git:20260317.cd2f3b3 to git:20260317.d99fae7

23 added, 0 removed. Audit A to B.

# AGENTS.md

Shared instructions for all AI coding agents (Claude Code, Gemini CLI, Codex, etc.) working in this repo.

## Skill Creator Workspaces

When using `/skill-creator` (or any skill evaluation workflow) to audit, test, or iterate on skills, **always** place workspace directories in `.workspaces/` at the repo root — not as siblings to the skill directory. This overrides skill-creator's default behavior of creating `<skill-name>-workspace/` next to the skill.

```

# Use this:

.workspaces/<skill-name>/iteration-1/...

# NOT this:

skills/workflows/<skill-name>-workspace/...

# or:

<skill-name>-workspace/...

```

Both `.workspaces/` and `*-workspace/` are gitignored, but consolidating into `.workspaces/` keeps the repo clean and avoids stray directories.

+ ## Environment Variables for Skills

+ All API keys and credentials for skills live in a **single root `.env` file** at the repo root — never in per-skill `.env` files. This keeps secrets centralized and avoids confusion as the skill count grows.

+ - **`.env`** — The live credentials file (gitignored, never committed)

+ - **`.env.example`** — Template documenting every variable across all skills (committed, no real values)

+ When a skill needs a new env var:

+ 1. Add the variable (with empty value and a comment) to `.env.example`

+ 2. Have the skill's scripts read from the repo root `.env`

+ 3. Document which skill uses it in the `.env.example` section header

+ Scripts should walk up from their location to find the repo root `.env`. Pattern:

+ ```python

+ import pathlib, os

+ _repo_root = pathlib.Path(__file__).resolve().parent

+ while _repo_root != _repo_root.parent:

+ if (_repo_root / ".env.example").exists():

+ break

+ _repo_root = _repo_root.parent

+ # Then load _repo_root / ".env"

+ ```