HarnessMatch

Search 52 product records by name, workflow, interface, or documented capability.

Quick access
Use ↑ ↓ to move, Enter to open, Esc to close.

Quick access links shown.

Menu

General-purpose agent

Kern

Self-hosted personal agent with coding tools, durable recall, channels, and subagents.

Status
Active
License
MIT
Evidence
Documented + code-verifiable, 9 sources
Product record checked
2026-07-28
Compare harnessesInspect sources

At a glance

A young MIT-licensed personal agent that qualifies as a coding harness through its iterative tool loop, repository and shell mutation, persistent JSONL and vector recall, tool scopes, background runtime, MCP, local Ollama path, and parallel read-only subagents across terminal, web, desktop, and messaging clients.

Good choice if

  • Self-hosters who want a personal agent that can also work directly in repositories
  • Long-running, multi-channel workflows with persistent local recall and scheduled tasks
  • Local Ollama users who want MCP and bounded parallel research subagents

Check before choosing

  • Kern is a general personal agent rather than a coding-first product, so its defaults and documentation span many non-repository workflows
  • Tool control uses coarse full, write, or read scopes rather than a documented per-call approval system; normal local execution remains host-first
  • Docker is an optional deployment boundary for the whole service, not proof of a per-task disposable sandbox
See 2 more considerations
  • Subagents are bounded read-only workers without the parent's MCP, notes, skills, or recall; running children do not survive restart and completed children are not reloaded into the live list
  • The project is young, has no repository security policy or contributor guide, and its engineering tests are not independent product evaluation

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.

    EvidenceKern overview

  • Repository tool execution

    Documented

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

    EvidenceKern tools

  • Task-aware context management

    Documented

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

    EvidenceKern context managementEvidenceKern durable memory

  • Model-independent runtime control

    Documented

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

    EvidenceKern configuration

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 & statePersistent stateDocumented mechanism, not a performance score.
  • Lifecycle & recoverySession resumeDocumented mechanism, not a performance score.
  • ObservabilityLogs/transcriptsDocumented mechanism, not a performance score.
  • VerificationTool-assistedDocumented mechanism, not a performance score.
  • Governance & permissionsPolicy controlsDocumented mechanism, not a performance score.

Public code audit

2/5public artifacts present
Security policy
Not found
CI workflow
Present at inspected commit
Automated tests
Present at inspected commit
Evaluation assets
Not found
Contributor documentation
Not found

The inspected young project contains 273 files, three workflows, and four test-like paths. It declares MIT in the README but has no root license file, security policy, contributor guide, dedicated evaluation suite, or immutable benchmark record; no product-performance score is inferred.

Inspect commit 8f82a046833128b2bf5f67fdf85a76b35b0fe847, checked 2026-07-28

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 2026-07-28The source establishes the mechanism, not its quality or availability in every mode.
Local models
DocumentedLocal or self-hosted model pathSource · checked 2026-07-28The source establishes the mechanism, not its quality or availability in every mode.
Agent parallelism
DocumentedDelegated or parallel agent workflowSource · checked 2026-07-28The source establishes the mechanism, not its quality or availability in every mode.
Runs without an open UI
DocumentedLong-running background agent service with automation and channel surfacesSource · checked 2026-07-28The 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 surfaceOptional whole-service container deploymentSource · checked 2026-07-28Docker isolates the deployed service; it is not a disposable per-task sandbox.
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 2026-07-28
Kern overviewAgent loop, coding and communication tools, clients, providers, background service, scheduled tasks, memory, and installationOfficial repositoryKern toolsRepository read and mutation, shell, web, memory, scheduled tasks, subagent spawning, status, and cancellationOfficial repositoryKern context managementContext-window trimming, persistent transcripts, compaction, recall injection, and session continuityOfficial repositoryKern durable memoryLocal JSONL history, sqlite-vector recall, notes, segmentation, summaries, persistence, and retrieval limitsOfficial repositoryKern subagentsParallel read-only workers, isolated sessions, model selection, result delivery, lifecycle, and restart limitationsOfficial repositoryKern MCPMCP server configuration, namespaced tool discovery, connections, environment, and trust implicationsOfficial repositoryKern configurationProvider and Ollama routes, token budgets, tool scopes, recall, subagent model, MCP, storage, and environment variablesOfficial repositoryKern Docker deploymentOptional container deployment, mounts, environment, service lifecycle, persistence, and boundary limitationsOfficial repository
View 1 additional sources
Additional first-party evidence1 sources
Kern inspected source treePinned source, CI, tests, desktop and web clients, documentation, license declaration, and absent security policyOfficial repository

Ecosystem context

OpenRouter coding app directory Used to discover Kern; the classification is established from its official source and documentation. Observed 2026-07-28.