covera-mcp
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| COVER_ACCESS_KEY | No | The COVER Open API Access Key. The build flow reads it from the Local Agent's cached credentials or this environment variable; the recovery flow reads it from this environment variable (or the cached key when the requested server matches the cached one). | |
| CODESCROLL_TM_SERVER_URL | No | URL of the Local Agent's remote service. | |
| CODESCROLL_TM_INSTALL_PATH | No | Path to the installed COVER EE Local Agent; the MCP bundle finds the Local Agent through this environment variable, which the Local Agent installation sets. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
COVERA drives Suresoft COVER EE on the local machine: coverage builds, module coverage lookups, and C/C++ toolchain-error recovery.
Coverage build: read covera_get_guide topic=build first. Plan with covera_coverage_build (no execute); explain what execute does -- it changes the project build setup and registers the module on the shared COVER server -- and get the user's approval before calling with execute. A WAITING_* status means resume with its jobContext, not a fresh start.
Measurement is read-only; give exactly one of a module name or a 24-hex id.
Recovery: covera_recovery_publish changes the shared COVER server; confirm first. Do not modify original sources.
The Open API Access Key is read from the Local Agent's cached credentials or COVER_ACCESS_KEY. Never ask for, print, or store it.
On setup errors run covera_doctor.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| covera_doctorA | Report whether this MCP server can drive COVER: the Python it runs on, whether CODESCROLL_TM_INSTALL_PATH points at a COVER EE Local Agent, whether cached Open API credentials exist, whether COVER_ACCESS_KEY is set, and which skill entry scripts are present. Needs no project. Run this first when other tools fail with setup errors. |
| covera_get_guideA | Return the procedure a COVERA skill follows, so a client without the skills can work the same way. Topics cover the coverage build flow, module measurement, and toolchain recovery, plus their reference documents. Read the matching topic before driving a flow. |
| covera_coverage_measureA | Read a module's coverage results from the COVER server: the module summary, and optionally per-file and per-function detail. Read-only; runs nothing and changes nothing. Give exactly one of a module name (a filename or path suffix to search for) or a 24-hex module id. The server URL and Access Key come from the Local Agent's cached Open API credentials; set CODESCROLL_TM_INSTALL_PATH if they are not found (see covera_doctor). |
| covera_coverage_buildA | Drive a COVER EE coverage build for a C/C++, C#, or Java project: identify the original build and real tools, plan the instrumented build keeping the original output paths, and -- when execute is true -- run it and keep the successful setup for later builds. Give projectDir to start a new build or jobContext to resume one (exactly one). execute runs the real builds and registers the coverage module on the COVER server, so it changes the project's build setup and server state: read covera_get_guide topic=build first, explain what it does, and get the user's approval before calling with execute. With execute omitted it only plans and writes job state. The result is a workflow envelope; a WAITING_* status means resume with its jobContext after supplying what it asks for. The Open API Access Key is read from the Local Agent's cached credentials or COVER_ACCESS_KEY; this server never prompts for it. |
| covera_recovery_discoverB | Read the build environment, COVER server settings, and available analysis-error material without changing anything, and report what is missing to collect it. Read-only. |
| covera_recovery_startA | Copy the analysis-error ZIP/folder (or collect the latest from the server) into a new run directory and record initial evidence and state. Does not modify original sources. May download from the COVER server, which needs COVER_ACCESS_KEY or cached credentials. |
| covera_recovery_collectA | Download the latest PA analysis-error material over the Open API into a new ZIP and start the case. Needs COVER_ACCESS_KEY or cached credentials. |
| covera_recovery_verifyA | Run a server-side PA verification of an evidence-backed candidate for a case. Runs analysis on the COVER server but makes no permanent server change. Needs COVER_ACCESS_KEY or cached credentials. |
| covera_recovery_inspectB | Read the recorded run history of a case. Local and read-only. |
| covera_recovery_publish_planA | Produce a server change plan for a verified case; reads the COVER server only. Returns a plan_digest to pass to covera_recovery_publish. Needs COVER_ACCESS_KEY or cached credentials. |
| covera_recovery_publishA | Apply a verified plan to the COVER server and connect the wrapper. This changes the shared COVER server, so confirm with the user first. Pass the plan_digest from covera_recovery_publish_plan. Needs COVER_ACCESS_KEY or cached credentials. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| coverage-build | Plan and run a COVER EE coverage build for a project, following COVERA's build procedure. |
| coverage-measure | Read a module's coverage results from the COVER server. |
| toolchain-recovery | Recover C/C++ files excluded by a COVER analysis error, following COVERA's recovery procedure. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 11 tools
Most tools map to distinct steps, but covera_recovery_start and covera_recovery_collect overlap heavily — start explicitly says it can 'collect the latest from the server' and record state, which is what collect does, so an agent may not know which to call. The three read-only diagnostics (covera_doctor, covera_recovery_discover, covera_recovery_inspect) also sit close together, distinguished mainly by prose rather than by name.
All tools share the `covera_` prefix and snake_case, with recognizable subgroups (`covera_recovery_*`, `covera_coverage_*`). Minor deviations exist: covera_doctor is a bare noun with no verb, and covera_get_guide breaks the sub-domain grouping used elsewhere.
11 tools is well within a comfortable range for a server spanning two workflows (coverage build/measure and PA error recovery). Each tool corresponds to a recognizable step, and none feel like filler, aside from the mild redundancy between start and collect.
The recovery lifecycle is nearly end-to-end (discover → start/collect → inspect → verify → plan → publish) and the build side covers guide → build → measure. Gaps are minor but real: no way to abort/stop a running build, no rollback or un-publish after covera_recovery_publish, and no tool to enumerate cases or runs beyond inspecting a single case.