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 Kaminoikari/product-playbook --skill dev-disciplinegit clone --depth 1 https://github.com/Kaminoikari/product-playbookWrote 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/kaminoikari/product-playbook/dev-discipline)<a href="https://agentmods.dev/skills/kaminoikari/product-playbook/dev-discipline"><img src="https://agentmods.dev/badge/skills/kaminoikari/product-playbook/dev-discipline/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/kaminoikari/product-playbook/dev-discipline"><img src="https://agentmods.dev/badge/skills/kaminoikari/product-playbook/dev-discipline.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 | $0.00126 | $0.02900 |
| Opus 5 | $0.00063 | $0.01450 |
| Sonnet 5 | $0.00025 | $0.00580 |
| Haiku 4.5 | $0.00013 | $0.00290 |
Grade A, and why
dev-discipline 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 12d 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 — 125 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Dev Discipline
Detect the user's language and reply in it; the protocol below is authored in English.
Provenance: this lens governs process, so code changes carry no framework tag. Skip the — Frameworks: line on code output; add dev-discipline to the provenance line only when the outcome is a document (a review report, a handoff note).
Right-sizing: small vs large
Decide the track before the first edit; ceremony scales with ambiguity, not with nervousness.
- Enter plan mode only when two or more reasonable architectures exist or picking wrong wastes significant rework (session vs JWT auth, Redis vs in-memory cache, WebSocket vs SSE). Do NOT plan for: obvious bug fixes, convention-following features, renames, formatting, test-only changes, or "update the error handling" requests — start working and ask specific questions inline. Research and exploration go to subagents, never to plan mode.
- Small change — a single-purpose diff under ~30 lines with a green focused test: skip the reviewer subagents; review the diff inline yourself and run the entry-point launch check from gate 6. The deterministic backstops still apply.
- Large change — architectural ambiguity, multi-file, or high-risk surface: write the plan contract below before implementing, then run every gate in full.
Plan contract — large tasks only
Write the plan to docs/plans/<slug>.md before implementing. The plan is a frozen contract on the OBSERVABLE OUTCOME the task asks for, never on how to build it — freezing file names or function signatures lets a reviewer refute correct work for diverging from them. Sections, in order:
# Plan: <one-sentence headline>
## Acceptance criteria 3-5 gating, outcome-based criteria; group related behaviors, never silently drop one
## Verification plan per criterion: the action + the observation that must hold; tag each step gating|evidence
## Non-goals required — park plausible-but-unrequested scope here so a reviewer sees it was deferred, not forgotten
## Assumed scope files / modules / deps this touches
## Task checklist 3-8 checkboxes, flipped as you work; guidance, never part of the judged contract
## Deviations single section, one terse bullet per deviation; never edit the criteria themselves
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.
- 12d ago First seen · 125 lines · 0 tokens per session scan A cb6316b6ca6c
dev-discipline is a skill published in the GitHub repository Kaminoikari/product-playbook (24 stars, last pushed 1mo ago), licensed MIT. It adds 126 tokens to every session and 2,900 once invoked, about $0.0006 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 2026-08-30.
Other skills, from other repositories
isolate-subagent-verification-from-live-worktree
A verification / gate-running subagent you dispatched into YOUR OWN worktree silently measures the branch PLUS your uncommitted edits, and reports the mixture as the branch's result. Nothing is clobbered and nothing errors — the suite really ran, the counts are real, they are just counts of a tree nobody will ever…
worktree-historical-test-replay-missing-dirs
Fix pytest exit 4 ("file or directory not found") when running a test command in a git worktree checked out at an OLD commit. Use when: (1) you're doing historical replay (incident replay, mutation testing, git bisect with tests) and the suite errors out in 1-2 seconds with no test execution, (2) the test command…
prove-test-failures-pre-existing-via-clean-worktree
After making a change you run the test suite and it shows failures — especially in files/areas your diff never touched, or a count that "feels unrelated." Before you either panic-debug them OR wave them off as "probably pre-existing," PROVE it: run the exact failing tests against a clean checkout of origin/main (a…
launch-plan
Generate a comprehensive launch checklist from a PRD, covering release criteria, QA, comms, rollback, metrics, and stakeholder prep. Slash command: /launch.
cw-gates
Use before claiming any Codewhale change is done, green, or ready to land: the focused-to-broad verification ladder, the budget checks CI enforces, and the rules for what counts as a passing test.
model-merging
Merge multiple fine-tuned models using mergekit to combine capabilities without retraining. Use when creating specialized models by blending domain-specific expertise (math + coding + chat), improving performance beyond single models, or experimenting rapidly with model variants. Covers SLERP, TIES-Merging, DARE, Task…