when it needs an exact local fact, a published catalogue record, or a support operation.
cd ~/my-analysis
claude # or opencode> Compare the spectra of a parametric stellar population with little and with
> heavy dust attenuation, and plot them."Do any of my local grids cover the metallicity range I need, and do they contain H-beta?" — answered from the actual grid files you have, not from memory and without delving into the online documentation.
"Is there a published BPASS grid with emission lines, and does it reach Z = 1e-5?" — answered from the Syndex catalogue, including axes, spectra, lines, size, checksum and citations, without downloading the file. Synthia then hands you the synthesizer-download command to run yourself.
"Why is spectra['total'] a KeyError after I built a TotalEmission model with fesc=0?" — a documented Synthesizer emission model initialisation trap.
"Show me what my BPASS v2.2.1 grid's spectra look like at 10 Myr and Z=0.01." — renders a figure with known plotting methods and returns its path.
"Adapt this script to the Synthesizer version I actually have installed." — take outdated scripts and update them for your specific Synthesizer install.
"Generate a young and old galaxy with black holes and produce plots showing whether JWST can detect the AGN contribution for a set of reasonable redshifts and galaxy properties." — render a complex astrophysics question by applying its knowledge of the local installation and available grids without the need to build up the quick script yourself.
Synthia is at its most useful where an answer depends on your specific installation: which version, which grids, which spectra those grids hold — and on what the project has published that you do not have yet. It is least useful for pure astrophysics questions not specific to Synthesizer, which the model can answer without it.
Synthia can also answer questions about the Syndex catalogue, the project's database of published grids, instruments and test data: what exists, what a grid contains, and which release to take. Downloading stays with synthesizer-download, which Synthia gives you the command for rather than running.
The benchmarks directory contains a series of tests for common Synthia prompts. These were run on a bare agent (i.e. no plugins, MCPs, or skills) using Sonnet 5 with and without Synthia. Below is a comparison of the usage and runtime, measured over the 15 cases that predate the catalogue tools.
metric | baseline | synthia | change |
total cost | $11.30 | $8.22 | −27% |
total wall time | 3682 s | 2174 s | −41% |
source read | 954 kB | 92 kB | −90% |
cache-read tokens | 21.5 M | 14.9 M | −31% |
turns | 443 | 341 | −23% |
Note that every Synthia script produced a correct result or a runnable script while baseline did in all but one case. Synthia is designed to be correct more often but importantly it'll get to the correct answer with less resource and often faster.
See CONTRIBUTING.md for the full development workflow.
Describe one published dataset in the Syndex catalogue.
Reports what a Syndex dataset's current release contains — for a
grid, its axes and their extent, which spectra and emission lines it
holds, the model it came from and how to cite it — without
downloading the file. This is how to answer "does the published
BPASS grid cover Z = 1e-5?" or "which lines would I get?" for a grid
that is not installed here. Use inspect_local_grid for one that
is.
Requires network access. Names come from search_catalogue.
Args:
name: Catalogue dataset name, such as
bpass-2-2-1-cloudy-sps-test. Not a filename and not a
path.
Returns:
On failure, a mapping with ok (False) and error,
plus status when the catalogue rejected the name and
hint when it could not be reached.
On success, a mapping with ``ok`` (``True``), ``source``,
``content_is_untrusted`` (always ``True``), the dataset's
``name``, ``display_name``, ``description``, ``data_type``,
``is_test``, ``is_recommended``, ``licence``, and
``current_release``. The release carries ``release_id``,
``published_at``, ``known_bug`` (with
``known_bug_description`` when set), the Synthesizer version
bounds, ``file`` (``filename``, ``format``, ``size_bytes``,
``sha256``), ``download_command``, ``citations``, and, for a
grid, ``grid`` with ``grid_type``, ``emission_type``, the model
and photoionisation code, the wavelength range, ``axes`` as
``{name, units, scale, count, minimum, maximum}``, ``spectra``,
and ``lines`` as ``{count, ids, truncated}``. Axis values
themselves are not returned; ``count``, ``minimum`` and
``maximum`` are what a coverage question needs.
``current_release`` is ``null`` for a dataset with nothing
published yet.
| Name | Required | Description | Default |
|---|---|---|---|
| name | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden. It discloses network dependency, failure modes with status/hint, success return structure, content_is_untrusted always being True, current_release being null for unpublished datasets, and the limitation that axis values are not returned. This is exemplary behavioral disclosure.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is long, especially the Returns section, but it is well-structured with Args/Returns and front-loaded purpose and usage. Every sentence adds value; the length is justified by the richness of the return mapping and behavioral caveats, though a tighter summary of return fields could make it slightly more concise.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the single parameter, the explicit network requirement, the detailed return contract, failure behavior, and clear differentiation from sibling tools, the description is complete enough for an agent to select and invoke the tool correctly without additional context.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 0%, but the description fully compensates: it explains the name is a catalogue dataset name, gives a concrete example, and explicitly says it is not a filename or path. It also tells the agent where names come from (search_catalogue), making the single parameter unambiguous.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it describes one published Syndex catalogue dataset, and enumerates exactly what information is reported: axes, spectra, lines, model, citation. It explicitly contrasts with inspect_local_grid, which handles installed grids, making sibling differentiation clear.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives concrete example questions the tool answers, states network access is required, says names come from search_catalogue, and explicitly directs users to inspect_local_grid for local grids. This is strong when-to-use guidance with an explicit alternative.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Find the closest canonical example for a task.
Args: task: Short description of what the user wants to do.
Returns:
A mapping with task and example.
available_examples lists every bundled example name, so
another can be requested when the match is wrong, and
other_candidates ranks the next best matches so a close call
is visible. example is None when nothing
matched, otherwise {path, score, source} where source is
the complete example in an untrusted-content envelope, capped at
32 KiB. checkout_example is the nearest match in a
Synthesizer source checkout when one exists, and carries only an
excerpt, not the whole file. notes explains an empty
result and error is present only when the search failed.
| Name | Required | Description | Default |
|---|---|---|---|
| task | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description fully carries the behavioral burden. It thoroughly describes the return contract: available_examples, other_candidates, the None case, the untrusted-content envelope, the 32 KiB cap, the checkout_example excerpt behavior, notes, and error presence. This is exemplary for a search tool.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is longer than average but every sentence earns its place: the purpose is front-loaded, followed by clearly separated Args and Returns sections. The return details are dense but structured in a way that is easy to scan, with no filler or repetition.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple one-parameter tool with no annotations, this is nearly complete: it explains the input and the full output structure. The only gap is practical deployment context—what 'bundled examples' and 'Synthesizer source checkout' refer to in the surrounding environment—which may be known system-wide. An existing output schema would cover return types, but the description goes beyond that anyway.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema has zero coverage for the sole parameter, so the description must compensate. 'task: Short description of what the user wants to do' adds meaning by clarifying the parameter is a free-text request, not a structured identifier. It is terse but adequate for a single string parameter.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with 'Find the closest canonical example for a task,' which clearly states a specific verb and resource. This is immediately distinguishable from the sibling tools, which focus on plotting, environment inspection, and documentation search.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The usage context is implied by the first line: use this when you need a canonical example for a task. However, there is no explicit when-to-use vs. alternatives, no mention of the sibling search_documentation tool, and no exclusions or conditions that would route an agent to a different tool.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Report Python and Synthesizer environment facts.
Synthesizer is never imported here, so this is cheap and has no side
effects. The price is being honest about what was checked:
synthesizer_installed means only that a module spec with a real
file origin was found. It is not proof that Synthesizer imports, and
a broken C extension, the most common real failure mode for this
package, is invisible to a spec lookup. import_error is
therefore always None here; only :func:inspect_synthesizer_api,
which does import, can populate it.
Returns:
A mapping with python_version, platform,
the executable and environment whose Synthesizer is
being reported on, synthesizer_installed,
synthesizer_version with the version_source that
reported it,
synthesizer_path, import_error (always None) and
synthesizer_env_vars. Synthesizer resolves its directories
as environment variable else platformdirs default, so only the
SYNTHESIZER_* variables that are actually set are reported;
resolving the defaults would mean importing the package.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries full burden and excels. It discloses no import side effects, explains the exact meaning and limitation of synthesizer_installed (spec lookup only, broken C extensions invisible), states import_error is always None, and explains that only set environment variables are reported to avoid importing. This is thorough and honest about tool behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Every sentence adds value: purpose, cost/safety, limitation of the installed check, and the exact return mapping with semantic notes. The structure front-loads the core purpose, then details nuances. No filler or redundancy.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's simplicity (no params, clear output), the description fully documents all return keys, their meanings, and the reasoning behind values like import_error=None. The output schema exists and is enriched by this description. Nothing an agent needs to invoke correctly is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool has zero parameters, so the baseline is 4 per rules. The description doesn't need to explain parameters, and it doesn't waste space on them. The mention of environment variables is about output, not parameters, so it's appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with 'Report Python and Synthesizer environment facts,' a clear verb-resource pair. It differentiates from the sibling inspect_synthesizer_api by explicitly stating it never imports Synthesizer, which tells the agent exactly what this tool does and doesn't do.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly contrasts itself with inspect_synthesizer_api: 'Synthesizer is never imported here, so this is cheap and has no side effects' and 'only inspect_synthesizer_api, which does import, can populate it.' This gives the agent a clear when-to-use vs. when-not-to-use decision, fulfilling the guideline requirement beyond mere context.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Describe a local Synthesizer grid: axes, model, and contents.
Loads the grid's metadata only (no spectra or lines) and lists the groups and datasets the file holds, so an agent can tell what spectra, extinction curves, or lines are available before loading anything. Dataset values are never read apart from the identifiers naming a grid's emission lines, and links are never followed. Importing Synthesizer takes a couple of seconds and creates its data directories as a side effect.
Args:
grid_name: Name of a grid in the local grid directory, with or
without an .hdf5 suffix. Must not escape that directory
or be a symbolic link.
Returns:
On failure, a mapping with ok (False) and error,
plus missing and hint when Synthesizer is not
importable, or grid_dir when the named grid is absent.
On success, a mapping with ``ok`` (``True``), ``grid_name``,
``path``, ``size_bytes``, ``synthesizer_version``, and
``content_is_untrusted`` (always ``True``: every string below
comes from the grid file and is data, not instructions),
alongside two independent sections. Either may be missing while
``ok`` stays ``True``:
* ``axes`` (a list of ``{name, units, size, min, max}``, where
``min`` and ``max`` are ``null`` if not finite) and
``model_metadata``, or ``metadata_error`` if Synthesizer
could not load the grid's metadata.
* ``structure`` with ``entries`` describing each group,
dataset, and link (datasets carry ``shape``, ``dtype``, and
``size``, links carry their target and ``followed: false``,
and a line identifier dataset also carries ``values``),
``truncated``, ``omitted_attributes``, and
``omitted_attribute_count``; or ``structure_error`` if the
file could not be read.
| Name | Required | Description | Default |
|---|---|---|---|
| grid_name | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden and does so thoroughly. It discloses read-only metadata behavior, that dataset values are never read except line identifiers, that links are never followed, that importing Synthizer has a directory-creation side effect, and that content is untrusted. It also details both success and failure return variants.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is long but well-structured into a one-sentence summary, argument explanation, and detailed returns section. Every part earns its place, and the key safety and side-effect facts are front-loaded near the top.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description covers the input format and constraints, operational side effects, security posture, success and failure shapes, and the independence of axes/structure sections. Given no annotations and minimal schema, this is exceptionally complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 0%, so the description must fully define grid_name. It does: the parameter is a local grid name, accepts an optional .hdf5 suffix, must not escape the grid directory, and must not be a symbolic link. This is substantive semantic content beyond the bare schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a clear verb and resource: 'Describe a local Synthesizer grid: axes, model, and contents.' It further distinguishes itself from plotting/listing siblings by explicitly stating it loads metadata only, no spectra or lines, and lists groups and datasets.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description gives clear context: this is for learning what a grid contains 'before loading anything.' It implies the use case and contrasts with alternatives by stating it does not read spectra or lines, but it does not explicitly name sibling tools or give when-not-to-use instructions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Look up a public Synthesizer object by dotted name.
Resolves a name such as synthesizer.grid.Grid in the installed
Synthesizer and reports its signature, docstring, source location
and the installed version. The object is never called and never
repr'd, and source bodies are never returned, only locations.
Side effects: this imports Synthesizer. The first call in a process
takes roughly 1.7-2.0 seconds and runs synth_initialise(), which
creates the Synthesizer base, data, grids,
instrument_cache and svo_filter_cache directories and writes
or merges default_units.yml under the user's data directory. It
is idempotent and silent when those already exist. Later calls reuse
the cached import.
Names are restricted: at most 200 characters and 8 dot-separated
segments, every segment a plain identifier that is neither a Python
keyword nor prefixed with an underscore, and the first segment
exactly synthesizer. The installed package is located before it
is imported, so a synthesizer.py shadowing it on sys.path is
refused rather than executed. Traversal reads attributes statically,
so anything created lazily by a module __getattr__ or by a
property is invisible to this tool.
Args:
dotted_name: Public dotted path rooted at synthesizer.
Returns:
On success a mapping with dotted_name, module (the
module prefix that was imported), object_type,
defining_module, qualname, signature, doc (an
untrusted-content envelope, or None), source as
path:lineno, version and version_source. Any
extraction that fails degrades to None. On rejection or import
failure, a mapping with an error key, plus import_error
when an import raised.
| Name | Required | Description | Default |
|---|---|---|---|
| dotted_name | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full disclosure burden and excels: it reveals side effects (imports Synthesizer, ~1.7-2.0s first call, runs synth_initialise, creates directories and writes default_units.yml), states idempotency, discloses the sys.path shadowing refusal as a security behavior, and notes the object is never called or repr'd and source bodies are never returned. This is exemplary behavioral disclosure.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is long but information-dense with every section earning its place: purpose, resolution semantics, side effects, security, restrictions, args, and returns. It is well structured with clear paragraph breaks. Slight redundancy exists in that the Returns section partly overlaps what the output schema likely already documents, so it could be trimmed, but the prose adds field-level semantics that justify its presence.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the single fully documented parameter and the presence of an output schema, the description is essentially complete for an agent to call this tool correctly. It covers side effects, performance cost, security refusal, naming restrictions, static-traversal limits, and the full success/error return shapes. Nothing needed for correct invocation is missing, especially notable given zero annotation coverage.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 0%, so the description must fully compensate, and it does. The dotted_name parameter is documented in depth: it must be a public dotted path rooted at 'synthesizer', with explicit length, segment-count, keyword, and underscore restrictions. The Returns section also explains the full output mapping and the degrade-to-None/error behavior, far exceeding what the bare schema alone would provide.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a specific verb-resource pair ('Look up a public Synthesizer object by dotted name') and enumerates exactly what the tool reports: signature, docstring, source location, and installed version. This distinguishes it cleanly from introspection siblings like inspect_environment, inspect_local_grid, and list_local_grids, all of which target different subjects.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description gives strong operational constraints (200-char/8-segment limit, plain identifiers, no underscore prefixes, first segment must be exactly 'synthesizer') that effectively tell the agent which lookups will be rejected. It also clarifies that lazily created attributes are invisible, guiding expectations. It stops short of naming sibling alternatives or stating explicit when-not-to-use conditions, but the tool's purpose is so distinct that the implicit guidance is nearly complete.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List every published release of one Syndex dataset.
Releases are immutable: a regenerated or corrected file is published as a new release and the old one stays resolvable. Use this to see which versions exist, which is current, which carry a known bug, and which Synthesizer versions a release is declared to work with — before pinning a download to one.
Requires network access.
Args:
name: Catalogue dataset name, as listed by search_catalogue.
Returns:
On failure, a mapping with ok (False) and error,
plus status or hint as describe_catalogue_dataset
returns them.
On success, a mapping with ``ok`` (``True``), ``source``,
``content_is_untrusted`` (always ``True``), ``dataset``,
``data_type``, and ``releases``: newest first, each with
``release_id``, ``published_at``, ``is_current``,
``known_bug`` (with ``known_bug_description`` when set), the
Synthesizer version bounds, ``file`` details including
``sha256``, and a ``download_command`` pinned to that release.
| Name | Required | Description | Default |
|---|---|---|---|
| name | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full behavioral burden and does so thoroughly. It discloses network requirements, release immutability, failure return structure with status/hint, success return fields, and the noteworthy fact that content_is_untrusted is always True. This is far beyond a minimal description.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with the core purpose, then provides relevant context, usage guidance, network requirement, and a structured Args/Returns section. Every part conveys necessary information without repetition or filler. The structure is easy to scan and well organized.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool has one parameterable, an output schema, and moderate complexity, the description is complete. It explains the return shape on both success and failure, including ordering ('newest first'), important flags, and the presence of a pinned download_command. Nothing necessary for correct invocation or output interpretation is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema only says 'name' is a string, but the description adds that it is a 'Catalogue dataset name, as listed by search_catalogue.' This gives the agent a concrete source for valid valuesholistic, compensating for the 0% schema coverage. It could be more detailed about format constraints, but for a single string parameter this is adequate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The opening sentence names a specific verb ('List'), a precise resource ('every published release of one Syndex dataset'), and clearly distinguishes the tool from siblings like search_catalogue and describe_catalogue_dataset. It is immediately obvious what the tool does and what domain it belongs to.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly states when to use the tool: 'Use this to see which versions exist, which is current, which carry a known bug, and which Synthesizer versions a release is declared to work with — before pinning a download to one.' It also notes that the dataset name comes 'as listed by search_catalogue.' It does not explicitly mention when not to use it, but the context is clear enough.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List the Synthesizer grid files available on this machine.
Reports the resolved grid directory and the grid files inside it without opening any of them, so an agent can pick a grid before paying to inspect it. Importing Synthesizer takes a couple of seconds and creates its data directories as a side effect.
Returns:
On success, a mapping with ok (True), grid_dir,
exists (whether that directory is present),
synthesizer_version, grids (a list of {name, filename, size_bytes, modified} entries, sorted by filename
and excluding symbolic links), and truncated (True when
more grids exist than were reported). On failure, a mapping
with ok (False) and error, plus missing and
hint naming the package to install when Synthesizer is not
importable, or grid_dir when the directory could not be
listed.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full transparency burden and does so thoroughly. It discloses that importing Synthesizer takes seconds and creates data directories as a side effect, that files are not opened, and that results exclude symlinks, are sorted, and may be truncated. These details go well beyond a basic list operation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with a clear one-sentence purpose, then provides exactly the behavioral details that matter: no opening, import side effects, and result shape. The Returns section is structured and informative without being padded, and every sentence earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description is complete for a no-parameter list tool. It covers success and failure modes, package-missing hints, directory-missing behavior, truncation, symlink exclusion, and sorting. An agent has enough context to invoke the tool and interpret its results without additional documentation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool has zero parameters, so there is no schema burden for the description to compensate for. Per the baseline for 0-parameter tools, this is adequately handled; the description correctly frames the operation as an unconditional listing with no required inputs.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb and resource: 'List the Synthesizer grid files available on this machine.' It also distinguishes itself from the sibling inspect-like tools by explicitly saying it reports files 'without opening any of them', making its scope clear for an agent choosing a grid.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies when to use the tool: 'so an agent can pick a grid before paying to inspect it.' This clearly situates the tool as a precursor to inspection tools like inspect_local_grid, though it does not explicitly name the alternative or state when not to use it.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Plot a grid's specific ionising luminosity over its axes.
This is the standard diagnostic for whether a grid covers the ionising output a study needs.
Args:
grid_name: Grid name, with or without a file suffix.
ion: Ion to plot, one of HI or HeII.
Returns:
A mapping with ok, the written path and the ion
plotted, or ok false with an error.
| Name | Required | Description | Default |
|---|---|---|---|
| ion | No | HI | |
| grid_name | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden. It discloses that the tool writes a file (via 'written path') and returns a mapping with ok/error, so the main side effect and possible failure are clear. It does not explicitly state whether it modifies the grid, but the nature of plotting implies non-destructive behavior, which is acceptable.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is compact and front-loaded with the core action, then adds the diagnostic context, followed by parameter and return details. Every sentence serves a purpose; there is no fluff or repetition. It is well-structured and easy to scan.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description covers the purpose, parameters, return format, and error state. An output schema exists, but the description still explains the return mapping clearly, making the tool fully callable without further investigation. For a two-parameter plotting tool, nothing essential is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, so the description must compensate. It does so well: it explains grid_name as a grid name with or without a file suffix, and ion as either HI or HeII, effectively documenting both parameters with meaningful contexts beyond the raw type. No information is left to guesswork.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific action (plot) on a specific resource (grid's specific ionising luminosity) and explains its diagnostic role. It clearly distinguishes from siblings like plot_grid_spectra and plot_grid_lines by naming the exact quantity plotted, so an agent can differentiate without opening schemas.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It says 'This is the standard diagnostic for whether a grid covers the ionising output a study needs,' giving clear guidance on when to use it. It does not explicitly mention alternatives or exclusions, but the context makes it obvious that for spectra or lines you would use the sibling tools, so it is still useful.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Plot the emission lines a local grid holds at one grid point.
Args:
grid_name: Grid name, with or without a file suffix.
point: Axis name to value in that axis's own units, defaulting
to the centre of the grid.
line_ids: Lines to draw, for example ["H 1 4861.32A"],
capped at forty. Defaults to every line in the grid.
Returns:
A mapping with ok, the written path, the grid_point
used and the number of lines drawn, or ok false with an
error.
| Name | Required | Description | Default |
|---|---|---|---|
| point | No | ||
| line_ids | No | ||
| grid_name | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden of behavioral disclosure. It reveals side effects by mentioning a 'written path', describes defaults for point and line_ids, states the forty-line cap, and describes success/error return shapes. This is useful and goes well beyond a bare 'plot' statement.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is compact and well-structured. The first sentence states the core behavior, and each Args/Returns line adds necessary operational detail without unnecessary prose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a three-parameter tool, the description covers parameters, defaults, output contract, and error behavior. An output schema exists, so the return shape need not be fully re-explained. Minor gaps remain around prerequisite conditions, such as whether the grid must already be loaded, but overall the tool is adequately specified.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, so the description fully compensates by explaining all three parameters. It clarifies grid_name suffix handling, point axis/units/default, and line_ids format with an example plus the forty-line cap.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a specific verb and resource: 'Plot the emission lines a local grid holds at one grid point.' This clearly distinguishes the tool from siblings like plot_grid_spectra and plot_grid_ionising_luminosity, which target different visualizations.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The purpose implies when to use the tool: whenever emission lines from a single grid point are needed. However, it does not explicitly state when to prefer this over alternatives, such as plot_grid_spectra, nor does it mention any exclusions or prerequisites.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Plot a local grid's spectra at one grid point.
Renders the spectra a grid holds at a chosen point, so the shape of
the grid's output can be seen without writing a script. Loading
spectra reads the full arrays, so this is slower than
inspect_local_grid, and importing Synthesizer creates its data
directories on first use.
Args:
grid_name: Grid name, with or without a file suffix.
spectra_types: Spectra to draw, defaulting to every spectrum the
grid provides, capped at six.
point: Axis name to value, each in that axis's own units, for
example {"ages": 1e7, "metallicities": 0.01}. Every axis
must be given. Defaults to the centre of the grid.
wavelength_range: [low, high] in the grid's wavelength
units. Defaults to the range carrying essentially all the
luminosity, because a grid's full axis spans many decades
and drawing all of it hides the interesting part.
Returns:
A mapping with ok, the written path, the grid_point
indices used, the spectra_types drawn, the
wavelength_range shown, and the axes values at that
point. On failure ok is False and
error explains why.
| Name | Required | Description | Default |
|---|---|---|---|
| point | No | ||
| grid_name | Yes | ||
| spectra_types | No | ||
| wavelength_range | No |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations present, the description carries the full burden and does so well. It discloses that loading reads full arrays (performance cost), that importing Synthesizer creates data directories on first use, and explains default behaviors such as capping spectra types at six and defaulting the wavelength range to the interesting luminosity range.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with the core purpose, then adds a performance/side-effect note, then systematically documents each parameter and return value. It is longer than average, but every sentence contributes necessary information for correct invocation.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a plotting tool with four parameters and no annotations, the description is complete: it covers all arguments, defaults, units, side effects, and failure behavior. The return mapping is also described even though an output schema exists, giving the agent a clear picture of the tool's contract.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, so the description must fully explain parameters, and it does. Each argument gets meaningful semantics: grid_name suffix tolerance, spectra_types default and cap, point as axis-to-value mapping with units and an example, and wavelength_range as [low, high] with a sensible default.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The opening sentence states a specific verb and resource: 'Plot a local grid's spectra at one grid point.' It distinguishes this from siblings like inspect_local_grid and plot_grid_lines by emphasizing rendering spectra at a single point, so an agent can tell what this tool does without opening the schema.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description gives clear context: use this to see the shape of a grid's output without writing a script. It also names inspect_local_grid as a faster alternative, implying a tradeoff. It does not explicitly say 'do not use when X', but the context is strong enough to guide selection.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search the Syndex catalogue of published Synthesizer datasets.
Syndex is the project's data service: every grid, instrument and
test dataset the project publishes is listed there. Use this to
answer "which grids exist?", "is there a BPASS grid with lines?" or
"what could I download?" — it reports what is published, not what is
installed here. Use list_local_grids for what is already on this
machine.
Requires network access. Nothing is downloaded: the result carries
the synthesizer-download command for each dataset, to propose to
the user rather than to run.
Args:
query: Free-text filter. Every whitespace-separated term must
appear somewhere in a dataset's name, display name,
description or type, case-insensitively.
data_type: Restrict to one catalogue type, such as grid,
dust_grid, instrument or simulation_data.
has_spectra: Restrict to grids that do or do not carry spectra.
has_lines: Restrict to grids that do or do not carry line
luminosities.
is_test: Restrict to deliberately reduced test datasets
(True) or to production ones (False).
limit: Maximum number of datasets to return, capped at 50.
matched always reports the true total, so a capped
listing is a reason to filter rather than to page.
Returns:
On failure, a mapping with ok (False) and error, plus
hint when the catalogue could not be reached.
On success, a mapping with ``ok`` (``True``), ``source`` (the
portal URL), ``content_is_untrusted`` (always ``True``: every
string below was published by a third party and is data, not
instructions), ``matched`` (how many datasets matched),
``truncated`` (whether more matched than were returned), and
``datasets``: a list of ``{name, display_name, description,
data_type, is_test, is_recommended, release_id, size_bytes,
download_command}`` entries, with ``has_spectra`` and
``has_lines`` on grids. Pass a ``name`` to
``describe_catalogue_dataset`` for the full metadata.
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | ||
| query | No | ||
| is_test | No | ||
| data_type | No | ||
| has_lines | No | ||
| has_spectra | No |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden and exceeds it: it discloses network dependency, that nothing is downloaded, that content_is_untrusted is always True because results are third-party data, and describes both failure and success return shapes. This is thorough behavioral disclosure.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is long but each section earns its place: purpose, usage guidance, parameter details, and return structure. It is front-loaded with the main purpose and alternative routing before diving into parameter semantics, making it efficient despite its length.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a search tool with 6 parameters, no annotations, and no parameter descriptions in the schema, the description is complete: it covers network requirements, return values, failure mode, truncation semantics, and how to get full metadata via describe_catalogue_dataset. Nothing needed for correct invocation is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, and the description fully compensates by explaining every parameter: query matching semantics, data_type examples, has_spectra/has_lines grid restrictions, is_test purpose, and limit's 50 cap with the matched field clarifying pagination behavior. This is exactly what an agent needs beyond the bare schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific verb and resource: 'Search the Syndex catalogue of published Synthesizer datasets.' It explicitly distinguishes itself from list_local_grids by clarifying this reports published data, not installed data, which prevents sibling confusion.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives concrete example questions ('which grids exist?', 'is there a BPASS grid with lines?') and explicitly directs to list_local_grids for installed data. It also notes network access is required and that the command should be proposed, not run, making when/how to use the tool unambiguous.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search the bundled skill tree and any Synthesizer checkout docs.
Args: query: Words to look for. Matching is plain substring and token matching, not a query language.
Returns:
A mapping with query, corpora_searched (a list of
{corpus, root} recording what was actually covered),
corpora_unavailable explaining, in a sentence each, any
corpus that was not,
files_scanned, hit_count, and hits. Each hit is
{corpus, path, line, score, snippet} where snippet is an
untrusted-content envelope. truncated records whether a cap
bit, notes explains an unsearchable query, and error is
present only when the search itself failed.
The bundled skill tree is always searched; a Synthesizer source
checkout is searched in addition to it when one is found, never
instead of it. Synthesizer's own documentation is not installed
by pip, so it is unavailable unless such a checkout exists.
| Name | Required | Description | Default |
|---|---|---|---|
| query | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden and fully discloses behavior: plain substring/token matching, the exact result mapping, corpora coverage/unavailability, the untrusted-content envelope for snippets, truncation, notes, and error-only-on-failure. This is unusually transparent.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with a one-sentence purpose and then structured into Args and Returns. It is dense and mostly efficient, but the lengthy return listing and a minor typo ('cap bit') keep it from a top score.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a one-parameter search tool, the description covers input semantics, output shape, hit entry fields, truncation, notes, errors, and corpus availability. Nothing an agent needs to call it correctly is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, but the description compensates fully by defining query as words to look for and specifying plain substring/token matching, not a query language. This adds exactly the semantic meaning an agent needs beyond the bare string type.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The opening sentence uses a specific verb ('Search') and names the two corpora ('bundled skill tree', 'Synthesizer checkout docs'), clearly distinguishing this from example-finding or grid-inspection siblings. It does not merely restate the tool name.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description clearly explains the tool's scope and availability: the bundled skill tree is always searched, while Synthesizer docs are only searched when a checkout exists and never instead of the bundled tree. It does not explicitly name an alternative tool, but the usage context is clear.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Statically validate a Synthesizer script.
The script is parsed, never run. Checks run least invasive first: syntax, import availability by specification lookup only, referenced Synthesizer attributes and keyword arguments, and locally available grids.
Args: source: Python source text.
Returns:
A mapping with ok (no error-level diagnostics),
script_was_run (always False), a summary sentence,
diagnostics (a list of {severity, code, line, message}
capped at 100) with diagnostics_truncated recording whether
that cap bit, imports as {found, missing} lists of
top-level module names, notes explaining any check that was
skipped, and suggested_commands the host may choose to run
itself. grids with {requested, local} grid names is
present only when the script builds a Grid and the local grid
listing is available, and error is present only when
validation itself failed.
| Name | Required | Description | Default |
|---|---|---|---|
| source | Yes |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description fully discloses behavior: the script is parsed, never run, checks run least invasive first, and the exact return structure is specified including the constant 'script_was_run' of False, conditions for optional keys like 'grids' and 'error', and limits like diagnostics capped at 100. This is extensive transparency beyond any structured metadata.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is detailed but well-structured, starting with the core purpose, then behavior and returns. The 'Args' and 'Returns' sections are clearly segmented and each sentence conveys necessary information about the complex return payload. It is not excessively verbose for the complexity, though it could be tightened slightly, earning a 4.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a static validation tool with one parameter and a rich output schema, the description covers all essential aspects: what it checks, what it never does, the order of checks, the complete return mapping with conditional fields, and even notes on skipped checks. An agent has everything needed to invoke it correctly and interpret results, making it fully complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, so the description must carry the parameter meaning. It provides 'source: Python source text' which exactly defines the single required parameter. While minimal, it is sufficient and unambiguous. Given the low schema coverage, this adequately compensates, though it could add more detail (e.g., encoding), so a 4 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states 'Statically validate a Synthesizer script' with a specific verb and resource, and clarifies it is 'never run.' It lists the checks performed (syntax, import availability, references, grids), distinguishing it from sibling tools that plot, inspect environments, or search documentation. This is a clear, non-tautological purpose that an agent can act on.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explains what the tool does and that it is static (never runs), but does not explicitly state when to use it instead of alternatives. It does not mention sibling tools or conditions under which validation is preferred over running the script. The behavior implies use for pre-run checks, but no direct guidance on alternatives is given, so it earns a 3.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.2.0describe_catalogue_datasetlist_catalogue_releasessearch_cataloguev0.1.0find_exampleinspect_environmentinspect_local_gridinspect_synthesizer_apilist_local_gridsplot_grid_ionising_luminosityplot_grid_linesplot_grid_spectrasearch_documentationvalidate_scriptScored across 13 tools
Most tools are clearly distinct: search vs list vs inspect vs plot families each target different resources and actions. Minor overlap exists between search_documentation and find_example for how-to questions, and the three plot_grid_* tools share a prefix but are differentiated by what they render.
Tool names follow a mostly consistent verb_noun snake_case pattern (search_, list_, inspect_, plot_, describe_, validate_). The main inconsistency is using 'find' vs 'search' for similar lookup actions and 'describe' vs 'inspect' for catalogue vs local metadata, but the pattern remains readable and predictable.
13 tools is well-scoped for a server covering documentation search, examples, catalogue browsing, local grid inspection/plotting, environment/API inspection, and script validation. Each tool has a clear role and none feel redundant or like filler.
The tool surface covers the main workflows: finding docs/examples, discovering and describing catalogue datasets, listing and inspecting local grids, plotting grid outputs, inspecting the environment/API, and validating scripts. Minor gaps exist, such as no tool to return raw spectral/line data as structured values rather than plots, but these are workable limitations rather than dead ends.
Your org's AI agents, tasks, runs, search, and brain files as MCP tools and resources.
Read-only MCP tools for AI agent discovery, structured resources, and NIULAI information.
Discover MCP servers, A2A agents, and shared agent knowledge through a read-only MCP gateway.
Agent-native catalogue of Baseframe Labs dev tools and MCP servers.
