Skip to content

joedaws/agent-techniques / .claude/skills/preflight

preflight

Verified

Use BEFORE implementing code in a project set up with this kit — loads the language rules for detected manifests, the project's CLAUDE.md, and the active task's prior handoff annotations. Replaces the heavy context that would otherwise be re-injected on every prompt. Trigger when the user says "preflight", "load the rules", "ground yourself", or before starting a fresh implementation session.

Preflight — load rules and grounding context

You are preparing to implement code. Load the project's rules and grounding context so you have the right constraints in mind. This replaces the heavy context that would otherwise be injected on every prompt.

1. Detect active languages and load rules

Check for project manifests and load the corresponding rule files:

ManifestLanguageRule file
package.jsonJavaScript (+ React if a dep).claude/rules/languages/javascript.md (+ javascript-react.md)
tsconfig.jsonTypeScript (+ React if a dep).claude/rules/languages/typescript.md (+ typescript-react.md)
pyproject.toml / requirements.txtPython.claude/rules/languages/python.md

Only read rule files for languages actually detected in this project. shell.md has no manifest to detect — load it when editing shell scripts (*.sh).

Also read .claude/rules/quality.md and .claude/rules/rigor.md — these apply regardless of language.

2. Read project-specific and task context

Read CLAUDE.md if present, for project-specific conventions.

Read the active task's prior handoff annotations so a resuming agent picks up where the last one left off:

task project:<repo> +ACTIVE list
task <id> information

3. Scan project tree

Build spatial awareness of the project (max depth 3, max 50 entries):

ls -R --max-depth=3 | head -50

4. Check dependency versions

Read the primary manifest file (e.g., package.json, pyproject.toml) to ground yourself on actual dependency versions. Do not guess versions.

5. Print summary

Print a brief summary of what was loaded:

Preflight loaded:
  - Language rules: TypeScript, TypeScript-React
  - Quality and rigor rules
  - CLAUDE.md read
  - Active task handoff read
  - Project tree scanned (N entries)
  - Dependencies: package.json (N deps)

You are now grounded. Proceed with implementation.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 465kUpdated Claude CodeCodexVerifiedInstall
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 465kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 318kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 318kUpdated Claude CodeCodexVerifiedInstall