pre-flight
/pre-flight
Pre-development briefing for a feature or bugfix the user already has in mind. Recons the repo and reflects the planned change against what exists, ending in a confirm gate before any code. Use before starting a feature or bugfix, or when the user says "pre-flight", "brief me on this feature", "recap before I build". Differs from grill-me (which interrogates the user about their plan) — pre-flight interrogates the repo; differs from whats-next (which proposes features) — here the user brings the feature.
- Author
- asrafilll ↗
- Repository
- skills ↗
- Licence
- MIT
- Last repository change
- 13.11.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 13.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Pre-development briefing for a feature or bugfix the user already has in mind. Recons the repo and reflects the planned change against what exists, ending in a confirm gate before any code. Use before starting a feature or bugfix, or when the user says "pre-flight", "brief me on this feature", "recap before I build". Differs from grill-me (which interrogates the user about their plan) — pre-flight interrogates the repo; differs from whats-next (which proposes features) — here the user brings the feature.
Triggers from the description
Original author description
Pre-development briefing for a feature or bugfix the user already has in mind. Recons the repo and reflects the planned change against what exists, ending in a confirm gate before any code. Use before starting a feature or bugfix, or when the user says "pre-flight", "brief me on this feature", "recap before I build". Differs from grill-me (which interrogates the user about their plan) — pre-flight interrogates the repo; differs from whats-next (which proposes features) — here the user brings the feature.
How it works
- Step 1 — Restate the request
- Step 2 — Recon the repo
- Step 3 — Reflect the change against the repo
- Step 4 — Output the briefing
- Step 5 — Confirm gate
- Rules
SKILL.md preview
--- name: pre-flight description: Pre-development briefing for a feature or bugfix the user already has in mind. Recons the repo and reflects the planned change against what exists, ending in a confirm gate before any code. Use before starting a feature or bugfix, or when the user says "pre-flight", "brief me on this feature", "recap before I build". Differs from grill-me (which interrogates the user about their plan) — pre-flight interrogates the repo; differs from whats-next (which proposes features) — here the user brings the feature. --- # Pre-Flight Produce a pre-development briefing for a feature or bugfix BEFORE any code is written. The goal: user reads the recap, confirms your understanding matches theirs, then development starts with shared context. This skill is read-only — never edit code while running it. Input: the user's description of the feature or fix (from the skill arguments or the conversation). If no description was given, ask for one first. ## Step 1 — Restate the request Rewrite the user's request in one short paragraph of plain words. If it is a bugfix, state the suspected broken behavior. Do not add scope the user did not mention. ## Step 2 — Recon the repo Map the area of the codebase the change touches. Prefer Lexa if available: ```bash lexa status # index missing/stale -> lexa index . lexa brief "<the task, with symbols/paths/domain keywords>" --max 8 lexa symbol-search <RelevantSymbol> --max 10 lexa text-search "<domain keyword>" --scope --compact --max 20 lexa text-search "<keyword>" --compact --path-glob "apps/api/**" # scope with --path-glob, not --path-prefix lexa word-refs <ExactSymbol> lexa outline <key-file> lexa trace-deps <key-file> ``` Flag gotchas: use `--max`, not `--max-results`; `text-search` scopes with `--path-glob` only. If a flag errors, run `lexa <cmd> --help` and adapt instead of retrying. If Lexa is unavailable, fall back to normal exploration: README, package manifest, route definitions, models/schema, the directories whose names match the feature's domain. Collect: which existing features/modules relate to this change, which files implement them, and how data currently flows through that area. ## Step 3 — Reflect the change against the repo Answer, from evidence found in Step 2: - What does the app do **today** in this area? - What will the change **add or alter**? New capability, changed behavior, or repaired behavior? - Which existing features does it touch, extend, or risk breaking? - What must be true for the plan to work (existing endpoints, schema fields, auth rules, third-party services)? Verify each in the code where possible; whatever you cannot verify becomes an assumption or open question. ## Step 4 — Output the briefing Print the recap in chat using exactly this structure. Cite real file paths from recon. Keep each section tight — this is a briefing, not a design doc. ```markdown ## Pre-Flight: <feature/fix name> **What** One paragraph: the feature/fix in plain words. **Current state** What the app does today in this area. Reference the modules/files that implement it. **What changes** Bullet list: capability added, behavior altered, or defect repaired. Note which existing features are touched or at risk. **Touched areas** | Area | Files/Modules | Change type (new / modify / risk) | **Assumptions** Numbered list. Things that must hold true, each marked ✅ verified in code or ⚠️ unverified. **Business flow impact** Before: how the process/flow works now for the user/business. After: how it works once this ships. **Expected outcomes** Bullet list: concrete results once done — what the user/business can do that they could not before, or what stops going wrong. **Open questions** Anything unresolved that could change the plan. Empty section is fine. ``` ## Step 5 — Confirm gate End with one question: does this match the user's intent? Wait for confirmation or corrections before any development starts. If the user corrects something, update the briefing and re-confirm. Only after confirmation does implementation begin — and implementation is outside this skill. ## Rules - Read-only: no file edits, no commits, no scaffolding while this skill runs. - Every claim about the current state must come from the repo, not from memory of similar apps. If you did not find it, mark it ⚠️ unverified. - Do not inflate scope. If the recap reveals the feature is bigger than the user's description, say so in Open questions instead of silently expanding "What changes". - Differs from grill-me: grill-me interrogates the user about their plan; pre-flight interrogates the repo and reflects the plan against it.
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · asrafilll/skills | 100 | MIT |
| 8778fkv694-hub/pre-flight | 90 | MIT |
Database snapshot: 13.11.2026 · package version cafc765a5059 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100whats-next
Product ideation from repo and market analysis — the AI proposes what to build next, backed by an evidence-based SWOT. Use when the user asks "what's next", "what should we build next", wants feature ideas, or a product roadmap suggestion for an existing repo. Differs from… - 90ship
Pre-flight check before opening a PR: runs the project's typecheck, lint and tests, reviews the diff for bugs with the code-reviewer agent, then drafts the PR description. Use when the user says they're ready to ship, open a PR, or wrap up a change. - 96test-driven-development
Use when implementing any feature or bugfix, before writing implementation code - 96test-driven-development
Use when implementing any feature or bugfix, before writing implementation code - 96test-driven-development
Use when implementing any feature or bugfix, before writing implementation code - 96test-driven-development
Use when implementing any feature or bugfix, before writing implementation code