Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
ECLIPSE_MUSTER_STATENoOverride for the location of the state file. State lives in data/state.json by default.data/state.json
ECLIPSE_MUSTER_POLICYNoOverride for the location of the policy file. Policy lives in policy.json by default.policy.json

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
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
record_ingestB

Ingest a canonical Customer Graph record (person / company / deal / ticket / conversation / activity). Every field must carry provenance; a Data Rights Ledger block is required. Customer-authored conversation text is tagged untrusted and instruction-shaped content is quarantined at ingest.

record_readA

Read one graph record with egress masking applied (surface: owner = full email, analytic = domain-only email; phone always partial; gov/payment/health fields null). Receipted.

record_searchB

Search graph records by kind + substring; returns masked summaries. The query itself is receipted by hash only.

merge_proposeA

PROPOSE an identity merge between two same-kind records. Returns evidence (matching keys by hash, conflicts) + a proposal hash. Executes nothing (Beacon propose-only doctrine).

merge_executeA

Execute a proposed graph merge. Requires the owner-confirm token bound to the proposal hash (confirm-merge-<proposal_id>-). Graph nodes only — never merges records inside a platform. The pre-merge snapshot is stored on the merge record for unmerge.

merge_unmergeA

Undo a confirmed merge from its receipt snapshot, restoring the pre-merge nodes exactly. Requires confirm-unmerge-<merge_id>.

reply_draftB

Draft a ticket reply ONLY from record facts, with a citation list. Untraceable facts become MISSING_FACT markers. Customer text is never a fact source; quarantined (instruction-shaped) messages are reported, never obeyed.

policy_showA

Show the tier floors, effective tiers, kill switches, and masking rules in force.

receipt_listB

List the append-only, hash-chained receipts and verify the chain. Receipts carry IDs + hashes only — never content.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

TDQS

A3.7/5.0

Scored across 9 tools

Disambiguation5/5

Each tool targets a distinct resource and action: record operations, merge lifecycle stages, reply drafting, policy inspection, and receipt listing. The staged merge tools (propose/execute/unmerge) are explicitly separated by required tokens and side effects, preventing confusion between them.

Naming Consistency5/5

All nine tools follow the same noun_verb resource_action convention: record_ingest, merge_propose, reply_draft, policy_show, receipt_list, etc. There is no mixing of camelCase, vague verbs, or inconsistent ordering.

Tool Count5/5

Nine tools is a well-scoped set for a governed customer graph service with ingestion, search, merge lifecycle, reply drafting, policy visibility, and receipts. Each tool has a clear role and none appear gratuitous.

Completeness3/5

The merge lifecycle is complete and governance operations are covered, but there is no explicit record update, delete, or erasure operation despite the data-rights framing. Agents can work around some gaps via re-ingest or merges, but lifecycle coverage is notably incomplete for a customer record graph.

Maintenance

ActivityMaintained
ResponsivenessNo issues