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.
npx skills add impactbrussels/AINativeOS --skill pay-down-agentic-debtgit clone --depth 1 https://github.com/impactbrussels/AINativeOSWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/impactbrussels/ainativeos/pay-down-agentic-debt)<a href="https://agentmods.dev/skills/impactbrussels/ainativeos/pay-down-agentic-debt"><img src="https://agentmods.dev/badge/skills/impactbrussels/ainativeos/pay-down-agentic-debt.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00138 | $0.01509 |
| Opus 5 | $0.00069 | $0.00754 |
| Sonnet 5 | $0.00028 | $0.00302 |
| Haiku 4.5 | $0.00014 | $0.00151 |
Grade A, and why
pay-down-agentic-debt 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 7d 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.
How it starts
The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Pay Down Agentic Debt
In a machine-learning system, changing anything changes everything. The debt an agent leaves behind does not sit in one place and wait like the old kind; the intelligence at the centre is a learned thing, and a learned thing entangles whatever it touches. So a surgical, obviously correct fix in one corner degrades a model three hops away you would swear is unrelated, and you are debugging at midnight certain the system has a ghost in it. It does not have a ghost. It has agentic technical debt, and the founders who survive growth are the ones who rank it by interest rate and pay it down without breaking ten things to fix one.
The method
Find it, rank it by interest, pay it down CACE-safe through the eval gate. Full framework: references/debt-method.md. Source: Handbook Chapter 07.
Step 1: Audit the repo for the six debts
Point your agentic-coding tool at the codebase and force it to find the categories below, not a vague "what's messy". An unranked list paralyses; a named taxonomy you can score.
| Debt category | What to grep for | Why it bites |
|---|---|---|
| Prompt sprawl | One mega-prompt; rules added per case | Maximum entanglement; nothing testable in isolation |
| Entangled context | A signal feeding more than one learned component | A fix here moves outputs you cannot predict |
| Untested paths | Code with no eval asserting on its content | You ship blind; silent failure has no alarm |
| God Agent | One agent doing many jobs in one context | Unauditable, unchangeable, brittle by construction |
| Missing evals | No held-out set, no eval suite at all | You cannot prove any change is safe, only hope |
| Dead retrieval / broken loop | A signal generated but never stored | The leak in the moat: thrown-away learning |
Step 2: Score each item by interest x principal
Interest is how fast the debt compounds; principal is the cost to fix. Rank by interest first, always. See the scoring grid in references.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 7d ago First seen · 90 lines · 138 tokens per session scan A 61cec084c31a
pay-down-agentic-debt is a skill published in the GitHub repository impactbrussels/AINativeOS (1 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 138 tokens to every session and 1,509 once invoked, about $0.0007 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.
Other skills, from other repositories
trace
Evidence-driven tracing lane that orchestrates competing tracer hypotheses in Claude built-in team mode.
ai-slop-cleaner
Clean AI-generated code slop with a regression-safe, deletion-first workflow and optional reviewer-only mode.
debug
Diagnose the current OMC session or repo state using logs, traces, state, and focused reproduction.
node-zombie-guardian
Use when diagnosing stale or orphaned Node.js processes launched by VCO, auditing ownership/liveness, or safely simulating cleanup without touching external Node workloads.
gh-fix-ci
Use when a user asks to debug or fix failing GitHub PR checks that run in GitHub Actions; use gh to inspect checks and logs, summarize failure context, draft a fix plan, and implement only after explicit approval. Treat external providers (for example Buildkite) as out of scope and report only the details URL.
providing-performance-optimization-advice
Produce a prioritized performance-optimization roadmap across frontend, backend, and infrastructure. Use as an explicit/manual helper after bottlenecks are known or suspected, not as the owner of regression detection, profiling capture, or test execution.