compounded-trust-tracker

A trust-management procedure for deciding how cautiously to use skills at different reliability levels, from unverified to autonomous.

In plain words
What is it for?
Use it when evaluating, invoking, confirming, or recording the results of skills marked verified, trusted, or autonomous.
Why use it?
It prevents unverified procedures from being used blindly and requires confirmation where the trust level calls for it.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/ankitkr3/compounded/compounded-trust-tracker
Any agent
npx skills add ankitkr3/compounded --skill compounded-trust-tracker
Clone the repo
git clone --depth 1 https://github.com/ankitkr3/compounded

Made for: Claude Code, Codex.

Per session 79 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,046 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00079 $0.01046
Opus 5 $0.00039 $0.00523
Sonnet 5 $0.00016 $0.00209
Haiku 4.5 $0.00008 $0.00105

Measured 2d ago against content hash f4784d96413f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

compounded-trust-tracker scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/compounded-trust-tracker/SKILL.md · 69 lines

How it starts

The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Trust Tracker (compounded)

compounded skills earn the right to act through demonstrated reliability. Your job is to apply the right level of caution at each trust state, and to honestly report outcomes so the gradient stays calibrated.

The four states, in detail

.proposed — untrusted, not in your skill index

Proposed skills are not loaded. You cannot invoke them. They exist only as candidates awaiting verification. Do not attempt to read or apply them. If the user asks you to use a proposed skill, explain that it has not been verified yet and offer to run /compounded:verify <skill> manually.

.verified — invoked by name only

A verified skill has passed exactly one replay verification. Treat it as a credible procedure but not yet a habit.

  • Activation: only when the user names it explicitly, or when you decide a verified skill is clearly relevant and you announce its use.
  • Confirmation: announce that you are loading the skill before applying it. e.g. "I'm going to use the express-to-fastify skill for this — it was verified once before."
  • Outcome reporting: if the skill works cleanly, compounded's logging will count this toward promotion to .trusted. If the user corrects you, the skill demotes back to .proposed. Do not edit the trust state yourself; just report the outcome accurately.

.trusted — auto-loads, asks before applying

A trusted skill has been used successfully 3+ times without correction. It is known to work in this user's context.

  • Activation: auto-load when the task description matches the skill's "When to use" section.
  • Confirmation: ask before applying. e.g. "This task matches the express-to-fastify trusted skill. Should I use it?"
  • Skip confirmation only if the user has explicitly told you to apply trusted skills automatically in the current session.

.autonomous — runs without asking, every time

An autonomous skill has 10+ uses, with zero corrections in the last 5. It has earned full autonomy.

Read the full file on GitHub · 69 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 69 lines · 79 tokens per session scan A f4784d96413f

Subscribe to this mod's changes

compounded-trust-tracker is a skill published in the GitHub repository ankitkr3/compounded (5 stars, last pushed 2mo ago), licensed MIT. It adds 79 tokens to every session and 1,046 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories