AI Agent

Technical Decisions

Maintains a decision log for technical architecture choices like folder structures, release packaging, and naming conventions. Records context, alternatives, consequences for Codex project continuity.

Git

Bash

Powershell

documentation

From pm-skills

Install

Run in your terminal

npx claudepluginhub product-on-purpose/pm-skills

Details

Tool AccessAll tools

RequirementsPower tools

Agent Content

Similar Agents

chief-of-staff

6 tools

Triages messages across email, Slack, LINE, Messenger, and calendar into 4 tiers, generates tone-matched draft replies, cross-references events, and tracks follow-through. Delegate for multi-channel inbox workflows.

everything-claude-code

139.9k

architect

3 tools

Software architecture specialist for system design, scalability, and technical decision-making. Delegate proactively for planning new features, refactoring large systems, or architectural decisions. Restricted to read/search tools.

everything-claude-code

139.9k

build-error-resolver

6 tools

Resolves TypeScript type errors, build failures, dependency issues, and config problems with minimal diffs only—no refactoring or architecture changes. Use proactively on build errors for quick fixes.

everything-claude-code

139.9k

Stats

Stars122

Forks18

Last CommitMar 23, 2026

Actions

View Source View Plugin View on GitHub View README

Technical Decisions

2026-01-22: Initialize Codex agent folder structure

Status: Accepted

Context: Codex sessions need a consistent place for context, decisions, and logs.

Decision: Create standard agent artifacts in AGENTS/codex/:

CONTEXT.md
DECISIONS.md
SESSION-LOG/ (with .gitkeep)

Alternatives Considered:

Keep only PLANNING/ docs
Store session notes in _NOTES/

Consequences:

Consistent Codex session continuity
Clear separation between working artifacts and project docs

Add new decisions above this line

2026-03-22: Implement the internal-doc packaging boundary in the release packagers

Status: Accepted

Context: The F-05 staging review exposed a broader release bug: scripts/build-release.sh and scripts/build-release.ps1 staged the entire docs/ tree, which caused docs/internal/** governance material to ship in the published ZIP.

Decision: Keep docs/internal/** tracked in-repo, but remove stage/docs/internal in both release packagers before zipping. Do not move the directory tree and do not broaden M-16 into a full public-doc allowlist redesign.

Consequences:

M-16 remains an isolated packaging fix instead of a repo restructuring effort.
Published ZIPs preserve public docs such as docs/releases/** while excluding internal governance docs.
Any future release-packaging refactor should preserve this boundary unless the team intentionally adopts a stricter allowlist model.

2026-02-13: Move Codex workspace path to AGENTS/codex

Status: Accepted

Context: AGENTS/codex-5.2 encoded model version in the path and created path churn. The desired stable working path is AGENTS/codex.

Decision: Adopt AGENTS/codex as the canonical Codex workspace path for context, decisions, planning, and session handoff artifacts.

Consequences:

Stable path independent of model version changes.
Less friction when rotating models.
Existing scripts/docs should reference AGENTS/codex going forward.

2026-02-13: Flat skill naming compliance

Status: Accepted

Context: Flat skill structures under agentskills spec must use hyphen-only names; underscores are invalid. Phase prefixes may be present in folder names, but MCP tool IDs should stay short and stable.

Decision: If skills are flattened, enforce hyphen-only folder names; keep MCP tool IDs unchanged (no phase prefixes), carry phase/order in metadata instead.

Consequences:

Prevents invalid skill names and marketplace discovery issues.
Avoids breaking existing MCP integrations.
Requires loader metadata mapping for phase context.