Skip to main content

A lightweight global Markdown spec and decision vault for AI coding agents.

Project description

Agent Spec Vault

agent-spec-vault is a tiny global Markdown vault for AI coding agents.

It stores durable specs, cross-repo contracts, and decisions in plain Markdown under a user-level directory. It is intentionally not a task tracker, Kanban board, daemon, hook system, or project management platform.

Why

Coding agents often need historical context:

  • previous product and architecture specs
  • cross-repository contracts
  • decisions and tradeoffs
  • validation rules that should survive chat resets

Tools such as Beads and Backlog.md are great when you need task tracking. This tool keeps the narrower part: Markdown-native, agent-readable specs and decisions that are global rather than tied to one repository.

Install

For most Python environments:

python3 -m pip install agent-spec-vault
specv init

If your Python installation blocks global CLI installs, or you prefer isolated command-line tools:

pipx install agent-spec-vault
specv init

Inside a virtual environment:

python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install agent-spec-vault
specv init

Unreleased development builds can still be installed from GitHub:

pipx install git+https://github.com/xixifast/agent-spec-vault.git

For local development from a checkout:

python3 -m pip install -e .

Or run from a checkout without installing:

python3 -m specv --help

Quick Start

specv init
specv new "Quality analysis display contract" \
  --repos lippi-smart-customer,aics-web-repos,lippi-dingagent-graph-engine \
  --tags quality-analysis,contract
specv decision "Use Markdown vault instead of Beads for historical specs" \
  --tags tooling,specs
specv list
specv search "quality analysis"
specv show spec-20260703-quality-analysis-display-contract
specv index --print
specv prime --repo lippi-smart-customer

Default vault location:

~/.agents/spec-vault

Override it with:

SPECV_HOME=/path/to/vault specv list
specv --home /path/to/vault list

Data Model

Everything is ordinary Markdown:

~/.agents/spec-vault/
  README.md
  index.jsonl
  templates/
    spec.md
    decision.md
  specs/
    2026-07-03-quality-analysis-display-contract.md
  decisions/
    2026-07-03-use-markdown-vault.md

Each document has a small frontmatter block:

---
id: spec-20260703-quality-analysis-display-contract
kind: spec
title: Quality analysis display contract
status: active
created: 2026-07-03
updated: 2026-07-03
repos:
  - lippi-smart-customer
  - aics-web-repos
tags:
  - quality-analysis
  - contract
---

specv index writes index.jsonl, a compact surface for scripts and agents.

Commands

Command Purpose
specv init Create the vault directories and templates
specv new <title> Create a new spec
specv decision <title> Create a decision note
specv list List specs and decisions
specv search <query> Search metadata and Markdown body
specv show <id> Print one document
specv index Regenerate index.jsonl
specv prime Print an agent-friendly summary
specv path Print the vault path

Suggested Agent Rule

Add a small rule to your global agent instructions:

When a task needs historical specs, cross-repo contracts, or prior design
decisions, search ~/.agents/spec-vault first with specv search or rg. When asked
to preserve a durable spec or decision, write it there instead of creating an
ad hoc file in a business repository.

Scope

This project does:

  • global spec and decision storage
  • Markdown-native files
  • frontmatter metadata for repos, tags, status, and dates
  • JSONL index generation
  • agent-friendly summary output

This project does not:

  • track task claim/close state
  • manage dependencies or blockers
  • run a daemon or hooks
  • replace GitHub Issues, Beads, or Backlog.md
  • write into business repositories by default

Development

python3 -m unittest discover -s tests -p 'test_*.py'
python3 -m specv --home /tmp/specv-demo init

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

agent_spec_vault-0.1.1.tar.gz (9.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

agent_spec_vault-0.1.1-py3-none-any.whl (9.6 kB view details)

Uploaded Python 3

File details

Details for the file agent_spec_vault-0.1.1.tar.gz.

File metadata

  • Download URL: agent_spec_vault-0.1.1.tar.gz
  • Upload date:
  • Size: 9.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for agent_spec_vault-0.1.1.tar.gz
Algorithm Hash digest
SHA256 99a712b8e529a868d6ee85caea7d9594f2394e457825e728b7a64e49fa89f39b
MD5 9de1e1a2c585851c322a0698dd21da5d
BLAKE2b-256 59463532d6ca161381d3b70c09a434c4bdc7793b2bc0df1b7b8187b933b6a620

See more details on using hashes here.

Provenance

The following attestation bundles were made for agent_spec_vault-0.1.1.tar.gz:

Publisher: publish-to-pypi.yml on xixifast/agent-spec-vault

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file agent_spec_vault-0.1.1-py3-none-any.whl.

File metadata

File hashes

Hashes for agent_spec_vault-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 181f0656f25fed4d68cf8c5eda46d5aa2b935789e5c01afc0506d3b479b36100
MD5 d3124e1a21d168ee0817a750f306e616
BLAKE2b-256 322f6b3a586fced12b1868187aa2c9ff3f385085c42eff88163769e32fc43f88

See more details on using hashes here.

Provenance

The following attestation bundles were made for agent_spec_vault-0.1.1-py3-none-any.whl:

Publisher: publish-to-pypi.yml on xixifast/agent-spec-vault

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page