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 CrumbtrailDev/crumbtrail-cli --skill stale-derived-stategit clone --depth 1 https://github.com/CrumbtrailDev/crumbtrail-cliWrote 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/crumbtraildev/crumbtrail-cli/stale-derived-state)<a href="https://agentmods.dev/skills/crumbtraildev/crumbtrail-cli/stale-derived-state"><img src="https://agentmods.dev/badge/skills/crumbtraildev/crumbtrail-cli/stale-derived-state.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 | $0.00085 | $0.01705 |
| Opus 5 | $0.00043 | $0.00852 |
| Sonnet 5 | $0.00017 | $0.00341 |
| Haiku 4.5 | $0.00009 | $0.00170 |
Grade A, and why
stale-derived-state 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 — 148 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The stale cache or derived state that never invalidates
Something is written. Something else is derived from it and was computed earlier. The write path never tells the derived thing to recompute, so the screen, the session config, the lookup table or the client store keeps serving the value it already had. A reload rebuilds everything from source and the problem disappears, which is exactly why it gets reported as intermittent when it is not.
The diagnostic move is to prove the write landed and the read did not reflect it. Crumbtrail is good at the first half and only partly good at the second.
Symptom
A user changes a setting, renames something, toggles a permission or saves a record. The interface accepts it. The new value does not appear. Reloading the page, logging out and back in, or navigating away and returning makes it appear.
A second phrasing is the opposite direction: a value that should have been cleared keeps coming back, or a list that should now be empty keeps rendering its previous contents.
What Crumbtrail can see
The honest limitation first, because it decides how you spend the next few turns. The cache that failed to invalidate is usually an in memory object: a client store, a module scoped map, a memoized selector, a server side lookup table. None of that is recorded. Crumbtrail instruments the boundaries a value crosses, not the heap it sits in. If the value never crossed a boundary, the recording cannot show it to you.
What the recording does carry, all of which are boundaries a stale value tends to cross:
- Storage writes with the storage type, the operation, the key, and the old and new values. Keys and values that look sensitive come back replaced with a redaction marker, so you may see that a key changed without seeing what it changed to.
- Cookie changes.
- Every request and response, so the write and the later read are both present with their bodies. This is the strongest available proxy: a read issued after the write that returns the old value proves the staleness is on the server or in a shared cache rather than in the browser, and a read that never happens at all proves the opposite.
- Navigation events, which matter because this archetype is defined by what a reload fixes.
- Database row changes with before and after column values, but only where the database adapter is installed. This is what turns "the write probably landed" into a fact.
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 · 148 lines · 85 tokens per session scan A 58612a969342
stale-derived-state is a skill published in the GitHub repository CrumbtrailDev/crumbtrail-cli (0 stars, last pushed yesterday), licensed MIT. It adds 85 tokens to every session and 1,705 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.
Other skills, from other repositories
backend-patterns
Backend architecture patterns, API design, database optimization, and server-side best practices for Node.js, Express, and Next.js API routes.
codex-log-guard
Diagnose excessive Codex local SQLite diagnostic log writes with read-only evidence by default. Use when a user mentions logs2.sqlite, logs2.sqlite-wal, blockloginserts, SSD/TBW wear, or explicitly asks to protect, clean up, verify, or restore Codex diagnostic logging.
solr-query
To build and debug Solr queries: eDisMax, block join, JSON facets, kNN, explain.
qdrant-monitoring-debugging
Diagnoses Qdrant production issues using metrics and observability tools. Use when someone reports 'optimizer stuck', 'indexing too slow', 'memory too high', 'OOM crash', 'queries are slow', 'latency spike', or 'search was fast now it's slow'. Also use when performance degrades without obvious config changes.
edge-serverless-db-expert
Expert guide for Serverless & Edge Databases (Neon Serverless Postgres, Cloudflare D1, Turso/libsql, Upstash Redis), cold-start mitigation, and connection pooling / Panduan ahli database Serverless & Edge (Neon, Cloudflare D1, Turso, Upstash).
spring-session
Spring Session for distributed session management with Redis, JDBC, or Hazelcast. Covers session configuration, security integration, and session events. USE WHEN: user mentions "spring session", "distributed session", "session Redis", "session JDBC", "session cluster", "session management", "@SessionScope" DO NOT USE…