token-compression
Output token compression for AA-MA workflows. Reduces response tokens by 40-75% while preserving full technical accuracy. Three intensity levels mapped to execution modes: lite (HITL), full (general), ultra (AFK). Auto-activates during AA-MA execution based on task Mode field. Manual activation via /compress lite|full|ultra.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-12
- Stars
- ★ 6
- Forks
- 1
01 前言 Front Matter
- name
- token-compression
- description
- Output token compression for AA-MA workflows. Reduces response tokens by 40-75% while preserving full technical accuracy. Three intensity levels mapped to execution modes: lite (HITL), full (general), ultra (AFK). Auto-activates during AA-MA execution based on task Mode field. Manual activation via /compress lite|full|ultra.
03 SKILL.md 正文
Token Compression
Reduce output tokens while preserving technical substance. Inspired by caveman prompt compression patterns, adapted for AA-MA execution mode integration.
Activation
- Automatic: During AA-MA execution, intensity is set by task Mode field
Mode: HITL→ lite (preserve clarity for human review)Mode: AFK→ ultra (maximum compression for autonomous execution)
- Manual: ask for "compress lite|full|ultra" to override during any session
- Default:
fullwhen activated outside AA-MA execution context
Rules
Drop: articles (a/an/the), filler (just/really/basically/actually/simply), pleasantries (sure/certainly/of course/happy to), hedging (it might be worth, you could consider). Fragments OK. Short synonyms (big not extensive, fix not "implement a solution for"). Technical terms exact. Code blocks unchanged. Errors quoted exact.
Pattern: [thing] [action] [reason]. [next step].
Not: "Sure! I'd be happy to help you with that. The issue you're experiencing is likely caused by..."
Yes: "Bug in auth middleware. Token expiry check use < not <=. Fix:"
Intensity Levels
| Level | Rules | When |
|---|---|---|
| lite | No filler/hedging. Keep articles + full sentences. Professional but tight. | HITL tasks, complex explanations, user-facing content |
| full | Drop articles, fragments OK, short synonyms. | General use, default level |
| ultra | Abbreviate (DB/auth/config/req/res/fn/impl), strip conjunctions, arrows for causality (X → Y), one word when one word enough. | AFK tasks, autonomous execution, subagent output |
Examples
"Why does the test fail?"
- lite: "The test fails because the mock returns an empty list instead of the expected user objects. Update the mock fixture to return test data."
- full: "Mock returns empty list, not expected user objects. Update mock fixture with test data."
- ultra: "Mock → empty list. Need test data in fixture."
"Explain the deployment pipeline."
- lite: "The pipeline runs lint, tests, and builds a Docker image. On merge to main, it deploys to staging automatically. Production requires manual approval."
- full: "Pipeline: lint → test → Docker build. Merge to main = auto-deploy staging. Prod needs manual approval."
- ultra: "lint→test→Docker→staging(auto). Prod=manual."
Auto-Clarity Exceptions
Drop compression for:
- Security warnings and vulnerability disclosures
- Irreversible action confirmations (DELETE, DROP, force-push)
- Multi-step sequences where fragment order risks misinterpretation
- User appears confused or asks for clarification
Resume compression after the clear section is complete.
Example — destructive operation:
Warning: This will permanently delete all rows in the
userstable and cannot be undone.DROP TABLE users;Compression resume. Verify backup first.
Boundaries
- Code blocks: always rendered normally, never compressed
- Commit messages: follow Conventional Commits format, never compressed
- PR descriptions: write normally
- Error messages: quote exact, never paraphrase
- "compress off" or "normal mode": deactivate compression
- Level persists until changed or session end
Integration with AA-MA Execution
When the aa-ma-execution skill detects a task's Mode: field:
Mode: AFK → Skill(token-compression) at ultra
Mode: HITL → Skill(token-compression) at lite
No Mode → Skill(token-compression) at full (default)
This mapping ensures autonomous agents produce minimal output (saving tokens for reasoning), while human-interactive sessions preserve enough context for the user to follow along.
04 同仓库更多 Skills
deploy-helper
You can use this to help with deploy stuff, deploying, deployments, releases, shipping, pushing to prod, going live, rollouts.
writing-for-agents
Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.
verify-plan
Slash entry for plan verification: runs plan-verification on an existing AA-MA plan, writes [task]-verification.md, and with --iterate loops until the plan converges. Use when the user types /verify-plan or wants the verification report saved.
verify-impl
Post-impl adversarial review symmetric to /verify-plan. Dispatches up to 5 parallel audit agents (code-reviewer, security-auditor, tdd-sequence-auditor, context7-evidence-auditor, future-proofing-auditor) based on the milestone's plan-declared Audit-Profile. CRITICAL findings surface via AskUserQuestion accept/dispute/defer panel before §7.3 user authorization. Introduced in v0.8.0 per ADR-0005. Invoked by Phase 6.8 of /execute-aa-ma-milestone.
understand-codebase
Onboard to a new, inherited, or shared codebase. Read it, understand it, map it, and learn its conventions, versioning, tests, tech stack, rules (AGENTS.md / CLAUDE.md / .cursorrules / etc.), build/run/CI, integrations, security posture, and repo health — then produce an honest pros/cons verdict, a "contribute safely" playbook, and an "add a feature" playbook, written to ONBOARDING.md at the repo root plus a .claude/onboarding/ set of deep-dives — and optionally author an AGENTS.md if one is missing, or review and propose improvements to an existing one. Tiered (Quick / Standard / Deep). Reuses codemem (index + diagrams), gsd-map-codebase, system-mapping, code-intelligence, impact-analysis rather than re-implementing them; Deep tier runs a TeamCreate agent-team. Keywords: new codebase, shared codebase, inherited code, onboard, understand this repo, how do I contribute, how do I add a feature, ramp up, get oriented, ONBOARDING.md, AGENTS.md, codebase walkthrough, joining a project.
system-mapping
Lightweight pre-flight checklist for understanding system context before code changes. Perform 5-point analysis of architecture, execution flows, logging, dependencies, and environment. Use before modifying unfamiliar code or multi-file changes.