Coding agent

Poolside Agent CLI

Provider-flexible terminal and ACP client with explicit policy, traces, and optional sandboxes.

Status
Active
License
Proprietary client
Evidence
Documented, 18 sources
Product record checked
2026-07-27

At a glance

Poolside's proprietary-distribution CLI combines interactive and JSONL automation modes, ACP client and server roles, OpenRouter and local OpenAI-compatible models, MCP, skills, worktrees, structured trajectories, and optional local or managed sandbox environments.

Good choice if

  • Developers who want one TUI to drive Poolside, OpenRouter, Ollama, local OpenAI-compatible models, or other ACP agents
  • Enterprise workflows that need reviewable policies, trajectories, and managed sandbox definitions
  • Scripts and ACP-compatible editors that need structured output, resumable sessions, MCP, and skills

Check before choosing

  • The local environment is enabled by default; container or managed sandbox isolation must be selected and configured
  • Remote HTTP or SSE MCP servers are outside local sandbox network restrictions
  • Rewind removes conversation turns but does not restore files, so no harness checkpoint or file rollback is claimed
See 2 more considerations
  • No delegated subagent workflow was verified; parallel isolation is limited to manually launched worktrees or separate sessions
  • The public repository distributes binaries, documentation, and a changelog rather than the proprietary client source, and its v1.0.13 tag points to a tree whose changelog still says 1.0.12

Why it qualifies as a coding harness

This confirms category fit, not product quality. Every required criterion links back to first-party evidence.

Qualifies4 of 4 required criteria evidenced
  • Adaptive agent loop

    Documented

    The system repeatedly observes results and chooses the next action instead of following a fixed one-pass graph.

  • Repository tool execution

    Documented

    The system can use tools to inspect and change a repository or its execution environment.

  • Task-aware context management

    Documented

    The runtime assembles, updates, compacts, retrieves, or persists task-relevant context while work proceeds.

  • Model-independent runtime control

    Documented

    Permissions, budgets, interruption, policy, or stop controls operate outside the model's own text generation.

Membership establishes category fit only. It does not score quality, safety, autonomy, model capability, or benchmark performance. · Read the membership rule.

How it works under the hood

Seven mechanisms mapped from first-party records. These labels describe what the harness provides, not how intelligent its model is.

7/7layers documented
  • Execution & isolationSandbox availableDocumented mechanism, not a performance score.
  • Tooling & integrationsExtensible toolsDocumented mechanism, not a performance score.
  • Context & stateManaged contextDocumented mechanism, not a performance score.
  • Lifecycle & recoverySession resumeDocumented mechanism, not a performance score.
  • ObservabilityStructured tracesDocumented mechanism, not a performance score.
  • VerificationTool-assistedDocumented mechanism, not a performance score.
  • Governance & permissionsPolicy controlsDocumented mechanism, not a performance score.

Public code audit

Unrankedsupport-only repository
Security policy
Not found
CI workflow
Not found
Automated tests
Not found
Evaluation assets
Not found
Contributor documentation
Not found

The public repository contains binary releases, a readme, a changelog, license notices, and issue tracking, but no client or service source, CI, engineering tests, or evaluation suite. The v1.0.13 Git tag resolves to ca5306676ab5fa9a0cfda2d4844470bf7d435741, where CHANGELOG.md still begins at 1.0.12; the 1.0.13 changelog entry was added later on main.

Inspect commit 39e9094bd5d49d2dc4df780753cc6da37fc88eb6, checked 2026-07-27

Measured configurations

No benchmark run passes the full metadata admission policy for this harness yet. Missing data is not scored as zero.

Benchmark policy and all runs

Capability support

Documented first-class product support, checked against the sources below.

External tools (MCP)
DocumentedProduct-supported MCP integrationSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Local models
DocumentedLocal or self-hosted model pathSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Agent parallelism
Not documentedNo first-class support established by the current recordAbsence of current documentation is not proof that the capability is impossible.
Runs without an open UI
DocumentedNon-interactive or automation surfaceSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Browser control
Not documentedNo first-class support established by the current recordAbsence of current documentation is not proof that the capability is impossible.
Isolated execution
Depends on surfaceIsolation depends on the managed or configured execution surfaceSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
Undo file changes
Not documentedNo first-class support established by the current recordAbsence of current documentation is not proof that the capability is impossible.

Primary evidence

Each capability claim is tied to a first-party record and a verification date.

Product record checked 2026-07-27
View 10 additional sources
Additional first-party evidence10 sources
Organization permissionsRBAC for agents, auto approval, credentials, MCP, repositories, sandboxes, and cross-user trajectoriesOfficial docsManaged agentsReusable model, instruction, tool, repository, credential, skill, MCP, and sandbox configurations with runtime access intersectionOfficial docsAgent skillsLocal, project, and organization-managed skills, access controls, and sandbox mount limitationsOfficial docsIndexed repositoriesOrganization-managed repository indexing and agent access controls for broader code contextOfficial docsPoolside model deploymentProprietary Poolside models deployed within the customer's environment; kept separate from harness capabilityOfficial docsPoolside platform deploymentOn-premises and air-gapped platform deployment, models, control plane, storage, and infrastructure requirementsOfficial docsPoolside CLI 1.0.13 releaseCurrent binary release and published distribution artifactsOfficial repositoryRelease repository readme at inspected commitOpenRouter, Ollama and OpenAI-compatible local endpoints, ACP client and server roles, worktrees, MCP, settings, and permission syntaxOfficial repositoryRelease changelog at inspected commitVersion history through 1.0.13, JSON logging, worktrees, session persistence, compaction, and provider additionsOfficial repositoryDistribution license at inspected commitProprietary end-user license notice for the distributed clientOfficial repository