ControlKeel
Turns engineering habits into enforceable policy gates, persistent evidence, and reusable memory for coding agents.
- Source repo
- aryaminus/controlkeel
- Stars
- ★ 11
- Last updated
- 3d ago
- License
- NOASSERTION
- Primary language
- Elixir
- FA score
- 58/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformCodexClaude Code (Partial support)
- Cost
- Free tier plus a paid hosted plan
- Setup effort
- Medium · a few setup steps
- You'll need
- Typical use
- A development team using Codex, Claude, OpenCode, or another supported host wants consistent checks before agent-generated changes reach the main branch.
- Not a fit if
- Teams that need fully offline operation with data never leaving the machine (cloud telemetry, workspace keys, and callbacks send evidence out).
- Teams wanting a fully unattended setup; attach may need workspace trust, manual provider config, and a host restart.
- Teams that need clear license terms before adoption; the license is reported as NOASSERTION.
- Source review
- 58/100 · Major gaps
What does this agent do, and when should you use it?
ControlKeel is a governance control plane for AI-assisted engineering that compares intended delivery with actual changes before risky work reaches the main branch. It ships a CLI, a full local stdio MCP tool set, project binding, host attachment, runtime exports, and records for findings, reviews, proof bundles, budgets, typed memory, and task outcomes. Its workflow captures scope, risk, budget, domain rules, and human review preferences, then applies deterministic validation and optional advisory review with human approval where policy requires it. Evidence can remain local, while opt-in cloud telemetry, workspace keys, cloud run packages, callbacks, and configurable synchronization provide additional team paths. Development uses Phoenix and Ecto on SQLite, with single-binary builds distributed through Burrito and GitHub Releases.
ControlKeel initializes governance state with controlkeel setup and connects a supported coding-agent host through controlkeel attach <host>; when only MCP is available, the documented path is ck_attach. controlkeel attach doctor diagnoses the host connection, controlkeel provider doctor checks provider configuration, controlkeel status reports state, and controlkeel findings displays recorded findings. During governed work it evaluates agent output against captured scope, risk, budget, domain packs, and human preferences, runs deterministic scans and optional advisory review, blocks or requests approval when policy calls for it, and persists findings, reviews, proof bundles, memory, cost, and task outcomes. Its observability loop also organizes timelines, trends, problem clusters, eval candidates, benchmark drafts and history, and promotion advisories so recurring failures can become bounded deterministic checks backed by regression evidence.
- A development team using Codex, Claude, OpenCode, or another supported host wants consistent checks before agent-generated changes reach the main branch.
- A DevSecOps group needs security or compliance rules that can block high-impact actions while retaining findings and proof bundles.
- An engineering organization that changes agent hosts wants project binding, MCP, host attachment, and runtime exports to preserve governance evidence and typed memory.
- A team with extensive domain conventions wants recurring review preferences and delivery habits enforced instead of repeatedly explaining them to a model.
- An agent-evaluation team wants to turn traces and repeated failures into regression evidence tied to specific suites and subjects.
How do you install or deploy this agent?
Choose one of the documented installation routes. The supplied material does not state minimum versions for Node.js, Homebrew, or supported operating systems, and it does not identify credentials required for the initial local run.
Homebrew:
brew tap aryaminus/controlkeel && brew install controlkeelnpm:
npm i -g @aryaminus/controlkeelRelease installer:
curl -fsSL https://github.com/aryaminus/controlkeel/releases/latest/download/install.sh | shWindows PowerShell:
irm https://github.com/aryaminus/controlkeel/releases/latest/download/install.ps1 | iexSource development requires the Elixir Mix toolchain; the documented development sequence is:
mix setup
mix phx.server
mix test
mix precommitHow do you use this agent?
Run the first setup from the repository you want to govern. This documented example attaches OpenCode; substitute another supported host name where applicable.
controlkeel
controlkeel setup
controlkeel attach opencode
controlkeel attach doctor
controlkeel provider doctor
controlkeel status
controlkeel findingsProject scope is the default and writes host files only inside that repository. Use --scope user only for targets that explicitly support user scope; it writes host-level user configuration but does not make project binding or proof state global. Restart the host after attachment or plugin changes. Workspace trust, manual provider configuration, a required restart, or a plan-review approval that cannot auto-wait may require human action. If ControlKeel is exposed only through MCP, call ck_attach for the current host.
What are this agent's strengths and limitations?
- It provides CLI, local stdio MCP, native host attachment, and runtime-export paths rather than requiring a single agent-host interface.
- It can perform deterministic policy validation, gate high-impact work, and persist findings, reviews, and proof bundles instead of only suggesting behavior.
- Typed memory, costs, task outcomes, eval evidence, and promotion advisories form a concrete path from recurring failures to regression-backed checks.
- Its bounded public baseline reports that version 0.3.45 caught 12 of 12 risky
host_comparison_v1scenarios, blocked 9, and used zero provider tokens for deterministic validation.
- Integration depth varies: the material distinguishes native attachment, a narrower OAuth-scoped hosted MCP set, minimal A2A, and fallback validation or proxy paths, so teams must verify the tier for each host.
- Initial integration may require workspace trust, manual provider configuration, and a host restart, preventing a uniformly unattended setup.
- The published evidence is limited to named suites, subjects, and scoring definitions; deterministic scanner results are not proof of model-backed performance across arbitrary repositories.
- The repository license is reported as
NOASSERTION, and no verifiable license terms are included in the supplied material. - Bidirectional cloud synchronization is dormant until configured, while cloud paths add decisions around workspace keys, telemetry, run packages, and callbacks.
How does this agent compare with similar options?
Compared with relying only on an *.md file that tells a model how to behave, ControlKeel is designed to inspect actual output and express recurring expectations as typed memory, policy checks, findings, and proof bundles. In the named host_comparison_v1 suite, null_policy_baseline caught 0 of 12 risky scenarios while controlkeel_validate caught 12 of 12; that result is bounded to the documented suite and should not be generalized to every host or repository.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| ControlKeel This agent | 58 · Major gaps | CLIFreemium | ★ 11 | 3d ago | Elixir | Codex |
| Alas Agent Workspace | 79 · Good | Desktop appFree + model costs | ★ 24 | today | Swift | Codex · Claude Code |
| Jev Review | 62 · Some gaps | Agent plugin / skillFree + model costs | ★ 249 | 1mo ago | TypeScript | Codex · Claude Code |
| Frad’s Claude Code Plugins | 45 · Major gaps | Agent plugin / skillFree | ★ 596 | 2d ago | JavaScript | Claude Code |
How does FollowAgents rate this agent?
Why each dimension lost points
The README and getting-started guide say project attach defaults to changes inside the governed repository, user scope is limited to hosts that support it, and high-impact actions require human approval. These support least privilege and confirmation, but the sample does not show the underlying tool permissions or approval implementation. The docs distinguish the full local MCP set from the narrower hosted OAuth set and identify custom gateways as trust boundaries. Secret handling details, dependency audit evidence, and recovery from installation side effects are thin. Install examples pipe remote scripts to a shell and npm downloads a release binary; those external effects are disclosed but no verification step is shown. Repository attribution is visible, but publisher identity is unverified and maintenance ownership or update commitments are unclear.
The README, getting-started guide, and integration skill are broadly consistent about project binding, scope, and local versus hosted MCP. They provide doctor, status, and findings commands for diagnosis. The sample lacks dependency lock or supply-chain review details and does not show CLI messages for download or network failures, limiting support for dependency availability and failure-message quality.
The docs address individuals and teams using coding agents and cover multiple hosts, MCP-only integrations, native skills, and local or cloud paths. The skill selects an installation approach by integration type and limits user scope to targets that advertise support. Triggers and scenarios are reasonably clear; project defaults and Windows, source, and packaged installation paths show environment fit. The sample does not include the full support matrix or every environment constraint, so full marks are not justified.
The task-oriented documentation map links getting started, support, integrations, benchmarks, and API/CLI references. Installation guidance covers several methods and verification commands, but remote installers have no checksum or signature verification step in the sample. Skills use stable names and frontmatter and include actionable examples; a complete FAQ is not shown. Docs disclose limitations such as the narrower hosted MCP set and bounded benchmark claims. The license clearly states Apache-2.0. Release markers and an automatic version-bump workflow provide some versioning evidence, but the changelog contents are not included. Maintenance responsibility is not stated.
The product describes a loop from policy capture through deterministic validation and conditional human approval to persistent evidence and regression evaluation, with bounded benchmark figures. Findings, proof bundles, status, and benchmark comparison are plausible usable outputs. The sample does not show actual output formats or a complete cost-benefit measurement, so some value claims remain product assertions.
The README bounds benchmark claims to a named suite, subject, and scoring definition and points to benchmark documentation and reproduction protocols; the getting-started guide supplies run and compare commands. Product descriptions and the integration skill corroborate local and hosted capability boundaries, and version information has a corresponding workflow clue. The sample omits benchmark tables, test results, and implementation code needed to independently verify the claims. It does distinguish deterministic scanner evidence from model-backed evidence, providing some separation of fact and inference.
- Installation can execute a remote script or download a release binary; the sample shows no checksum, signature verification, or install rollback procedure. Verify the selected release artifact and its origin before use.
- Benchmark figures are limited to the documented suite, subject, and scoring definition. This static sample cannot independently reproduce them or establish model-driven agent effectiveness.
Maintain this project? Add the FARS badge to your README
[](https://followagents.com/en/agents/controlkeel)<a href="https://followagents.com/en/agents/controlkeel"><img src="https://followagents.com/badge/controlkeel.svg" alt="FARS review"></a>Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
Release notes excerpt
ControlKeel v0.4.21 Install - Homebrew: brew tap aryaminus/controlkeel && brew install controlkeel - npm bootstrap: npm i -g @aryaminus/controlkeel - Unix installer: curl -fsSL https://github.com/aryaminus/controlkeel/releases/latest/download/install.sh | sh - Raw GitHub shell installer: curl -fsSL https://raw.githubusercontent.com/aryaminus/controlkeel/main/scripts/install.sh | sh - PowerShell installer: irm https://github.com/aryaminus/controlkeel/releases/latest/download/install.ps1 | iex …View on GitHub ↗ -
Release notes excerpt
ControlKeel v0.4.20 Install - Homebrew: brew tap aryaminus/controlkeel && brew install controlkeel - npm bootstrap: npm i -g @aryaminus/controlkeel - Unix installer: curl -fsSL https://github.com/aryaminus/controlkeel/releases/latest/download/install.sh | sh - Raw GitHub shell installer: curl -fsSL https://raw.githubusercontent.com/aryaminus/controlkeel/main/scripts/install.sh | sh - PowerShell installer: irm https://github.com/aryaminus/controlkeel/releases/latest/download/install.ps1 | iex …View on GitHub ↗ -
Release notes excerpt
ControlKeel v0.4.19 Install - Homebrew: brew tap aryaminus/controlkeel && brew install controlkeel - npm bootstrap: npm i -g @aryaminus/controlkeel - Unix installer: curl -fsSL https://github.com/aryaminus/controlkeel/releases/latest/download/install.sh | sh - Raw GitHub shell installer: curl -fsSL https://raw.githubusercontent.com/aryaminus/controlkeel/main/scripts/install.sh | sh - PowerShell installer: irm https://github.com/aryaminus/controlkeel/releases/latest/download/install.ps1 | iex …View on GitHub ↗
FAQ
Does local validation necessarily incur model-provider cost?
Will setup write outside my repository?
--scope user writes user configuration only for targets that explicitly support it, and it does not globalize project binding or proof state.Is the cloud service mandatory?
What should I check when attachment fails?
controlkeel attach doctor, controlkeel provider doctor, and controlkeel status; also verify workspace trust and restart the host after attachment or plugin changes.Do the benchmark numbers prove equivalent results on every model and host?
Thanks — we use this to improve the page.