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 wsauret/flywheel --skill compoundgit clone --depth 1 https://github.com/wsauret/flywheelWrote 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/wsauret/flywheel/compound)<a href="https://agentmods.dev/skills/wsauret/flywheel/compound"><img src="https://agentmods.dev/badge/skills/wsauret/flywheel/compound/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/wsauret/flywheel/compound"><img src="https://agentmods.dev/badge/skills/wsauret/flywheel/compound.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 | $0.00032 | $0.01379 |
| Opus 5.5 | $0.00013 | $0.00552 |
| Sonnet 5.5 | $0.00006 | $0.00276 |
| Haiku 4.5 | $0.00003 | $0.00138 |
Grade A, and why
compound 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 — 206 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Compound Skill
Document solved problems to build searchable institutional knowledge.
Why "compound"? Each documented solution compounds your team's knowledge like compound interest. First time solving a problem takes research. Document it, and the next occurrence takes minutes.
Compounding vs Compaction: Flywheel uses two strategies (see README):
- Compaction = Reduce context mid-work (atomic
progress.jsoncheckpoints, structuredspec.json.contextarrays) - Compounding = Accumulate knowledge post-work (this skill)
Preconditions:
- Problem has been solved (not in-progress)
- Solution has been verified working
Organization: Single-file per problem in category directory (e.g., docs/solutions/performance-issues/n-plus-one-query.md).
Step 1: Detect Confirmation
Auto-invoke after: "that worked", "it's fixed", "working now", "problem solved"
OR manual: /compound command
Non-trivial problems only:
- Multiple investigation attempts needed
- Tricky debugging
- Non-obvious solution
Skip: Simple typos, obvious syntax errors, trivial fixes.
Additional categories:
pattern= Successful approach worth reusing →docs/solutions/patterns/mistake= Failed approach with prevention guidance →docs/solutions/mistakes/
Step 1.5: 3-Strike Integration
When invoked after 3-Strike escalation (error resolved with user help), capture:
- What failed: All 3 attempts (sanitized - remove credentials, API keys, PII, internal URLs)
- Why it failed: Root cause analysis
- What worked: User-provided solution
- How to prevent: Future guidance
This creates institutional memory from hard-won debugging sessions.
Step 2: Gather Context
Extract from conversation history:
- Module - Which component had the problem
- Symptom - Exact error messages, observable behavior
- Investigation - What was tried, what didn't work
- Root cause - Technical explanation
- Solution - Code/config changes that fixed it
- Prevention - How to avoid in future
What ships with it
5 files 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.
- 1mo ago First seen · 206 lines · 32 tokens per session scan A e4c4490369d3
compound is a skill published in the GitHub repository wsauret/flywheel (14 stars, last pushed 4d ago), licensed MIT. It adds 32 tokens to every session and 1,379 once invoked, about $0.0001 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 2026-08-30.
Other skills, from other repositories
handoff
Write a compact, decision-ready handoff so the next session (or the user) can continue without reconstructing the current one. Use when the session is ending, context is running low, the user asks for a handoff / 'pass the baton' / 'hand off', or a long-running operation needs a durable state checkpoint.
forget
Help the user remove a stored fact using Codewhale's Context Lens when they ask to forget or delete a memory.
maintaining-context
Use at the start of any substantial task and whenever you make a non-obvious decision or learn a reusable fact — keeps .yoke/context/ (PROJECT, DECISIONS, KNOWLEDGE) the durable source of truth so fresh-context work never drifts.
startup-context
When the user wants to create or update their startup context document. Also use when the user mentions "set up context", "tell you about my startup", or any other skill needs context that doesn't exist yet.
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.
money
Skill "money" from Hmbown/Codewhale, covering money, when to use, setup, workflow and non-goals.