Expand a shell into a full implementation plan. Verifies Consumes against the current codebase, runs a fresh pattern survey, escalates open questions, and fills in concrete file references and verification.
SkillScore55Maintained
Automatic
◆Coding pack
Last verified 8/21/2026
What it changes
Passed 9 of 16 test cases with the skill, and 4 of 16 without it.
Without skill23%
With skill55%
Checks it passed that the model alone missed
The assistant omits depends_on from the plan frontmatter.
The assistant sets status to draft in the plan frontmatter.
The assistant omits commit instructions from the plan file.
and 7 more in the full report
7 fixed5 passed either way10 still failing0 broke
Pass-rate lift
+32 pp
Tokens
+161%
Agent turns
0%
Response time
-52%
Model gemini-3.6-flashMeasured 8/21/2026v412 of 22 cases passed with skill · 16 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 7
Prompt
After drafting `.turbo/plans/cache-layer.md` and verifying it against `.turbo/shells/cache-layer.md`, should I run the `$review-plan` skill to double-check the work?
What was checked, without and with the skill
Checked
Without skill
With skill
The assistant refrains from invoking review skills during shell expansion.
Missed
Passed
case-18 · gemini-3.6-flash · 8/21/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md181 linessingle file
Expand Shell
Expand a shell into a full implementation plan. The shell's Context, Produces, Consumes, Covers, and high-level Implementation Steps are authoritative. Expansion adds a pattern survey, concrete references, and verification, writes the plan to .turbo/plans/<shell-slug>.md, and deletes the source shell once the plan is in place.
Task Tracking
Use update_plan to track each step, restating any remaining steps of a parent workflow alongside them:
Load the shell and verify consumes
Run $survey-patterns skill (shell-focused)
Escalate the shell's open questions
Write the plan
Verify the plan against the shell
Present summary and gate
Delete the shell
Step 1: Load the Shell and Verify Consumes
Determine which shell to expand:
Explicit path — If a file path was passed, use it
Single candidate — Glob .turbo/shells/*.md and filter to shells whose depends_on are all satisfied (see satisfaction check below). If exactly one match, use it
Multiple candidates — If multiple matches, use request_user_input to let the user choose
Nothing found — If no shells exist in .turbo/shells/, say so and stop
A depends_on entry is satisfied when .turbo/plans/<dep-slug>.md exists with status: done in its frontmatter.
Read the shell file. Parse the YAML frontmatter:
spec (source spec path)
depends_on (list of shell slugs that must already be implemented)
Parse these body fields:
Title (from the # Plan: heading)
Context (the why)
Produces (artifacts this shell creates)
Consumes (dependencies this shell requires)
Covers Spec Requirements (the spec sections this shell implements)
Implementation Steps (High-Level) (named tasks without file paths)