Skip to main content
Glama
theosis-ancient-context-mcp

theosis-ancient-context-mcp

by PrattyT85

theosis-ancient-context-mcp

FastMCP stdio service for ancient text corpora. Part of the Theosis ancient-context integration.

Corpora

Corpus

Language(s)

Status

Licence

TLA (Thesaurus Linguae Aegyptiae)

Egyptian

Remote/status-only

Free non-commercial research

Coptic SCRIPTORIUM

Coptic

Local optional

CC-BY (exceptions exist)

HPM/HDivT (Hittite ritual texts)

Hittite

Remote/status-only

Academic

TLHdig (Hittite XML editions)

Hittite

Local optional

CC BY 4.0 (verify files)

CUC (Copenhagen Ugaritic Corpus)

Ugaritic

Local optional

CC BY-NC 4.0

DASI

Arabic epigraphic

Remote/status-only

Academic

OCIANA

Akkadian/Sumerian

Remote/status-only

Academic

CDLI

Akkadian/Sumerian/Elamite

Adapter ready (remote)

Open access

DPPC

Phoenician/Punic

Deferred

Not published

CIP

Punic

Deferred

Not published

Related MCP server: receipts

What's integrated vs. what's not

  • DPPC and CIP are NOT integrated — no public API, data, or licence has been confirmed. They are registered as deferred with no scraper or adapter.

  • TLHdig is local optional — set HITTITE_TLHDIG_DIR to the extracted TLHbasisONLINE25.1_ZENODO directory from Zenodo record 15459134. The adapter reads bounded XML and labels the dataset/version.

  • CUC is CC BY-NC and local-only — commercial use is prohibited. The adapter only works when CUC_CORPUS_DIR is configured to point at a local checkout of the corpus.

  • TLA, HPM, DASI, OCIANA are remote/status-only until a stable public API contract is verified. They return structured remote_only results with provenance envelopes.

  • CDLI has a read-only HTTP adapter targeting the verified cdli.earth REST JSON API. Endpoints: /search/?q=...&format=json, /artifacts/<id>.json, /inscriptions/<id>.json. Strict query/id validation, 15s timeout, 5 MiB response cap. Open access — verify licence for your use case. Do not claim full text when the API returns metadata only.

  • Coptic SCRIPTORIUM requires a local corpus directory. The adapter supports meta.json-backed metadata and bounded text-file search.

Source-layer routing guidance

When choosing a corpus for research, use the following routing by language/domain:

Domain / Language

Primary Source

Notes

Egyptian (hieroglyphic, hieratic, Demotic)

TLA

Thesaurus Linguae Aegyptiae — remote-only until API contract confirmed

Coptic

Coptic SCRIPTORIUM

Local optional — requires COPTSCRIPTORIUM_CORPUS_DIR

Hittite (ritual texts)

HPM/HDivT

Remote-only; TLHdig XML dataset available separately

Hittite (XML editions)

TLHdig

Local optional — requires HITTITE_TLHDIG_DIR (Zenodo DOI: 10.5281/zenodo.15459134)

Ugaritic

CUC

CC BY-NC 4.0 — commercial use prohibited. Local optional

Akkadian / Sumerian / Elamite (cuneiform)

CDLI

Open access adapter ready — REST JSON API

Arabic epigraphic

DASI

Remote-only — no stable API confirmed

Ancient North Arabian / Safaitic / Dadanitic

OCIANA

Remote-only — no stable API confirmed

Phoenician / Punic

DPPC, CIP

Deferred — no public data or API; not integrated

Key warnings

  • ORACC is NOT a Ugaritic corpus. ORACC (Open Richly Annotated Cuneiform Corpus) publishes Sumerian, Akkadian, and other cuneiform texts. It is not a source for Ugaritic primary texts. Use CUC for Ugaritic.

  • CUC is CC BY-NC. Commercial use of the Copenhagen Ugaritic Corpus is prohibited. Respect the licence when using KTU text data.

  • Primary texts vs. translations vs. scholarly synthesis: This service targets primary-source corpora and their machine-readable editions. Translations, commentaries, and scholarly interpretation may be available through upstream platforms but are not bundled or re-hosted here. Check each source's licence notes for redistribution constraints.

Tools

  1. list_corpora — all registered corpus records with metadata

  2. get_corpus_status(corpus) — detailed status, local path, availability

  3. get_source_manifest() — machine-readable source manifest with version/DOI/review dates

  4. search_corpus(corpus, query, limit) — search with explicit status results

  5. get_text(corpus, reference) — text retrieval with provenance envelope

  6. get_text_metadata(corpus, reference) — metadata retrieval with provenance

Setup

# Install
uv sync --extra dev

# Run tests
uv run pytest -v

# Compile check
uv run python -m compileall src/theosis_ancient_context -q

# Health check (offline, CI-safe — no network requests)
uv run python scripts/healthcheck.py

# Health check (live — tests CDLI adapter, bounded read-only HTTP)
uv run python scripts/healthcheck.py --live

# Health check (text output)
uv run python scripts/healthcheck.py --text

# Run the server (stdio)
uv run python -m theosis_ancient_context

Optional local corpora

# Coptic SCRIPTORIUM
export COPTSCRIPTORIUM_CORPUS_DIR=/path/to/corpora

# Copenhagen Ugaritic Corpus
export CUC_CORPUS_DIR=/path/to/cuc

# Hittite TLHdig (extracted Zenodo archive)
export HITTITE_TLHDIG_DIR=/path/to/TLHbasisONLINE25.1_ZENODO

Integration

All results include a provenance envelope with:

  • source_layer — integration type (local_optional, remote_only, deferred, etc.)

  • corpus_id — canonical corpus identifier

  • source_url — upstream URL

  • licence — licence summary

  • licence_warning — detailed licence notes and restrictions

  • access_status — current integration status

Safety

  • Path traversal prevention for all local file access

  • Bounded search results and text retrieval

  • No persistent cache

  • No secrets required

  • CDLI adapter makes bounded read-only HTTP requests (stdlib urllib, 15s timeout, 5 MiB cap)

Available Tools

5 tools
get_corpus_statusGet Corpus StatusA

Get detailed status for a specific corpus.

Includes configured local path, availability checks, and version/commit info when discoverable.

Inputs: corpus (required). Returns: object

get_textGet TextB

Retrieve text content by reference identifier.

For local adapters, returns bounded text with provenance. For deferred/remote-only sources, returns explicit status.

Inputs: corpus (required), reference (required). Returns: object

get_text_metadataGet Text MetadataA

Retrieve metadata for a text reference.

For local adapters, returns file metadata and provenance. For deferred/remote-only sources, returns explicit status.

Inputs: corpus (required), reference (required). Returns: object

list_corporaList CorporaA

List all registered ancient text corpora with metadata.

Returns registry records for TLA, Coptic SCRIPTORIUM, HPM/HDivT, CUC, DASI, OCIANA, CDLI, DPPC, CIP — each with source_type, language, period, access_status, source_url, licence notes, and integration notes.

Inputs: no named parameters. Returns: result

search_corpusSearch CorpusA

Search a corpus for a query string.

For enabled/local adapters, returns matching results. For deferred or remote-only sources, returns an explicit not_ready/remote_only result with provenance envelope rather than failing ambiguously.

Inputs: query (required), corpus (required), limit. Returns: object

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4/5.0

    Scored across 5 tools

    Disambiguation5/5

    Each tool targets a distinct resource and action: listing all corpora, checking one corpus's detailed status, searching within a corpus, retrieving text content, and fetching text metadata. There is no meaningful functional overlap, and even the status-related tools are cleanly separated by all-versus-one scope.

    Naming Consistency5/5

    All tools use a consistent verb_noun snake_case pattern: list_corpora, get_corpus_status, search_corpus, get_text, get_text_metadata. Plural and singular nouns are used naturally for collection-level versus single-item operations, so the naming is predictable.

    Tool Count5/5

    Five tools is a well-scoped size for a read-only ancient-text corpus access server. Each tool earns its place in the workflow without redundancy or unnecessary surface area.

    Completeness5/5

    The tool surface covers the apparent domain completely: discover available corpora, inspect corpus status, search for texts, retrieve text content, and retrieve metadata. Deferred or remote-only sources are handled explicitly rather than resulting in dead-end errors, and no CRUD operations are expected for this read-only context.

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      B
      maintenance
      Exposes forensic case and evidence management operations over stdio, including case creation, evidence registration with SHA-256 hashing, and integrity verification, so clients can manage DFIR investigations programmatically.
      7
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables any agent to query the same interview corpus as the UI over stdio, exposing tools to list transcripts, search quoted lines, retrieve verified themes with citations and dropped reasons, and fetch source lines.
      9 npm
      MIT