Skip to main content
Glama
ahines99

portco-mcp

by ahines99
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Hybrid

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
PORTCO_HTTP_TOKENSNoBearer tokens for HTTP transport, each carrying a role and an explicit tenant scope (token=principal:role:company1|company2, '*' for all). The HTTP app refuses to start without it; not needed for the stdio server.
PORTCO_LLM_ENABLEDNoEnables the optional LLM mapping judge. Off by default; it can only reorder deterministic candidates or abstain.false

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

CapabilityDetails
tools
{
  "listChanged": false
}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
healthcheckA

Return service health for diagnostics.

start_onboarding_runA

Start read-only onboarding of a registered source (e.g. fixture:portco_a).

    Runs connection validation, profiling, entity and join inference and canonical mapping,
    then stops at the first human review gate. Returns the run status and pending review items.
get_run_statusC

Current status, step history, gate and pending review items of a run.

resume_runA

Continue a paused or failed run. Gates re-check approvals; completed steps are reused, not recomputed.

list_pending_reviewsC

Review items a human must decide before the run can continue.

submit_mapping_reviewA

Record a human reviewer's decisions for the run's current gate. Reviewer principals only.

    Decisions: approve | reject | approve_with_override (override keys: canonical_field, transform,
    pii_handling). The approval is bound to the content hash of the items under review.
certify_runA

Certify the generated bundle and its metrics. Reviewer principals only.

    `subject_hash` is the `subject_hash` of the pending certification items: it binds the approval to the
    exact packet under review (bundle, tests, waivers, open findings). `metric_decisions` maps
    metric -> approve|reject; unknown metric names are rejected, unlisted metrics are approved. Rejected
    metrics (and metrics derived from them) are not published.
profile_schemaA

Profile a source read-only: row counts, null rates, distinct counts, patterns, PII classes.

    Aggregates only (`sample_rows=0` by construction); values never leave the source adapter.
    Starts a run that pauses after profiling; continue it with propose_canonical_mapping.
propose_canonical_mappingA

Infer entities and joins and propose canonical mappings for a profiled run.

Deterministic scoring; anything uncertain, PII-bearing, metric-bearing or conflicting is routed to human review rather than guessed. Stops at the mapping-review gate.

generate_dbt_artifactsA

Generate the dbt project and semantic layer from the reviewed mapping.

    Requires `approval_id` from submit_mapping_review; the server verifies it exists, is not revoked
    or expired, belongs to this run, and is bound to the current mapping content (no boolean flags).
run_sandbox_testsA

Build the generated project with dbt in a disposable sandbox and reconcile metrics exactly.

On success the run pauses at the certification gate; on failure at the test-failures gate.

publish_runA

Publish certified artifacts. Fails closed without a valid certification bound to the bundle.

Prompts

Interactive templates invoked by user choice

NameDescription
review_runReview a workflow run, separating facts, assumptions and recommendations.
explain_mappingExplain one mapping proposal to a human reviewer.
onboarding_kickoffKick off onboarding of a new portfolio-company source using the project Skills.

Resources

Contextual data attached and managed by the client

NameDescription
policiesOperating and safety policies, rendered from the enforced policy registry.
ontologyThe canonical PE ontology: entities, fields, synonyms, relationships and metrics.
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A3.6/5.0

Scored across 12 tools

Disambiguation4/5

Most tools map to distinct pipeline stages (profile, propose, review, generate, test, certify, publish), so an agent can largely tell them apart. However, start_onboarding_run and profile_schema both 'start a run' with profiling, and get_run_status overlaps list_pending_reviews on surfacing pending review items, creating minor boundary ambiguity.

Naming Consistency4/5

Tools follow a clear snake_case verb_noun pattern (start_onboarding_run, get_run_status, resume_run, submit_mapping_review, certify_run, publish_run, etc.). The lone exception is healthcheck, a bare noun that breaks the otherwise consistent convention.

Tool Count5/5

12 tools is well-scoped for a governed data-onboarding pipeline with multiple human gates. Each tool corresponds to a meaningful lifecycle stage or control operation, with no obvious redundancy bloating the surface.

Completeness4/5

The surface covers the full happy-path lifecycle: start, status, resume, profile, propose mapping, submit review, generate artifacts, sandbox test, certify, and publish, plus review listing and healthcheck. Missing abort/cancel and any run-listing or cleanup operations are minor gaps agents can work around.

Maintenance

ActivityNo data
ResponsivenessNo issues

Resources