Repository intelligence

FairladyZ625/harness-anything

A self-evolving harness for software projects that captures AI agent decisions, failures, facts, and reviews as a durable git ledger to enforce structural completion gates.

CLOUDM0N decision

Trust FAIL · 35/100

Good fit if

Teams utilizing AI coding agents who need to preserve developer-agent reasoning and decision histories directly inside their codebases

Watch out for

Maintains no backward compatibility across generation changes, preventing in-place ledger upgrades and requiring full-history replay migrations

Practical intelligence

What matters before you adopt it

Decision first · detail on demand

Problem it solves

The loss of context and engineering momentum when disposable agent runs evaporate into chat transcripts, causing teams to reopen settled decisions, repeat past mistakes, and accept completion claims without structured verification.

Best for

Teams utilizing AI coding agents who need to preserve developer-agent reasoning and decision histories directly inside their codebases

Repositories requiring programmatic completion gates to ensure that task completion is a verified state rather than a confidence claim

Main trade-offs

Maintains no backward compatibility across generation changes, preventing in-place ledger upgrades and requiring full-history replay migrations

Why it stands out

Stores all project memory (tasks, decisions, facts) as simple, accessible Markdown files in a nested git directory

Provides a six-field Submission Packet that gives human reviewers a traceable claim and clean entry points for inspection

Does not require CI or a minimum number of recorded facts universally to function, maintaining a flexible core kernel

Trust & CVEs

Security evidence without the noise

Trust remains a decision signal; CVEs and scanner evidence explain what is driving the risk.

Security findings

CLOUDM0N scanner findings

Critical

High

Medium

Low

View trust evidence & security findings▼

Why this score

No trust rationale was stored for this scan.

CLOUDM0N findings

CRITICAL

Private-key material appears to be committed in executable/config scope. 2 sample match(es) found.

MEDIUM

The project can spawn operating-system processes; review command construction and input handling. 5 sample match(es) found.

MEDIUM

Broad permission or elevated-command pattern detected. 1 sample match(es) found.

Architecture from code15 modules · 26 edges▼

Structural evidence

Modules and dependency edges extracted from repository code. This is code evidence, not README inference.

Code files

1501

Modules

Dependency edges

Core modules

(root)

2 files

packages/adapters

3 files

packages/application

29 files

packages/cli

86 files

packages/daemon

397 files

packages/gui

365 files

packages/kernel

323 files

packages/preset

43 files

scripts

1 files

tools

tool implementations

252 files

Dependency flow

(root)→tools

packages/application→packages/kernel

packages/application→tools

packages/cli→packages/application

packages/cli→packages/daemon

packages/cli→packages/kernel

packages/cli→tools

packages/daemon→packages/adapters

packages/daemon→packages/application

packages/daemon→packages/cli

packages/daemon→packages/gui

packages/daemon→packages/kernel

packages/daemon→packages/preset

packages/daemon→tools

Detected languages

TypeScript · JavaScript · Shell · Ruby

Detected frameworks

—

Architecture evidence details

flowchart TD
    %% harness-anything — high-level architecture (DRAFT, refine me)
    n0["(root) · 2 files"]
    n1["packages/adapters · 3 files"]
    n2["packages/application · 29 files"]
    n3["packages/cli · 86 files"]
    n4["packages/daemon · 397 files"]
    n5["packages/gui · 365 files"]
    n6["packages/kernel · 323 files"]
    n7["packages/preset · 43 files"]
    n8["scripts · scripts · 1 file"]
    n9["tools · tool implementations · 252 files"]
    n0 --> n9
    n2 --> n6
    n2 --> n9
    n3 --> n2
    n3 --> n4
    n3 --> n6
    n3 --> n9
    n4 --> n1
    n4 --> n2
    n4 --> n3
    n4 --> n5
    n4 --> n6
    n4 --> n7
    n4 --> n9
    n5 --> n2
    n5 --> n4
    n5 --> n6
    n5 --> n9
    n6 --> n9
    n7 --> n6
    n9 --> n0
    n9 --> n2
    n9 --> n3
    n9 --> n4
    n9 --> n5
    n9 --> n6
    class n8 infra
    class n9 shared
    classDef infra fill:#b35c00,color:#ffffff,stroke:#8f4a00
    classDef shared fill:#79706e,color:#ffffff,stroke:#5d5654

Evidence, security & integrations▼

Integrations

Node.jsGit

Security notes

• Ledger content, decisions, and tasks are kept fully local and private inside a nested git repository structure.

• Gating architectures systematically regulate which state transitions are permitted during task cycles.

Still unknown

• The README does not detail the exact database schema or markdown structure of the queryable ledger projection.

• No metrics or performance data regarding daemon CPU/RAM resource footprint are provided in the text.

• The explicit fields of the six-field Submission Packet are not detailed in the source documentation.

Adoption guidance▼

Adopt if

+ You rely heavily on AI coding agents and want to systematically capture their context, rationale, and failures to continuously harden your repository rules.

+ You want a copyleft-compliant (AGPL-3.0-or-later) project memory and governance standard for your engineering workflows.

Avoid if

– You cannot support Node.js 24+ or do not want to manage a local, git-nested file database inside your codebase.

– You require long-term legacy format stability and cannot accommodate ledger format migration workflows.

How it works & getting started▼

How it works

1.The user sets up Harness Anything within their repository from a source checkout or via an agent installation skill

2.Agent activities, decisions, and observations are logged as Markdown files inside a private nested git ledger to preserve historical context

3.The platform compiles these records into a rebuildable, queryable projection of the repository's history

4.Reviewers audit submissions through a standardized six-field Submission Packet to verify correctness

5.Upon approval, the system evaluates the state transitions against preset contracts to officially transition the task to a verified 'done' state

Getting started

→ Obtain a source checkout of the Harness Anything repository.

→ Ensure your environment is running Node.js 24+.

→ Run the 30-second smoke demo to compile the CLI, establish a temporary workspace, execute a sample task loop, and generate initial ledger records.

→ For new project adoption, instruct your coding agent to execute the 'harness-install' skill.

Agent handoff

Use with any agent

JSON API

Alternatives

Nearby repositories worth comparing before adoption.

Compare top options →

anything-llm

Mintplex-Labs/anything-llm

Fit

Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent experience

Trust REVIEW · 55

Compare →