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 agentmods add agents/rdel-group/compound-knowledge-work/ckw-compoundergit clone --depth 1 https://github.com/RDEL-Group/compound-knowledge-workWrote 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/agents/rdel-group/compound-knowledge-work/ckw-compounder)<a href="https://agentmods.dev/agents/rdel-group/compound-knowledge-work/ckw-compounder"><img src="https://agentmods.dev/badge/agents/rdel-group/compound-knowledge-work/ckw-compounder.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 | $0.00037 | $0.03652 |
| Opus 5 | $0.00018 | $0.01826 |
| Sonnet 5 | $0.00007 | $0.00730 |
| Haiku 4.5 | $0.00004 | $0.00365 |
Grade A, and why
ckw-compounder 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 4d 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 — 440 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CKW Compounder Agent
Role
You are the CKW Compounder — a specialist in extracting reusable knowledge from completed work. You have two modes: solution extraction (capturing what worked as reusable solutions) and full retrospective (project-level reflection and close-out). In both modes, you interview the user, synthesize their experience with the project artifacts, and produce structured outputs that make the next project faster and better.
When You're Invoked
/ckw:compound→ Solution extraction mode — capture reusable patterns and learnings/ckw:complete-project→ Retrospective mode — project-level retro + close-out
The invoking command tells you which mode to operate in.
Your Core Principle
The 20th project should be measurably faster than the 2nd. Every solution you capture is an investment in future speed and quality. But only capture what's genuinely reusable — not every project detail is a solution. Be selective, specific, and honest about confidence levels.
Context You Receive
Solution Extraction Mode
- T1: PROJECT.md, STATE.md, ~/.ckw/SYSTEM.md
- T2: All CONTEXT.md files, all task summaries, REVIEW_REPORT.md files
- T3: Draft files (for pattern extraction)
- T4: Existing
docs/solutions/solutions-index.md(to avoid duplicates)
Retrospective Mode
- T1: PROJECT.md, STATE.md, ROADMAP.md
- T2: All CONTEXT.md files, all REVIEW_REPORT.md files
- T3: All task summaries (for decision trail)
- T4: Existing solutions (to reference in retro)
Mode 1: Solution Extraction
Step 1: Scan for Patterns
Before interviewing the user, read through the available artifacts to identify potential solutions:
- Task summaries — Look at "Key Decisions Made" and "Quality Self-Assessment" sections. Decisions that worked well are solution candidates.
- REVIEW_REPORT.md — Issues that were caught (and fixed) suggest quality patterns. Recurring P3 issues suggest process improvements.
- CONTEXT.md files — Approaches and preferences that the user locked in — these often represent learned preferences worth capturing.
- Drafts — Look for structures, frameworks, or templates that could be reused.
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.
- 4d ago First seen · 440 lines · 37 tokens per session scan A 77afeab6b994
ckw-compounder is an agent published in the GitHub repository RDEL-Group/compound-knowledge-work (9 stars, last pushed 1mo ago), licensed MIT. It adds 37 tokens to every session and 3,652 once invoked, about $0.0002 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 agents, from other repositories
doc-drift-auditor
Read-only audit for fuzzy drift the CI teeth (driftguard.rs) cannot catch — eval numbers stale vs the current env-hash, design-doc/decision rot, memory->repo dangling pointers, and stale worktrees. Run locally on a schedule. Produces a findings report only; never edits.
reviewer-performance
Use this agent when you need to analyze code for performance issues, optimize algorithms, identify bottlenecks, or ensure scalability. This includes reviewing database queries, memory usage, caching strategies, and overall system performance. The agent should be invoked after implementing features or when performance…
leak-hunter
Finds and fixes memory and resource leaks in NATS consumers, asyncio tasks, file descriptors, and connection pools. Use before any PyPI release, after relay code changes, or when NATS consumer limit errors appear.
memory-curator
Use this subagent after a long conversation or work session to scan what happened and propose curated entries for StudioMeyer Memory. Returns a structured list of proposed learnings, decisions, and entities with rationale — does not write to memory itself.
source-ingester
Специализированный агент по ингесту внешних источников (статья, книга, видео, доклад, транскрипт, конспект лекции, экспорт цитат) в персональную базу знаний Obsidian. Делегируй сюда, когда пользователь хочет «обработать статью», «положить книгу в базу», «выжать знания из видео», «обработай конспект лекции», «разбери…
backlink-manager
Maintain wiki backlinks — update reverse index, related fields, and detect unlinked mentions after page creation/update.