Skip to content
verify mcp Beta VerifyMCP is currently in beta. If you notice any issues, get in touch and we’ll put it right.

Hatun MCP - Governed Agent Gateway

OCI · GHCR.IO/SZL-HOLDINGS/HATUN-MCP

Governed MCP gateway: policy gate on every tool call, hash-chained DSSE receipt on every result.

−1 this week 63 Trust /100

Last scanned (14 hours ago)

Scores higher than 98% of OCI servers

Category Security

Trust breakdown (7 categories)

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →

Supply Chain Security0
  • No malware scan is available for this kind of package: the supply-chain vendors we use do not cover it. This is a permanent gap in our coverage, not a finding about the package.Unverified
  • Known CVEs could not be checked: this artifact ships no SBOM or dependency manifest, so there is no dependency list to read.Unverified
  • Install-script risk not yet assessed.Unverified
  • Dependency health could not be checked: this artifact ships no SBOM or dependency manifest, so there is no dependency list to read.Unverified
Provenance & Transparency100
  • Source repository is publicly reachable at the declared URL. View diagnostics → Pass
  • Cryptographically verified build provenance (signed, bound to szl-holdings/hatun-mcp). View diagnostics → Pass
  • Clear OSI-approved license (Apache-2.0).Pass
  • Actively maintained (last published 5 days ago).Pass
  • Publishes a security disclosure policy (SECURITY.md).Pass
Schema Quality & AI Usability78
  • 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).Pass
  • AI-judged instruction clarity (good).Pass
  • Tool/resource definitions use about 4094 tokens (~83/item across 49 items; 47 tools + 2 resources), lean.Pass
  • Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management95
  • Stability check failed: the tool surface changed between 1.0.1 and 1.0.4: 2 tool removals, 0 breaking changes, 12 additions. See how to fix → Fail
Tool Coverage67
  • 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
  • 0% of tool parameters carry a description.Fail
Tool Safety100
  • No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
  • We read all 47 captured tool definition(s), and no name or description among them implies an irreversible operation.Pass
  • An AI judge read all 49 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
  • Implements a current MCP spec version (2026-07-28).Pass

Unverified: 1 category

A category scored 0 because we could not verify it: a data source with nothing on this package, evidence we could not reach, or a check we could not run. We only credit what we can confirm.

What we tested

Checks we ran (12)

  • Source repository
  • Provenance
  • License
  • Maintenance
  • Security disclosure
  • Schema quality
  • Stability
  • Tool coverage
  • Injection markers
  • Destructive annotations
  • Judged manipulation
  • MCP protocol

Checks we could not complete (4)

These are gaps on our side, not findings about this package. We only credit what we can confirm, so a category made up entirely of them scores 0.

  • Malware scan in Supply Chain Security (scores 0 for this)No malware scan is available for this kind of package: the supply-chain vendors we use do not cover it. This is a permanent gap in our coverage, not a finding about the package.
  • Known CVEs in Supply Chain Security (scores 0 for this)Known CVEs could not be checked: this artifact ships no SBOM or dependency manifest, so there is no dependency list to read.
  • Install scripts in Supply Chain Security (scores 0 for this)Install-script risk not yet assessed.
  • Dependency health in Supply Chain Security (scores 0 for this)Dependency health could not be checked: this artifact ships no SBOM or dependency manifest, so there is no dependency list to read.

Does not apply to this package (2)

  • Endpoint SecurityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
  • Transport & ReachabilityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.

The score does not measure everything. What the score does not measure →

Changelog

Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.

  • 19 Nov 26 +1

    No change was recorded against any check on this day. Stability & Change Management went from 91 to 95.

  • 18 Nov 26 −1

    No change was recorded against any check on this day. Stability & Change Management went from 95 to 91.

  • 15 Nov 26 −1
    • Stability: pass → fail ▼ security
    • Schema quality: 2968 → 4094 ▼ functional
    • MCP protocol: Implements a current MCP spec version (2026-07-28). functional
    • Package version: 1.0.1 → 1.0.4 functional
    • Package version: 1.0.1 → 1.0.3 functional
    • Package version: 1.0.1 → 1.0.2 functional
  • 12 Nov 26 +1
    • Stability: 0.97 → pass security
  • 10 Nov 26 +1

    No change was recorded against any check on this day. Stability & Change Management went from 90 to 93. That category is still filling its 30-day observation window: 27 days of observed history at the previous scan, 28 at this one. The score rises as the window fills, whether or not the server changes.

  • 8 Nov 26 +1

    No change was recorded against any check on this day. Stability & Change Management went from 83 to 87. That category is still filling its 30-day observation window: 25 days of observed history at the previous scan, 26 at this one. The score rises as the window fills, whether or not the server changes.

  • 6 Nov 26 −3
    • Stability: pass → 0.80 functional
  • 31 Oct 26 +1
    • Stability: 0.97 → pass security
Diagnostics

Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.

Captured 21 Nov 2026 · Analysed oci/ghcr.io/szl-holdings/hatun-mcp@1.0.4

Provenance Verified

A signed build attestation was found and verified, binding this exact artifact to the source repository it claims to come from.

Result Verified
Ecosystem oci
Reason Verified
Discovered via OCI referrers API
Source repo szl-holdings/hatun-mcp
Certificate issuer https://token.actions.githubusercontent.com
Certificate SAN https://github.com/szl-holdings/hatun-mcp/.github/workflows/publish-mcp.yml@refs/tags/v1.0.4
Rekor log index 3898619383
Predicate type SLSA build provenance https://slsa.dev/provenance/v1
Subject digest sha256:7bdb7b6be2108983534e0d2cad49ad95a4b44d3c202675139b1197337620577c

Background: How many MCP packages publish verified provenance →

MCP tools · 47 exposed · ~3,939 tokens

The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →

Tool Tokens
a11oy_status ~69

[a11oy] HONEST status. This organ is currently UNREACHABLE (organ unreachable: [Errno -3] Temporary failure in name resolution); it exposes zero live MCP tools. Call this tool to see why and when it will self-heal.

NameTypeReqDescription
noop–––

No output schema declared.

No examples provided.

anatomy_observe ~44

[anatomy] Accept a sanitized observation event. Cannot change the decision.. Live organ tool aggregated by Hatun-MCP under PURIQ governance.

NameTypeReqDescription
arguments–––

No output schema declared.

No examples provided.

anatomy_organs ~42

[anatomy] List the five anatomy organs. Status only.. Live organ tool aggregated by Hatun-MCP under PURIQ governance.

NameTypeReqDescription
arguments–––

No output schema declared.

No examples provided.

companion_act ~51

[companion] Request an operator action from the a11oy companion.. Live organ tool aggregated by Hatun-MCP under PURIQ governance. · governance-critical: routed through Byzantine quorum.

NameTypeReqDescription
arguments–––

No output schema declared.

No examples provided.

companion_ask ~61

[companion] Ask the a11oy companion to reason over a question — answers ONLY from live platform data and refuses to fabricate (discloses grounded=false).. Live organ tool aggregated by Hatun-MCP under PURIQ governance.

NameTypeReqDescription
arguments–––

No output schema declared.

No examples provided.

companion_recommend ~43

[companion] Consolidated platform recommendations from the a11oy companion.. Live organ tool aggregated by Hatun-MCP under PURIQ governance.

NameTypeReqDescription
arguments–––

No output schema declared.

No examples provided.

dsse_sign ~112

Sign `payload` with a real ECDSA P-256 DSSE envelope. Uses the server's live signer, initialized from HATUN_MCP_SIGNING_KEY or HATUN_MCP_SIGNING_KEY_PATH. An unusable or absent key returns honesty='UNSIGNED' with empty signatures[]. key_env_var is retained as an advisory compatibility argument; it never selects or reloads a key.

NameTypeReqDescription
key_env_varstring––
payloadobjectyes–

No output schema declared.

No examples provided.

governance_pacbayes_bound ~73

Compute the published PAC-Bayes governance-ensemble bound (F7). REAL closed-form; inputs caller-supplied. Emits a Khipu receipt.

NameTypeReqDescription
deltanumber––
emp_risknumberyes–
klnumberyes–
nintegeryes–

No output schema declared.

No examples provided.

immune_status ~65

[immune] HONEST status. This organ is currently UNREACHABLE (immune unreachable: [Errno -3] Temporary failure in name resolution); it exposes zero live MCP tools. Call this tool to see why and when it will self-heal.

NameTypeReqDescription
noop–––

No output schema declared.

No examples provided.

khipu_append_and_verify ~76

Append one Khipu link carrying `payload` and recompute-verify the chain. parent_hash is advisory (the live chain links to its own last hash); when supplied and it does not match the live tip, we disclose the mismatch.

NameTypeReqDescription
parent_hash–––
payloadobjectyes–

No output schema declared.

No examples provided.

killinchu_status ~69

[killinchu] HONEST status. This organ is currently UNREACHABLE (organ unreachable: [Errno -3] Temporary failure in name resolution); it exposes zero live MCP tools. Call this tool to see why and when it will self-heal.

NameTypeReqDescription
noop–––

No output schema declared.

No examples provided.

llm_tiers ~58

[llm] List the a11oy open-LLM router tiers (id, rank, use, why) from the product route.. Live organ tool aggregated by Hatun-MCP under PURIQ governance.

NameTypeReqDescription
arguments–––

No output schema declared.

No examples provided.

mesh_quorum_status ~117

Byzantine n>=3f+1 mesh-quorum status over the named organs. Returns n, f, threshold (=floor((n+f)/2)+1), the presence set, and the arithmetic quorum boolean. `present` is caller-declared data, not a live cluster probe; live_polled and operational_quorum therefore fail closed. Traces to lutar-lean KhipuConsensus.lean::khipu_consensus_safety.

NameTypeReqDescription
organ_idsarrayyes–
present–––

No output schema declared.

No examples provided.

puriq_master_tool ~98

THE named PURIQ entrypoint over MCP. PURIQ(input, context) -> (Lambda in [0,1], DSSE-signed Khipu receipt, traceparent chain, 13-axis breakdown). Composes Yuyay-13 -> quorum -> HUKLLA -> Khipu append -> DSSE-sign using the server's LIVE chain + signer.

NameTypeReqDescription
context–––
inputstringyes–

No output schema declared.

No examples provided.

second_brain_status ~58

[second_brain] HONEST status. This organ is currently UNREACHABLE (GRANT_REQUIRED); it exposes zero live MCP tools. Call this tool to see why and when it will self-heal.

NameTypeReqDescription
noop–––

No output schema declared.

No examples provided.

szl_a11oy_code_chat ~76

Chat with a11oy.code — the unified open-LLM router (7-tier, GREEN-first). Calls a11oy POST /v1/router. messages = OpenAI-style [{role, content}, ...].

NameTypeReqDescription
messagesarrayyes–
model–––
narrative_axesboolean––

No output schema declared.

No examples provided.

szl_a11oy_operator_reason ~75

a11oy Operator reasoning — a11oy-named twin of szl_companion_reason. Reason over a question + context. Calls the live a11oy companion organ POST /api/a11oy/v1/companion/ask.

NameTypeReqDescription
context–––
questionstringyes–

No output schema declared.

No examples provided.

szl_a11oy_sentinel_scan ~88

a11oy Sentinel inline immune screen — a11oy-named twin of szl_immune_scan. Screen code, an SBOM, or a container image for threat signatures. Calls the live a11oy immune organ POST /api/a11oy/v1/immune/verdict.

NameTypeReqDescription
code–––
image–––
sbom–––

No output schema declared.

No examples provided.

szl_anatomy_3d_render ~51

Return a URL to a Three.js scene snapshot of an organ in the SZL 3D anatomy.

NameTypeReqDescription
animation_statestring––
organstringyes–

No output schema declared.

No examples provided.

szl_axle_call ~167

Consolidated allowlisted passthrough for ANY of the 16 AXLE endpoints (POST api/v1/<endpoint>). `endpoint` must be one of the fixed allowlist (verify_proof, check, extract_decls, extract_proof_states, sorry2lemma, disprove, merge, normalize, rename, theorem2lemma, theorem2sorry, have2lemma, have2sorry, simplify_theorems, repair_proofs, extract_theorems); `payload` is the AXLE request body (content/environment/...). Anything not on the allowlist is declined before any network call. Secondary oracle: corroborates, never authorizes; fail-closed abstain.

NameTypeReqDescription
endpointstringyes–
payloadobjectyes–

No output schema declared.

No examples provided.

szl_axle_check ~98

Check Lean code for errors, as a SECONDARY oracle (POST api/v1/check). Sorry/axiom findings are warnings in AXLE's `okay`; the estate surfaces the raw verdict and classifies only corroborated / not_corroborated / abstain. It never authorizes; the local kernel stays primary.

NameTypeReqDescription
contentstringyes–
environmentstringyes–
timeout_seconds–––

No output schema declared.

No examples provided.

szl_axle_disprove ~103

Attempt to DISPROVE theorems, as a SECONDARY oracle (POST api/v1/disprove). Aligned with PROOF/ABSTAIN doctrine: a successful AXLE disproof of an estate claim must be visible immediately — an adversarial probe is evidence, never a rubber stamp. Fail-closed abstain when unreachable.

NameTypeReqDescription
contentstringyes–
environmentstringyes–
timeout_seconds–––

No output schema declared.

No examples provided.

szl_axle_environments ~54

List AXLE's managed Lean environments (GET /v1/environments). Secondary oracle: informs which pinned environment a corroborating call will run in. It never authorizes; the local kernel stays primary.

Input schema present but exposes no named parameters.

No output schema declared.

No examples provided.

szl_axle_extract_proof_states ~96

Extract the proof state at each line of Lean code (POST api/v1/extract_proof_states). Feeds the estate's self-proving Salad lanes proof STATES, not whole files (instill Tier 1). Secondary oracle: the states corroborate; the local kernel authorizes any discharge.

NameTypeReqDescription
contentstringyes–
environmentstringyes–
timeout_seconds–––

No output schema declared.

No examples provided.

szl_axle_sorry2lemma ~113

Extract every `sorry` (and elaboration error) into standalone named lemma targets (POST api/v1/sorry2lemma). The sorry-gate ratchet hook: turns open sorries into named proof-state targets the estate's self-proving lanes can attack. Secondary oracle: the extracted targets corroborate; the local kernel authorizes their discharge.

NameTypeReqDescription
contentstringyes–
environmentstringyes–
names–––
timeout_seconds–––

No output schema declared.

No examples provided.

szl_axle_status ~67

AXLE secondary-oracle health probe (GET /v1/status). Corroboration infrastructure check only: it never authorizes anything, and an unreachable AXLE is an honest abstain, never a pass. The estate's local Lean kernel remains the authorizing oracle.

Input schema present but exposes no named parameters.

No output schema declared.

No examples provided.

szl_axle_verify_proof ~183

Verify a candidate Lean proof against a formal statement, as a SECONDARY oracle (POST api/v1/verify_proof). SZL tweaks over the raw AXLE tool: strict by default (permitted_sorries defaults to NONE — a proof with any sorry does not corroborate); verify_negation is explicit opt-in; the pinned `environment` is always named so environment drift is receipted. The result corroborates only: the estate's local kernel (szl_lean_verify) authorizes, and this is not human review. Fail-closed: no key / unreachable => honest abstain.

NameTypeReqDescription
contentstringyes–
environmentstringyes–
formal_statementstringyes–
permitted_sorries–––
timeout_seconds–––
verify_negationboolean––

No output schema declared.

No examples provided.

szl_battery_score ~76

Read one estate-battery receipt from the fixed public frontier dataset and return its score AS MEASURED (typed contract enforced: kind szl.estate-battery/mass-training, state MEASURED; anything else is an honest contract mismatch, never a guess).

NameTypeReqDescription
receipt_pathstringyes–

No output schema declared.

No examples provided.

szl_companion_reason ~94

Ask the a11oy companion to reason over a question + context. Calls the live a11oy companion organ POST /api/a11oy/v1/companion/ask (answers only from live platform data; refuses to fabricate). (Renamed 2026-06-16 from the retired codename tool; see DEPRECATED.md.)

NameTypeReqDescription
context–––
questionstringyes–

No output schema declared.

No examples provided.

szl_doctrine_lookup ~51

Semantic lookup across SZL Doctrine v11/v12/v13 + thesis v20. Routed via the a11oy companion grounded-RAG over the doctrine corpus.

NameTypeReqDescription
querystringyes–

No output schema declared.

No examples provided.

szl_drone_lookup ~45

Return the canonical drone DB entry from the killinchu drone database. Calls killinchu GET /v1/drones.

NameTypeReqDescription
model_or_signaturestringyes–

No output schema declared.

No examples provided.

szl_formula_evaluate ~113

Evaluate a named doctrine formula primitive (real closed-form arithmetic). Known: 'puriq' (master operator P(x,t)), 'kl_divergence', 'sigmoid', 'liu_hui_pi'. Unknown names are forwarded to the live lutar-lean kernel /formula-eval route. Optional receipt_challenge binds a caller-owned challenge to the signed call; this does not consume the challenge or prevent replay.

NameTypeReqDescription
args–––
namestringyes–
receipt_challenge–––

No output schema declared.

No examples provided.

szl_frontier_chain_read ~79

Observe one sub-tree of the fixed public SZLHOLDINGS/frontier-54-runs dataset (bounded GET, fixed origin, path-validated tree). Returns file names/sizes and a canonical digest. Accepts no URL, token, or mutation.

NameTypeReqDescription
limitinteger––
treestring––

No output schema declared.

No examples provided.

szl_github_estate_snapshot ~116

Observe the fixed public szl-holdings GitHub estate with bounded GETs. Returns structural repository and open-PR/check evidence plus a canonical SHA-256 digest. It never accepts a URL, organization, token, cursor or mutation. Private repositories, default-branch head SHAs, deployments, model training and runtime health are explicitly outside this v1 scope. Missing pages, rate limits or malformed evidence produce INCOMPLETE or UNAVAILABLE. Signing availability is reported separately by the envelope.

Input schema present but exposes no named parameters.

No output schema declared.

No examples provided.

szl_holo_cells ~79

Pure local deterministic port of the anatomy v4 Λ·HOLO cell math: each name is placed by its FNV-1a hash on the declared trefoil spine. Same chain renders the same organism; the picture is the hash. No network call.

NameTypeReqDescription
namesarrayyes–
treestring––

No output schema declared.

No examples provided.

szl_immune_scan ~108

a11oy Immune (Hukulla) inline screen — screen code, an SBOM, or a container image for threat signatures. Calls the live a11oy immune organ POST /api/a11oy/v1/immune/verdict (the screen IS the signed verdict route). (Renamed 2026-06-16 from the retired codename tool; see DEPRECATED.md.)

NameTypeReqDescription
code–––
image–––
sbom–––

No output schema declared.

No examples provided.

szl_khipu_verify ~65

Verify a Khipu receipt hash (+ optional merkle proof) on a flagship. Calls <flagship> POST /khipu/verify.

NameTypeReqDescription
flagshipstring––
merkle_proof–––
receipt_hashstringyes–

No output schema declared.

No examples provided.

szl_killinchu_cue ~80

Generate a signed Ballistic-of-Effect (BoE) target cue package. STATE-CHANGING: requires a 2-person Yuyay gate (provide X-Second-Approver). Calls killinchu POST /v1/cue.

NameTypeReqDescription
asset_value_polygonobjectyes–
trackobjectyes–

No output schema declared.

No examples provided.

szl_killinchu_detect ~61

Detect/identify a drone or track from RF signature, Remote-ID, or ADS-B. Calls killinchu POST /counter-uas/identify.

NameTypeReqDescription
adsb–––
remote_id_data–––
rf_signature–––

No output schema declared.

No examples provided.

szl_lambda_quorum ~357

Governance-critical Λ verdict under a Byzantine n>=3f+1 quorum (n=5, f=1). Fans the action out to the five SZL organs' policy/verdict routes, collects each organ's verdict, and decides only if >= floor((n+f)/2)+1 (=4) organs AGREE and >= 3f+1 (=4) organs are reachable. Each organ's contribution mints a Khipu receipt; the participating receipts are BLS-aggregated into one signature. The full quorum tally + aggregate ride in `governance.quorum`. HONEST: each organ's reachability is captured from the real HTTP status; any organ whose policy route is not live contributes an honest miss and the degradation is disclosed. OUROBOROS LOOP (in-code, receipt-closed). The organ fan-out — the repo's actual multi-step orchestration path — runs inside `hatun_mcp.loop.run_bounded_loop`: a HARD step budget (env `HATUN_MCP_LOOP_MAX_STEPS`, safe default 12), per-step trace entries, and an HONEST exit reason (`converged` | `budget_exhausted` | `error`, never fabricate convergence). The trace rides in `governance.loop` and in the top-level Khipu receipt detail. Default behavior is unchanged: the default 5-organ set is well under the budget, so it always exits `converged`. Doctrine: "bounded, terminating, receipt-closed" (Λ = Conjecture 1, never a theorem).

NameTypeReqDescription
actionobjectyes–
context–––
organs–––

No output schema declared.

No examples provided.

szl_lean_verify ~41

Verify a Lean theorem against the live lutar-lean kernel. Calls lean-kernel POST /lean-verify.

NameTypeReqDescription
theorem_namestringyes–

No output schema declared.

No examples provided.

szl_puriq_evaluate ~77

Compute the PURIQ master operator P(x,t) for an action + return the factor breakdown (Λ, Yuyay-13, HUKLLA, Khipu, Hatun_MCP). Also queries a11oy policy.

NameTypeReqDescription
actionobjectyes–
contextobjectyes–

No output schema declared.

No examples provided.

szl_thesis_query ~43

RAG query against the thesis-corpus-v18 HF dataset. Calls the a11oy companion grounded-RAG.

NameTypeReqDescription
questionstringyes–

No output schema declared.

No examples provided.

szl_wayra_recent ~59

Recent WAYRA ingestions. WAYRA agent is not yet deployed — returns an honest 'not yet deployed' payload (disclosed in the receipt, never faked).

NameTypeReqDescription
source–––
top_kinteger––

No output schema declared.

No examples provided.

szl_yachay_dome_predict ~56

Yachay-Dome impact prediction for a track over a horizon. Calls killinchu POST /v1/predict-impact.

NameTypeReqDescription
horizon_secondsinteger––
track_idstringyes–

No output schema declared.

No examples provided.

szl_yuyay_score ~37

Return the 13-axis Yuyay breakdown of any content (the heart's scoring).

NameTypeReqDescription
contentstringyes–

No output schema declared.

No examples provided.

yuyay_gate_check ~95

Run the 13-axis Yuyay gate over `input` (input-as-data defense). Returns the per-axis verdicts, the Lambda contribution (0/1 conjunctive AND), and a Khipu receipt. axes_to_run optionally restricts which axes are reported (the gate itself always evaluates all 13; this only filters the report).

NameTypeReqDescription
axes_to_run–––
inputstringyes–

No output schema declared.

No examples provided.

Install

How do I install the Hatun MCP - Governed Agent Gateway server?

Hatun MCP - Governed Agent Gateway runs locally as a container image, launched with docker run --rm -i -e HATUN_MCP_DISABLE_DYNAMIC -e HATUN_MCP_SIGNING_KEY ghcr.io/szl-holdings/hatun-mcp:1.0.4 python -m hatun_mcp.server. Ready-made configuration for Claude, Cursor, VS Code, Codex and 3 more is on this page, copied from each client's own documentation.

oci · ghcr.io/szl-holdings/hatun-mcp

# add to Claude Code
claude mcp add szl-holdings-hatun-mcp -- docker run --rm -i -e HATUN_MCP_DISABLE_DYNAMIC -e HATUN_MCP_SIGNING_KEY ghcr.io/szl-holdings/hatun-mcp:1.0.4 python -m hatun_mcp.server

This image reads HATUN_MCP_DISABLE_DYNAMIC and HATUN_MCP_SIGNING_KEY. Set them in your client's env block for this server; docker run -e passes each one through to the container.

Common questions

What is the Hatun MCP - Governed Agent Gateway server?

Hatun MCP - Governed Agent Gateway is listed in the public MCP registry as io.github.szl-holdings/hatun-mcp. Governed MCP gateway: policy gate on every tool call, hash-chained DSSE receipt on every result. This page covers its container image (ghcr.io/szl-holdings/hatun-mcp).

Is the Hatun MCP - Governed Agent Gateway server safe to use?

Hatun MCP - Governed Agent Gateway scores 63 out of 100 on VerifyMCP. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

What tools does the Hatun MCP - Governed Agent Gateway server expose?

Hatun MCP - Governed Agent Gateway exposes 47 tools: szl_a11oy_code_chat, szl_killinchu_detect, szl_killinchu_cue, szl_immune_scan, szl_companion_reason, and 42 more. Their descriptions and schemas cost roughly 3,939 tokens of context every time the server is loaded.

Is the Hatun MCP - Governed Agent Gateway server still maintained?

Hatun MCP - Governed Agent Gateway is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

What licence is the Hatun MCP - Governed Agent Gateway server under?

Hatun MCP - Governed Agent Gateway declares the Apache-2.0 licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.

What did VerifyMCP test on the Hatun MCP - Governed Agent Gateway server?

VerifyMCP ran 12 checks on Hatun MCP - Governed Agent Gateway, covering Source repository, Provenance, License, Maintenance and 8 more. 4 items could not be completed, a gap on our side rather than a finding about Hatun MCP - Governed Agent Gateway: Malware scan, Known CVEs, Install scripts and Dependency health. One category scores 0 for that reason, because we only credit what we can confirm. Endpoint Security and Transport & Reachability do not apply to a container image, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.