Skip to content

Tags: gl0bal01/contract-agents

Tags

v3.0.0

Toggle v3.0.0's commit message
v3.0.0 — Quality over count

125 → 52 agents. 10 → 8 divisions. Every agent has real domain rules.
contract-orchestrator hardened with mandatory delegation enforcement.
Makefile now actually implements SCOPE/LOCAL_PATH/DIVISION per README.

See CHANGELOG.md for the full list.

v1.3.1

Toggle v1.3.1's commit message
v1.3.1 - Stable release

Fixes agent count inconsistencies, renames the orchestrator,
and replaces vague marketing claims with verifiable language.

All 121 agents are human-guided and contract-governed.
Stable for human use.

v1.5.0

Toggle v1.5.0's commit message
docs: add teasing tagline and badges to README

v1.3.0

Toggle v1.3.0's commit message
v1.3.0 - Contract v2.0 with universal language

v1.2.0

Toggle v1.2.0's commit message
v1.2.0 - Simplified Makefile, cleaner repo structure

v1.1.0

Toggle v1.1.0's commit message
Release v1.1.0 - 20 new agents including architecture workflow, cyber…

…security, data & AI, and business intelligence

v1.0.2

Toggle v1.0.2's commit message
Contract-Agents v1.0.2 - Ultra-Slim

Maximal efficiency achieved:
- Shared structure moved to AGENTS_CONTRACT.md
- Agents reduced to 11-44 lines
- Average ~25 lines per agent
- All emojis and verbose sections removed
- Total: 101 agents, 2823 lines

v1.0.1

Toggle v1.0.1's commit message
Contract-Agents v1.0.1 - Black Box Style

Massive token reduction through black-box architecture:
- All agents rewritten in clean, consistent format
- No emojis, no verbose sections
- Average 36 lines per agent
- ~89% token reduction (23,854 lines removed)
- Protocol-based: Role → Session Start → Protocol → Domain Rules → Output

v1.0.0

Toggle v1.0.0's commit message
Contract-Agents v1.0.0

Professional AI agents governed by shared contract.

Features:
- 101 agents across 10 divisions
- Flattened structure with division prefixes (eng-*, test-*, design-*, etc.)
- AGENTS_CONTRACT.md with shared rules
- Evidence requirements (3+ file:line citations)
- Approval gates for risky operations
- Self-improvement loop via tasks/lessons.md

Based on agency-agents by @msitarzewski and black-box-architecture by @gl0bal01