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 thaitype/chief --skill chief-loopgit clone --depth 1 https://github.com/thaitype/chiefWrote 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/thaitype/chief/chief-loop)<a href="https://agentmods.dev/skills/thaitype/chief/chief-loop"><img src="https://agentmods.dev/badge/skills/thaitype/chief/chief-loop/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/thaitype/chief/chief-loop"><img src="https://agentmods.dev/badge/skills/thaitype/chief/chief-loop.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00070 | $0.01313 |
| Opus 5.5 | $0.00028 | $0.00525 |
| Sonnet 5.5 | $0.00014 | $0.00263 |
| Haiku 4.5 | $0.00007 | $0.00131 |
Grade A, and why
chief-loop 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 1mo 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 — 141 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run chief-agent across the full milestone — batch after batch — until both the goals and the contracts are satisfied, recording a decision-aware report for every task instead of one combined report per batch.
This builds directly on chief-autopilot's auto mode. If you want the "stop and ask a
human on ambiguity" behavior, use chief-autopilot safe instead — chief-loop only runs
in auto mode; it doesn't have a safe-mode equivalent.
Prerequisite Check
Before doing anything:
- Identify the active milestone directory under
.chief/. - Check that
_goal/has at least one non-empty file. - Check that
_contract/has at least one non-empty file.
If either is missing → STOP. Tell the user:
"Goals and contracts are required for chief-loop. Run
/chief-planfirst."
Do NOT proceed.
Entry Confirmation
Present the current goals and contracts to the user in a brief summary (file names + 1-line description each).
Ask one question:
"Goals and contracts look correct? Proceed with chief-loop, or use
/chief-planto revise first?"
If the user says revise → stop. If the user confirms → proceed.
Optional: if the loop-readiness skill is available, offer to run it against this
milestone's plan before proceeding — it reviews whether the plan has enough
feedforward/feedback coverage to run safely unattended. If it flags pre-existing gates
the plan depends on, also offer loop-preflight to actually run them and confirm they
still pass. Both are suggestions, not requirements; proceed without them if the user
declines.
Outer Loop (spans as many batches as it takes)
1. Create or update TODO
Read existing _plan/_todo.md if present. Create the next batch of 3–5 tasks based on
goals, contracts, and what's already done — same small-batches principle as
chief-autopilot.
Write _plan/_todo.md. Do NOT wait for approval — this is autopilot-style, no pausing
for plan review.
2. Work the batch, one task at a time
For each uncompleted task in the current batch:
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.
- 1mo ago Changed · +2 lines f98662233c84
- 2mo ago First seen · 139 lines · 70 tokens per session scan A 6cda4b4863c8
chief-loop is a skill published in the GitHub repository thaitype/chief (68 stars, last pushed 1mo ago), licensed MIT. It adds 70 tokens to every session and 1,313 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-09-03.
chief-loop compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| chief-loopthaitype/chief | Skill | 68 | 1mo ago | MIT | A | 70 | Claude Code |
| imagegenopenai/codex | Skill | 134k | today | Apache-2.0 | A | 113 | Codex |
| local-ai-agentsmicrosoft/ai-agents-for-beginners | Skill | 79k | yesterday | MIT | A | 200 | Claude Code, Codex |
| playwright-climicrosoft/playwright | Skill | 100k | yesterday | Apache-2.0 | A | 19 | Claude Code |
| browser-automationopenclaw/openclaw | Skill | 394k | today | MIT | A | 36 | Claude Code |
| systematic-debuggingobra/superpowers | Skill | 314k | 4d ago | MIT | A | 21 | Claude Code |
| next-cache-components-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 95 | Claude Code |