RocqSolid
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ROCQ_LSP_BINARY | No | The `coq-lsp` executable (default `coq-lsp`) | |
| ROCQ_LSP_TIMEOUT | No | Deadline per call; set slightly below the MCP tool timeout (default `55`, max `3600`) | |
| ROCQ_LSP_USE_OPAM | No | Run via `opam exec --`; `0` runs the binary directly (default `1`) | |
| ROCQ_LSP_WORKSPACE | No | Override the inferred workspace root (default unset) | |
| ROCQ_LSP_DUNE_BUILD | No | Build dune dependencies before checks; `0` disables (default `1`) | |
| ROCQ_LSP_MAX_ERRORS | No | Max errors `coq-lsp` collects (default `1000000`) | |
| ROCQ_LSP_DUNE_TIMEOUT | No | Limit per dune detection or build command (default `55`, max `3600`) | |
| ROCQ_LSP_MESSAGE_LIMIT | No | Max errors returned per check (default `10`, max `50`) | |
| ROCQ_LSP_COMMAND_TIMEOUT | No | Default per-sentence timeout; `0` disables (default `5`) | |
| ROCQ_LSP_IDLE_TTL_SECONDS | No | Idle time before cached sessions are stopped (default `1800`) | |
| ROCQ_LSP_HARD_TIMEOUT_GRACE | No | Extra wait after cancellation before the session is killed (default `5`) | |
| ROCQ_LSP_TIMEOUT_AFTER_ERROR | No | Overall deadline for the rest of a check after its first error; `0` disables (default `10`) |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": true
} |
| resources | {
"subscribe": false,
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| rocq_check_fileA | Check a Rocq file and report diagnostics. Use this to check files for errors. (To find warnings, recompile the file manually.) Error diagnostics include the source sentence and the proof state at error positions. The first error in a file is always reported, along with later errors reached within the after-error deadline. The list of errors may be truncated. The tool also reports Rocq sentences running for long time. This tool is efficient due to per-file caching. For dune-based Rocq projects,
it attempts to rebuild dependencies before checking and restarts the cached
coq-lsp process when that applies. Use Timeouts are handled gracefully, and partial diagnostics are returned when possible. Repeated checks after timeouts use the cache. |
| rocq_run_atA | Run Rocq commands at a file position and return their output. Use this for running Rocq commands, such as:
Prefer calling this tool right after imports or between lemmas/definitions. Only use within proofs when necessary. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 2 tools
The two tools have distinct primary purposes: file-level diagnostics versus running Rocq commands at a specific position. There is minor overlap around proof-state inspection, but the descriptions make the intended use clear.
Both names use the same `rocq_` prefix and snake_case, which is predictable. The action phrasing is slightly inconsistent (`check_file` versus `run_at`), but still readable and coherent.
Two tools is thin for a Rocq interaction server, even though both are broad and useful. The surface relies heavily on `rocq_run_at` for many query and debugging tasks.
The pair covers file checking and arbitrary Rocq command execution, which enables most core workflows. Minor gaps remain, such as warning retrieval requiring manual recompilation and no dedicated proof-state/inspection tool beyond using `rocq_run_at`.