HarnessMatch
Menu

Extensible harness

ForgeCode

One terminal coding harness for many models, scripts, and specialist agents.

Status
Active
License
Apache-2.0
Evidence
Documented + code-verifiable, 24 sources
Product record checked
2026-07-27
Compare harnessesInspect sources

At a glance

ForgeCode is an open terminal coding harness that works interactively or as a one-shot command. The audited source defines 47 provider entries spanning hosted APIs, subscription access, local endpoints, and a context engine, plus MCP, skills and delegated agents. It runs on your machine by default; its `--sandbox` option separates Git work but does not contain commands.

Good choice if

  • Vibe coders who want to keep one terminal tool while switching among subscriptions, API providers, and local models
  • People who want an interactive coding session now and repeatable one-shot scripts later
  • Power users who want reusable skills, MCP tools, semantic code search, and delegated research agents

Check before choosing

  • The `--sandbox` flag only creates a separate Git worktree and branch; commands still run on the host
  • Restricted permissions are off by default, the generated policy allows all built-in actions, and MCP tools bypass it
  • Optional semantic search uploads source chunks and embeddings to ForgeCode Services; logging out does not delete indexed data
See 1 more considerations
  • Browser work requires an external MCP server, and file undo restores one path's last snapshot rather than a project checkpoint

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.

    EvidenceAudited ForgeCode readme

  • Repository tool execution

    Documented

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

    EvidenceBuilt-in implementation agent

  • Task-aware context management

    Documented

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

    EvidenceBuilt-in agent roles

  • Model-independent runtime control

    Documented

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

    EvidencePermission policy

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 & isolationWorkspace isolationDocumented 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.
  • ObservabilityLogs/transcriptsDocumented mechanism, not a performance score.
  • VerificationTool-assistedDocumented mechanism, not a performance score.
  • Governance & permissionsHost accessDocumented mechanism, not a performance score.

Public code audit

4/5public artifacts present
Security policy
Not found
CI workflow
Present at inspected commit
Automated tests
Present at inspected commit
Evaluation assets
Present at inspected commit
Contributor documentation
Present at inspected commit

No repository security policy was found, and project evaluation assets are not independent evidence. The contribution guide is stale and calls the code proprietary despite the root Apache-2.0 license.

Inspect commit 1ca089a52fd2, 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
DocumentedDelegated or parallel agent workflowSource · checked 2026-07-27The source establishes the mechanism, not its quality or availability in every mode.
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
Not documentedNo first-class support established by the current recordAbsence of current documentation is not proof that the capability is impossible.
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
Installation and setupPlain-language product identity, installation, subscriptions, cloud and local models, and provider loginOfficial docsAudited ForgeCode readmeInteractive, one-shot, and Zsh modes; conversations, agents, skills, providers, MCP, and semantic searchOfficial repositoryLatest verified releaseRelease recency, signed release commit, current provider fixes, and versioned distribution artifactsOfficial announcementApache-2.0 licenseRoot source-code license at the audited repository commitOfficial repositoryCLI implementationInteractive default, one-shot prompt and stdin, conversation resume and export, logs, agents, and worktree flagOfficial repositoryZsh workflowOptional shell integration, prompt routing, agent switching, conversation commands, and diagnosticsOfficial docsCustom and local providersSelf-hosted endpoints, enterprise gateways, localhost examples, provider overrides, and session selectionOfficial docsBuilt-in provider registryForty-seven audited provider definitions including subscriptions, hosted APIs, local runtimes, and compatible endpointsOfficial repository
View 16 additional sources
Additional first-party evidence16 sources
MCP integrationOptional project and user MCP servers, transports, trust, OAuth, CLI management, and external browser examplesOfficial docsReusable skillsBuilt-in and custom SKILL.md workflows, loading precedence, resources, discovery, and invocationOfficial docsCustom agentsAgent definitions, model selection, tool restrictions, MCP tool globs, and agent-as-tool supportOfficial docsBuilt-in agent rolesForge, Muse, and Sage roles, write access, task selection, context continuity, and version-control guidanceOfficial docsDelegated task toolSubprocess agents, isolated task context, concurrent delegation, agent selection, results, and resume identifiersOfficial repositoryBuilt-in implementation agentBroad host tools, MCP and task access, code-edit workflow, tool-assisted compilation, tests, and verification instructionsOfficial repositoryPermission policyRestricted-mode prerequisite, allow, deny and confirm rules, allow-all generated policy, exemptions, and MCP bypassOfficial docsConfiguration defaultsRestricted mode off by default, tool and request limits, retries, semantic search, HTTP, and compaction settingsOfficial docsGenerated permission defaultsAllow-all defaults for built-in reads, writes, commands, and network fetches when the policy file is generatedOfficial repositoryGit worktree implementationThe sandbox-named option creates or reuses a Git worktree and branch without an operating-system boundaryOfficial repositoryForgeCode ServicesOptional workspace login, sync, semantic search, stored source chunks and embeddings, ignore rules, and deletionOfficial docsService privacy policySource-file and embedding storage for semantic search, direct model-provider routing, and service data handlingOfficial docsFile snapshots and undoPer-path content snapshots used for file-level undo rather than multi-file workspace or conversation checkpointsOfficial repositoryLocal log commandsListing, tailing, and selecting local ForgeCode log files for operational inspectionOfficial repositoryProject evaluation frameworkFirst-party task runner, parallelism, timeouts, data-driven cases, validations, and debug artifacts; not independent evidenceOfficial repositoryContinuous integration workflowCross-platform build and test automation, coverage, performance checks, and release jobs at the audited commitOfficial repository