Skip to content

Arnie016/codex-prompt-templates / skills/low-token-context

low-token-context

Verified

Build compact Codex task context and avoid token blowups. Use when work spans large repos, repeated sessions, long logs, many files, expensive models, or the user mentions limits, context, tokens, compression, summaries, or keeping Codex fast.

Low-Token Context

Goal: get enough evidence to act without loading the whole project.

Workflow

  1. Run a compact repo snapshot when the repo is unfamiliar: scripts/repo_snapshot.sh <repo>.
  2. Estimate heavy files before reading broadly: scripts/token_budget.sh <repo>.
  3. Search first with rg, git diff, package manifests, and exact symbols.
  4. Read only the files needed to state:
    • edit surface
    • invariants
    • verification command
    • known risks
  5. Write or update a small task brief if the work will continue across turns.

Task Brief Contract

# Task Brief
## Goal
## Relevant Files
## Current Behavior
## Required Change
## Constraints
## Verification
## Open Risks

Stop Rules

  • Stop exploring once new files are repeating facts.
  • Summarize command output instead of pasting logs.
  • Store durable project facts in markdown memory; do not repeat them in every prompt.
  • For codebase graph tools, request exact symbols or files rather than broad "explain everything" queries.

Read references/low-token-patterns.md for budgets and anti-patterns.

Validation

  • Context includes the goal, edit surface, constraints, verification, and risks.
  • Large logs or broad file reads are summarized instead of pasted.
  • Durable facts are written to memory/wiki only when they will help future sessions.
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: 467kUpdated 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: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 319kUpdated Claude CodeCodexVerifiedInstall