agentleFS
IgorGanapolskyIgorGanapolsky/mac-yolo-safeguards/.agents/skills/agent-harness-router/SKILL.md

agent-harness-router

Skill3 starsChanged 4 days agoMITView on GitHub

Agent harness router consolidating TrueFoundry TrueForge (harness as control point, $10/mo budget enforcement), Martech (scattered AI discovery capture), GLM-5.3 (private benchmark validation, model portability), and Futurism (ChatGPT Computer History fail-closed alternative). Routes model calls by cost/latency/quality/privacy, tracks spend against hard cap, captures reusable prompt patterns, validates models on private benchmarks. Never captures keystrokes, clicks, or local Mac activity.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never captures keystrokes, clicks, or local Mac activity timelines (Futurism)

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Trigger

Any agent task involving model selection, cost optimization, budget enforcement, discovery capture of AI patterns, or private benchmark validation.

Usage

# Route a task to the best model (cost/latency/quality/privacy/balance)
node tools/agent-harness-router.js route '{"requiresTools":true,"category":"coding","preference":"quality"}'

# Check current budget + discovery/benchmark stats
node tools/agent-harness-router.js check --json

# Capture a useful prompt/pattern for reuse
node tools/agent-harness-router.js discover "Write a function to sort arrays" --model zai-org/glm-5.3 --task coding --tags utility,beginner

# Private benchmark: add a task, record results, get portability scores
node tools/agent-harness-router.js benchmark add --name "Sort test" --category coding --prompt "Write sort" --expected "sorted"
node tools/agent-harness-router.js benchmark portability

# List all known models
node tools/agent-harness-router.js registry --json

Health Check

node tools/agent-harness-router.js check --json
# Exit code 0 = budget not exhausted, storage healthy
# Exit code 1 = budget exhausted (fail-closed)

Architecture

  • MODEL_REGISTRY: 9 models with cost/latency/quality/privacy metadata
  • routeModel(): Optimal model for a task based on 5 routing preferences
  • BudgetGuard: Fail-closed $10/mo spending cap (per Anti-Babysitting + GLM-5.3)
  • DiscoveryCapture: Captures useful prompt/task patterns with secret rejection + input-capture refusal
  • PrivateBenchmark: Validates models on real tasks, computes portability scores
  • ModelRegistry: isFirstParty/isVerifiedThirdParty/isSafeForSensitiveData classification

Safety Guarantees

  • Never captures keystrokes, clicks, or local Mac activity timelines (Futurism)
  • Fail-closed budget enforcement at $10/mo
  • Secret detection rejects AWS keys, GitHub PATs, JWTs, Bearer tokens in captured discoveries
  • Input-capture detection refuses Computer History / event_tap / keystroke data
  • Third-party models blocked when budget exhausted or sensitive data required

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.