deploy

A workflow for making an Elliot connector available to agents at runtime. A connector packages tools for agent use, and the workflow builds, checks, evaluates, and publishes it.

In plain words
What is it for?
Use it when a connector is ready to ship: build it, lint it, run its evaluation cases, and publish it to Elliot Cloud.
Why use it?
It catches build, lint, evaluation, or runtime problems before the connector is made live. Publishing only happens after the required validation steps succeed.

Skill for Claude CodeCodex

Part of the elliot plugin — 10 skills, 2 hooks, 1 MCP server shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/elibarak12/elliot/deploy
Any agent
npx skills add EliBarak12/Elliot --skill deploy
Clone the repo
git clone --depth 1 https://github.com/EliBarak12/Elliot

Made for: Claude Code, Codex.

Or install elliot, the plugin that ships this one along with the rest of its 10 skills, 2 hooks, 1 MCP server.

Per session 42 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 782 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 2 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00042 $0.00782
Opus 5 $0.00021 $0.00391
Sonnet 5 $0.00008 $0.00156
Haiku 4.5 $0.00004 $0.00078

Measured 3d ago against content hash 03de5ae444ca, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

deploy scanned grade C with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Downloads and executes remote codehighSupply chain

curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.

Runtime health: !`curl -s http://localhost:3001/health 2>/dev/null | python3 -c "import sys,json; d=json.load(sys.stdin); print(d.get('status','unknown'))" 2>/dev/null || echo "not running"`

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

Runtime health: !`curl -s http://localhost:3001/health 2>/dev/null | python3 -c "import sys,json; d=json.load(sys.stdin); print(d.get('status','unknown'))" 2>/dev/null || echo "not running"`
skills/deploy/SKILL.md · 64 lines

How it starts

The opening of the file, as written. The whole thing — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Deploy Connector Workflow

Pre-flight status

!ls connectors/*.connector.json 2>/dev/null || echo "(no connectors)" Runtime health: !curl -s http://localhost:3001/health 2>/dev/null | python3 -c "import sys,json; d=json.load(sys.stdin); print(d.get('status','unknown'))" 2>/dev/null || echo "not running"

Deploy pipeline (must complete in order)

1. Build

Call elliot_build_connector so the connector reflects the latest tools. Lint, eval, and export all operate on this built connector.

2. Lint

Call elliot_lint_connector (no arguments — it lints the built connector). Do not proceed if any errors are reported. Fix all issues, call elliot_build_connector again, re-lint, and confirm zero errors.

3. Eval

If an eval suite exists: call elliot_run_eval. All cases must pass. Fix failures before deploying.

4. Ship it

On Elliot Cloud (the marketplace default) — one call: elliot_cloud_publish. It runs a publish-time smoke gate first (a cache-safe runtime build + tools/list

  • a per-tool execute of each auto-callable READ tool), so a connector that would 404 on every agent call is blocked instead of shipped. On success it deploys to a stable public MCP URL on your tenant and returns it. If it blocks (lint_errors, missing_secrets, smoke_failed), fix and re-publish. Choose who can call it with auth_mode: api_key (default, shared key), personal (your workspace only), third_party_oauth (each user connects their own upstream account), or open (anyone, no key — non-sensitive data only). This is the whole deploy on Cloud — skip the local steps below.

Running Elliot locally instead:

  • elliot_export_connector — write the connector file (path, or the .elliot/connector.json default).
  • elliot_start_runtime — launch the connector-runtime subprocess. Pass connector_path if you exported to a non-default path, or it loads the wrong (or no) connector. Confirm with elliot_runtime_logs; get the agent URL with elliot_get_connection_config; stop later with elliot_stop_runtime. (These tools are local-only — they aren't served on Elliot Cloud.)

Read the full file on GitHub · 64 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 64 lines · 42 tokens per session scan C 03de5ae444ca

Subscribe to this mod's changes

deploy is a skill published in the GitHub repository EliBarak12/Elliot (11 stars, last pushed 6d ago), licensed MIT. It adds 42 tokens to every session and 782 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.