From Markdown to agents in production.
NeoSpec turns a Markdown file into a deployed agent — authorized, monitored, budget-capped, and fail-safe.
---name: QA Agentmodel: anthropic-opus-5budget_usd: 25permissions: - linear [read_issue] - github [pr_comment]trigger: - github [pr_open] - slack [message]---You are a QA agent, for testing features in side branches beforethey get merged. Launch an environment and test the features asdescribed in Linear, and make sure it's working as expected.Your team already has agents running in production.
A Slack bot here, a scheduled script there — built fast, shipped without review, and trusted with real access. They work, until something goes wrong and there's no record of what happened.
No audit trail
Nobody can say what an agent actually did once it touched production.
No spending limit
A bad prompt loop or runaway retry can burn through API budget overnight.
No access boundary
Agents run with whatever credentials they were handed, scoped to nothing in particular.
From Markdown file to running agent
No new DSL to learn and no dashboard wizard to click through — an agent is just a file in your repo.
Write an agent as a .md file
Describe the agent's model, budget, permissions, and trigger in a YAML frontmatter block, then write its instructions in plain language below it.
Push to GitHub
Commit the file to your repo like any other change. No separate dashboard step required to define an agent.
NeoSpec detects & deploys it
NeoSpec watches your repo for pushes, finds agent definitions automatically, and deploys them.
Run, monitor, and audit
Watch runs live, message an agent or trigger it on demand, keep it inside its budget, and review a full audit trail of every action it took.
Everything you need to run agents in production
Not just a place to run a prompt — the controls that make it safe to hand an agent real access.
Budgets & cost control
Cap every agent at a hard USD budget per run. NeoSpec halts execution the moment it's reached — no surprise bills.
Scoped permissions
Grant an agent exactly the integrations and actions it needs — read a Linear issue, comment on a PR — nothing more.
Event triggers
Kick off agents from real events: a pushed commit, an opened PR, a new issue, a schedule, or a webhook.
BYOK
Bring your own key for any LLM provider. Your usage, your billing relationship, your rate limits.
Failure retries
Transient failures don't kill a run. NeoSpec retries with backoff and picks up where the agent left off.
Custom Docker runtimes
Give an agent exactly the tools it needs by running it inside your own container image.
Audit log
Every action, tool call, and dollar spent is recorded — searchable, exportable, and attributable to a run.
Agent workflows
Chain steps within a run or hand off between stages — define multi-step behavior right in the agent's file.
Multi-agent orchestration
Compose agents that call on each other, so complex jobs get split across specialists instead of one do-it-all bot.
Connect once, use everywhere
Authorize an integration in the dashboard, then grant any agent scoped access to it — no re-authenticating per agent.
Built to be trusted with real access
Handing an agent write access to your systems only works if the platform underneath it is airtight.
Your data stays in your infra
Self-host NeoSpec and your code, secrets, and agent runs never leave your network.
Isolated runtime sandboxing
Every agent executes inside its own Docker runtime, sandboxed from your host and other agents.
Encrypted key custody
BYOK credentials are encrypted at rest and scoped to the agents that are explicitly granted access.
Compliance roadmap
SOC 2 and data-residency options are on the roadmap for NeoSpec Cloud customers.
Self-host it free, or let us run it for you
Same runtime either way — pick based on how much infrastructure you want to own.
Self-Host
Free & open source
- Run the full runtime on your own infrastructure
- MIT-style license, no seat fees
- Community support on Discord
NeoSpec Cloud
Managed for you
- Zero-ops, one-click agent deploys
- Team roles & shared integrations
- Priority support & uptime SLA
Frequently asked questions
Want a sneak peek?
We're in private alpha. Talk to us and we'll show you NeoSpec running on your use case.