offload-harness (dmmdea/offload-harness) is an MCP server listed on the M8ven Trust Index. It scores 67 out of 100, grade C. It declares 47 tools. No publisher has claimed this listing.

Limited view

67/100

offload-harness

Item: offload-harness
Rating: 67
Author: M8ven

Delegate summarize/classify/extract/triage to a FREE local Gemma-4 cascade via llama.cpp. Go CLI + MCP server; never calls a cloud model. Includes a turnkey setup skill for Claude Code.

→ dmmdea/offload-harness · listed on github_repo_search

Limited view. Automated analysis covers part of this stack. Findings reflect what we verified. Grades reflect the full trust pyramid: code, verification depth, and reputation. New projects cap at C until adoption is earned.

Limited view: static analysis for Go is partially covered.

How we verified

⚡ Live Monitored: not connected

Verified is a snapshot. Live keeps it current, and builds your track record.

⚡ Connect GitHub → continuous verification on every push why connect →

Who stands behind it

dmmdea

Source: github_repo_search

Is this your MCP?

Claim it to get a verified publisher badge, a free copy of our full audit findings, and direct contact for any high-priority issues we find. Or connect your repo for our deepest verification, Live Monitored: read-only, revoke anytime. What we access →

⚡ Connect GitHub for Live Monitored

Install from

GitHub

The grade above is for the source repository. Registries can serve a different version, so we mark the ones we were not able to read.

// key findings

✅

No credential exfiltration, no sensitive file access, no obfuscation

Static analysis found nothing flowing your secrets to unexpected places.

✅

Open source with a license and README

Anyone can audit the code, the license is declared, and the publisher documents what it does.

// tools this server exposes44 tools

These names and descriptions are the publisher's own, read from the source code. We print them as written. Our assessment is the findings above, not this list.

offload_status

offload_summarize

Classify text into one of the given labels on the LOCAL model cascade (free, on-box, no cloud). Returns {label, confidence}; low-confidence results are deferred back to you. Triggers: classify / categorize / label / tag / bucket / route text into one of a known set.

offload_classify

offload_extract

Extract structured fields from text on the LOCAL model cascade (free, on-box, no cloud), constrained to the provided JSON schema. Returns the extracted object or defers. Triggers: extract / parse / pull out structured fields from text into a schema (names, dates, amounts, entities).

offload_triage

Answer a yes/no/unsure question about text on the LOCAL model cascade (free, on-box, no cloud). Returns {decision, reason} or defers. Triggers: a yes/no/unsure check on text — 'does this contain X?', 'is this relevant/spam/safe?', 'should this be flagged?'.

offload_vqa

Answer a question about a VIDEO on a free local vision model. It samples frames from the video and reasons over them. video is a LOCAL file path; question is what to ask. Returns {answer} (which notes what the relevant frames show); if it can't answer confidently it returns deferred:true and you sho…

offload_video_describe

offload_video_watch

offload_transcribe

Extract structured fields from an IMAGE on a free local model: it OCRs the image, then extracts the fields from the transcribed text constrained to the provided JSON schema (values are grounded against the OCR text). image is a local file path or a data:image/... URI; schema is a JSON schema with a …

offload_extract_image

offload_assess_image

offload_ocr

offload_generate_image

offload_run_graph

offload_generate_svg

offload_generate_video

offload_animate_character

offload_generate_audio

offload_edit_image

offload_inpaint_image

offload_upscale_image

Rewrite a local image from a TEXT INSTRUCTION on the LOCAL ComfyUI for FREE — no mask (Qwen-Image-Edit class: the model reads the source through its own vision encoder and re-renders the whole frame). This is the route for instruction edits that have no drawable region: \

offload_edit_image_generative

offload_media

offload_nim

agent_rig

agent_run

offload_ask

offload_review_diff

agent_delegate

offload_research

comfyui_search

Search the comfyui API for endpoints matching a natural-language query. Returns a ranked list of {endpoint_id, method, path, summary} entries. Call this first to find the endpoint to execute.

comfyui_get

Get metadata for one GET endpoint by its endpoint_id (from comfyui_search). This registry-only lookup never calls the API.

comfyui_execute

Execute one comfyui API endpoint by its endpoint_id (from comfyui_search). Params are passed as a JSON object; path placeholders and query strings are resolved automatically.

diagnose_an_invisible_model

Separates unregistered model class from not-listed and no-such-input, each with its own remedy, instead of reporting a generic missing file.

preflight_a_graph_offline

The server has no validate-only endpoint, so this is the only dry run that does not consume a queue slot.

patch_a_graph_without_silently_hitting_the_wrong_node

Refuses the write when node 6 no longer holds the expected class, which is how a template revision fails loudly rather than quietly.

search

Full-text search across all synced data. Faster than paginating list endpoints. Requires sync first.

sql

Run read-only SQL against local database. Use for ad-hoc analysis, aggregations, and joins across synced resources. Requires sync first.

context

Get API domain context: resource taxonomy, auth requirements, query tips, and unique capabilities. Call this first.

llamaswap_search

Search the llamaswap API for endpoints matching a natural-language query. Returns a ranked list of {endpoint_id, method, path, summary} entries. Call this first to find the endpoint to execute.

llamaswap_get

Get metadata for one GET endpoint by its endpoint_id (from llamaswap_search). This registry-only lookup never calls the API.

llamaswap_execute

Execute one llamaswap API endpoint by its endpoint_id (from llamaswap_search). Params are passed as a JSON object; path placeholders and query strings are resolved automatically.

free_vram_safely_before_a_render_job

Drains in-flight work first and always excludes the protected keep-set, then confirms via /running.

prove_the_memory_stack_survived_a_config_change

Asserts embedder cosine and reranker score against stored calibrated baselines — catches dropped flags roster checks miss.

// environment variables

To run this server yourself, you supply these values. They go in your own MCP client configuration and stay on your machine. The secret label means the value is sensitive, not that the server mishandles it.

configCOMFY_API

configCOMFY_BATCH_MAX_CONSEC_FAIL

configCOMFY_CKPT

configCOMFY_CM_CLI

configCOMFY_COMPUTE_DEVICE

configCOMFY_EDIT_MEGAPIXELS

configCOMFY_EDIT_PRESET

configCOMFY_EDIT_UNET

configCOMFY_EXTRA_ARGS

configCOMFY_START_WAIT_SEC

configCOMFY_SUBMIT_DEADLINE_MS

configCOMFY_UPSCALE_MODEL

configCOMFY_VAE

configCOMFY_WAIT_SEC

configGGML_VK_VISIBLE_DEVICES

configLLAMA_SWAP_API

configMEMORY_STACK

configOPENCODE_LOCAL_OFFLOAD_OPTIONS

configSDCPP_BIN

configTTS_PY

configCORAL_BIND

configCORAL_PORT

configCORAL_IDLE_SEC

configCORAL_MODELS_DIR

configCORAL_MANIFEST

configCORAL_ENABLED

configCORAL_DELEGATE

configCORAL_LOG

configTTS_DEVICE

// quality suggestions

Tool annotations

No tools have read-only/destructive annotations

Add readOnlyHint or destructiveHint annotations to every tool so hosts can warn users before invoking.

All four hints declared on every tool

47/47 tools missing one or more hints — offload_status (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); offload_summarize (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); offload_classify (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint), +44 more. OpenAI's directory rejects tools where any of the four hints are missing or non-boolean.

For every tool, set all four hints (readOnlyHint, destructiveHint, idempotentHint, openWorldHint) to explicit true/false values that match the handler’s actual behaviour.

Tool test coverage

35/47 tools referenced in tests (74%)

Write tests that reference each tool by name so every tool has at least one test.

Domain consistency

npm scope @offload-harness doesn't match GitHub owner dmmdea

Use the same org name across GitHub, npm, and your homepage so users can verify the publisher.

Claim the listing to review these findings one by one and send us a correction where you disagree, straight to the team. Claiming also means we tell you when the grade moves, and reach you first if we find anything urgent.

// full audit trail

The findings above are the summary. The full trail, every check we ran, each deduction, the network hosts observed and the dependency advisories, goes to verified publishers, along with an alert whenever a new one lands. Verified publishers can also review each finding and dispute it in one click. Publisher corrections have sharpened several of our checks this month, because the maintainer knows the codebase better than any scanner.

// improvement guidance — verified publishers only

We have 4 concrete improvements we can share with the publisher of this MCP. Each comes with specific guidance to raise the trust score.

// embed badge in your README

[![M8ven Score](https://m8ven.ai/badge/mcp/dmmdea/offload-harness)](https://m8ven.ai/mcp/dmmdea/offload-harness)

Shows your grade and updates automatically. Prefer no grade? Append ?variant=verified to the badge URL.

commit: 31dbe2122253dfa92b144c40a8c13c59dd8c3611

code hash: 5793b8fba03feb4d4e9d485ef20c162f2e6b68cdcb7295012f833258e3af9c32

view raw JSON →

Check MCPs from inside your assistant

Tool Check · MCP

Vetting this one by hand? Tool Check is an MCP that scores other MCPs. Add it once and ask Claude, ChatGPT, or any MCP client to grade a server, surface CVEs, check the publisher, and suggest safer alternatives — before you install.

https://m8ven.ai/api/mcp/tool-check

check_toolsearch_toolscompare_toolsrecommend_alternativescheck_publisherreport_concern

How to add it →Free · no account needed · works in any MCP client