Micron Document
🎖️GitЯра🎖️


Displaying Rendered • View rawDownload

.specify/extensions/agent-governance/templates/agent-governance-template.md bd2863243bab6eb213401d949839a2bc74dde7e2 (bd286324) Text, 1.47 KB

Agent Governance Cache

<!--
Sync Impact Report
-->

Final Output

• active agent platform governance file
• managed T383838SPECKIT GOVERNANCE section
• cache: internal

Directory Governance

• Responsibility: one primary purpose per directory.
• Depth: 2.
• Coverage: include visible, hidden, generated, cache, config/env, tool, and agent directories.
• Mixed concerns: follow existing repo convention or split responsibility.
• Change impact: review linked code, tests, docs, config/env, data, assets, generated files, and tool outputs; update only when in scope and authorized.

Scope

• agent collaboration rules
• tool and MCP permissions
• write boundaries
• skill invocation contracts
• project governance: external

Write Boundaries

• Scope: active task only
• Preserve: user-authored edits
• Preserve managed markers verbatim: T383838<!-- SPECKIT GOVERNANCE START --> and T383838<!-- SPECKIT GOVERNANCE END -->
• Protected files: implementation, CI, MCP config, secrets, permissions, tool settings
• Protected-file writes: explicit user request only

Skill Contract

• Repository-local skill specs should declare purpose, trigger, allowed read paths, allowed write paths, forbidden paths, outputs, and validation command.

MCP Policy

• Default: read-only
• Mutation: explicit user intent
• External writes: target, action, result
• Secrets: never log, never write

Validation

• changed files
• commands run
• tests/validation result
• unresolved risks

Served by rngit 1.5.2 - Generated in 0.07s