m0n0x41dCommunity·12 Agent Skills Included

haft

Durable project memory and decision governance for AI coding agents

Gives AI coding agents a durable project memory of problems, decisions, evidence, and specifications. Records why choices were made, detects stale decisions and code drift, and verifies predictions against reality. Keeps binding decisions and execution authority under explicit human control across Claude Code, Codex, and other hosts.

1,387|102

↗

npx skills add m0n0x41d/haft --all -g -y

Available:

Instructs the AI agent on how to apply Haft's reasoning discipline, query the FPF source, persist project memory only when justified, and route binding decisions through explicit human authority.

All Skills in This Repository (12)

Pure Emerald Level Indicators

📦 In Repo

m0n0x41d

h-reason

Frame problems and guide structured reasoning for architecture decisions.

h-onboard

Bootstraps Haft project memory and prepares profile reviews through one onboarding surface.

h-note

Persists lightweight non-binding facts and observations into Haft project memory.

h-status

Reports read-only Haft project status, decision coverage, drift, and spec lifecycle state.

h-spec

Manages typed specification lifecycle, semantic repair, and human-gated spec changes in Haft projects.

h-decide

Routes operator-approved binding choices into validated DecisionRecord artifacts via the haft CLI.

h-frame

Shape engineering problems into structured frames with signals, constraints, and acceptance criteria.

h-verify

Verifies recorded decisions against measured evidence with drift detection and decay tracking.

h-diagnose

Diagnose failures by testing rival hypotheses in parallel with read-only probes.

h-compare

Compare candidates under declared parity and selection policy to produce a non-dominated set.

h-commission

Creates a bounded WorkCommission execution-authority grant from an approved DecisionRecord.

h-explore

Generates 3-5 distinct candidate approaches with weakest links for a current engineering question.

Community

Advanced

Frequently Asked Questions

FAQPage Schema

How to install Haft?▼

Run `npx skills add m0n0x41d/haft --all -g -y` in your terminal to install all skills in this suite globally.

What problem does Haft solve for AI coding agents?▼

It gives your agent a durable project memory: which problems were framed, which options were compared, which decisions the human made, and what evidence still supports them.

How does Haft keep decisions from going stale?▼

The h-verify skill checks recorded decisions against current code, tests, and measurements, flags drift on affected files, and surfaces decisions whose evidence has expired.

Does Haft work with Claude Code and Codex?▼

Yes. Haft installs skills, MCP configuration, and instruction sections for Claude Code and Codex as stable hosts, with experimental adapters for Grok, Cursor, Zed, and others.

Can the AI make binding decisions on its own with Haft?▼

No. Binding decisions and execution commissions always require a direct, unambiguous human request; the agent can only prepare options, comparisons, and evidence.