| name | tutor-project |
| description | Start a tutoring session on the current working directory — convenience wrapper for /tutor-codebase against $PWD. Accepts an optional aspect: /tutor-project (defaults to overview) or /tutor-project 'how requests flow'. Triggers: '/tutor-project', 'tutor me on this project', 'tutor me on what I'm working on', 'teach me this'. |
Tutor Project
A thin wrapper around /tutor-codebase that uses the user's current working directory as the path. Use this when you want to tutor on whatever you're already inside — no need to type a path.
This skill does not duplicate /tutor-codebase's logic. It resolves CWD as the codebase path and follows the same procedure that lives in skills/tutor-codebase/SKILL.md.
When to use this vs /tutor-codebase
/tutor-project — tutor on the directory you're currently in/tutor-codebase <path> — tutor on a different directory (e.g., a repo you've cloned but aren't cd-ed into)
The slug format and aspect handling are identical in both skills.
Procedure
Step 0a — Resolve CWD
ABS_PATH=$(pwd)
Validate it's a directory you can read. (It almost certainly is — you're already in it — but check defensively.)
Step 0b — Resolve aspect
If the user provided an argument (e.g., /tutor-project 'how requests flow'), use that as the aspect.
If not, default to "overview" per locked decision C2.
Step 0c — Compute slug
Per locked decision B2: slug = "{basename-of-PWD}-{aspect-slug}".
REPO_BASENAME=$(basename "$ABS_PATH")
ASPECT_SLUG=$(printf "%s" "$ASPECT" | tr '[:upper:]' '[:lower:]' | sed 's/[^a-z0-9-]/-/g' | sed 's/-\+/-/g; s/^-\|-$//g')
SLUG="${REPO_BASENAME}-${ASPECT_SLUG}"
Step 1 onward — Follow /tutor-codebase's steps
With $ABS_PATH, $ASPECT, and $SLUG resolved, follow steps 0d–6 from /tutor-codebase/SKILL.md exactly. The flow is identical from this point: resume-check → explore the codebase → generate path → charge cycle → calibrate → position on path → hand off to persona.
Do not re-implement the exploration, path generation, calibration, or handoff logic here. Read tutor-codebase/SKILL.md as your source of truth for those steps.
Hard rules
Same hard rules as /tutor-codebase — they all apply unchanged:
- One thing at a time (don't read 30 files in one batch)
- Cap exploration at ~10 file reads
- Don't summarize the codebase during exploration — that's not what the user asked for
- Never reveal canonical solutions during calibration
- Cycle charged exactly once per (codebase, aspect) pair (the existing-topic resume check in /tutor-codebase's Step 0e handles this)
- Refuse on too-broad scope
The full rationale and worked examples for each rule are in tutor-codebase/SKILL.md.
What this skill adds beyond /tutor-codebase .
Strictly speaking, /tutor-codebase . does exactly the same thing. So why have this skill at all?
- Discoverability. "tutor me on what I'm working on" is a common framing; surfacing it as its own slash command makes it findable in the picker.
- Convenience. Skipping the
. argument feels right when the intent is "this thing I'm already in."
- Trigger phrases. This skill matches
tutor me on this project / tutor me on what I'm working on more naturally than /tutor-codebase would.
The implementation cost is small (this short file) and the UX win is real.
