| name | work |
| description | Switch working context to a specific ecosystem repo |
| user-invocable | true |
| disable-model-invocation | true |
| allowed-tools | Bash, Read, Grep, Glob |
Switch to a specific repo and load its context so the human can immediately give instructions.
The argument $ARGUMENTS is the repo name (e.g., kr-beneish, kr-derivatives, jfia-forensic).
Steps
-
Resolve the path: The repo lives at C:\Users\pon00\Projects/$ARGUMENTS/. If the directory doesn't exist, list available repos and ask the user to pick one.
-
cd to the repo:
cd /c/Users/pon00/Projects/$ARGUMENTS
-
Read CLAUDE.md in the repo root to load project context.
3b. Show Known Gaps. Extract the ## Known Gaps table from CLAUDE.md (if present) and display it. Highlight any Unblocked items that relate to the current task or repo area.
-
Show git status: git status --short — any uncommitted changes.
-
Show recent commits: git log --oneline -5 — what happened last.
-
Run tests (if the project has tests):
- kr-company-registry:
pytest tests/ -v
- kr-trading-calendar:
uv run pytest tests/ -v
- kr-beneish:
uv run pytest tests/ -v
- kr-derivatives:
uv run pytest tests/ -v
- jfia-forensic:
uv run pytest tests/ -v
- kr-enforcement-cases:
uv run pytest tests/ -v
- kr-forensic-core:
uv run pytest tests/ -v
- krff-shell:
uv run pytest tests/ -v
- kr-dart-pipeline:
uv run pytest tests/ -v
- kr-anomaly-scoring:
uv run pytest tests/ -v
- kr-stat-tests:
uv run pytest tests/ -v
- jfia-catalog: no tests
- forensic-accounting-toolkit: no tests
- kr-real-estate: no tests
-
Summarize in 2-3 lines: project purpose, current state, anything needing attention.
Rules
- This skill loads context. Do NOT start working on tasks unless the human asks.
- If tests fail, report the failures clearly — do not try to fix them.
- Keep the summary brief. The human knows these projects.