principle-trace-token-burn-loop SKILL.md skillA
principle-trace-token-burn-loop from edbertchan/catstack is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add edbertchan/catstack/principle-trace-token-burn-loop@git:20260924.a859be2Pins git:20260924.a859be2 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- edbertchan/catstack
- Path
corpus/skills/principle-trace-token-burn-loop/SKILL.md- Versions recorded
- 2
- Latest version
- git:20260924.a859be2, committed Sep 24, 2026
- First version recorded
- Sep 4, 2026
- Size
- 2,783 bytes
- Repository stars
- 6
- License
- none declared
Apply when analyzing token burn or agent session spend. Prevents mistaking the top task_type or prompt_type for the root cause. Forces a trace from the session back to the loop script or plan entry point before prescribing a fix.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# Trace token burn back to the loop script When a session miner reports a task_type or prompt_type consuming tokens, the symptom (what the agent was doing) is not necessarily the cause (what kept re-spawning the agent). **Why:** session JSONL contains `cwd`, `workflow_id`, and the first user prompt, but the scheduler name is missing. An unbounded outer loop — a worker `intervalMs`, a `while true` shell driver, a cron plan, or a plan task with no iteration limit — can make a cheap plan become an expensive burn. **Pattern:** 1. Rank spend by task_type / prompt_type, then open one representative session file. Fleet-wide: `scripts/scan_session_tokens.py <session-dirs>` emits one JSONL row per session file (codex exec, codex rollout, claude, OMP formats; `--summary` for the agent x workload table), and is pipeable over ssh — `cat scan_session_tokens.py | ssh host python3 - ~/.invoker/agent-sessions ~/.claude/projects ~/.codex/sessions`. For a single session's pricing / thrash detail use `engine/skills/reflect/scripts/token_audit.py`. 1a. Check whether the burn is session-lifetime (few long sessions, context …
Read the whole file at its exact version.
How to install
mdr add edbertchan/catstack/principle-trace-token-burn-loop@sha256:d976c4066dc3b0d6The line at the top of this page pins the label git:20260924.a859be2, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_j5hr53lie5gmb6le)
1 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 2 versions of this file, between Sep 4, 2026 and Sep 24, 2026, about one every 20 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20260924.a859be2 latest | 2026-09-24 | a859be2 | 2,783 B | A | view · diff |
| git:20260904.5579d02 | 2026-09-04 | 5579d02 | 1,720 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (2783 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
edbertchan/catstack · 6 stars · license none declared · pushed 2026-10-22 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_j5hr53lie5gmb6le GET https://markdownregistry.com/api/v1/resolve?ref=edbertchan/catstack/principle-trace-token-burn-loop GET https://markdownregistry.com/api/v1/blob/d976c4066dc3b0d645dcc84e4724ad47453a2943de39ab7af072962d12a8b580
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.