| name | consult-pro |
| description | Build, refine, and synthesize ChatGPT Pro consultation prompts. Use when the user asks for a paste-ready Pro prompt, wants Pro to audit, review, or critique a repo or plan, needs a follow-up after a Pro response, wants Pro feedback checked against local evidence, or wants a manual Pro consultation loop without browser monitoring. |
Consult Pro
Purpose
Use Pro as an outside high-context critic, not as an oracle. Produce prompts
that give it enough source truth, constraints, and deliverables to find real
blind spots, then validate its response against local evidence before acting.
Boundary
This skill owns the manual Pro consultation loop: prompt packaging, follow-up,
and response synthesis. It does not own the underlying repo decision, skill
design, or instruction rewrite; use the relevant project skill or
$writing-great-skills for that work after Pro's claims are checked.
Workflow
- Clarify the consultation lane:
- First pass: create a paste-ready prompt for Pro.
- Follow-up: summarize local changes, user reactions, and remaining
questions for another Pro pass.
- Synthesis: evaluate a Pro response and turn it into adopt/reject/defer
actions.
- Gather current facts before drafting:
- Repo/project path, branch, commit, PR/issue links, current diff, and any
relevant local findings.
- What Pro can inspect directly, such as GitHub connector access to a branch.
- What Pro cannot know, such as local-only files, unpushed changes, hidden
user preferences, or recent command output.
- Draft the prompt with a concrete objective, evidence routes, explicit
constraints, and requested output shape.
- Ask Pro for convergence-oriented criticism: findings, tradeoffs, and a clear
end-state recommendation, not an endless review loop.
- When Pro responds, validate material claims against current repo truth before
changing files.
Resources
- Read
references/prompt-guide.md when creating or revising a Pro prompt.
- Read
references/templates.md when the user wants a paste-ready prompt.
- Read
references/response-synthesis.md when processing Pro's response.
Output Rules
- Return prompts in fenced Markdown blocks so the user can paste them manually.
- Include paths, branch names, SHAs, dates, and links when known.
- Prefer direct asks such as "find remaining high-ROI infra cruft" over vague
asks such as "give feedback."
- Ask Pro to distinguish
adopt, purge, defer, and reject when the output
affects repo changes.
- Skip praise, apologies, and model-identity boilerplate; ask for
evidence-backed findings and a decision-oriented recommendation.
