Claude Octopus is an orchestration project that sends research, design, and coding tasks to Claude Code and other AI model providers so their results can be compared. Developers use it for multi-model work, disagreement detection, reviews, persistent context, and an optional workflow that moves from discovery through delivery. The catalogue entries are its commands, skills, agents, instructions, hooks, plugins, and settings.
Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/nyldn/claude-octopusnpx agentmods add skills/nyldn/claude-octopus/skill-writing-plansWrote 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/nyldn/claude-octopus/skill-writing-plans)<a href="https://agentmods.dev/skills/nyldn/claude-octopus/skill-writing-plans"><img src="https://agentmods.dev/badge/skills/nyldn/claude-octopus/skill-writing-plans/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/nyldn/claude-octopus/skill-writing-plans"><img src="https://agentmods.dev/badge/skills/nyldn/claude-octopus/skill-writing-plans.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 326 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00021 | $0.02192 |
| Opus 5.5 | $0.00008 | $0.00877 |
| Sonnet 5.5 | $0.00004 | $0.00438 |
| Haiku 4.5 | $0.00002 | $0.00219 |
Grade A, and why
skill-writing-plans 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 — 359 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Writing Plans
Load skills/blocks/engineering-method-selection.md from the installed plugin
and apply only the methods relevant to this task. Preserve this entry point's
execution contract and output format. Read referenced skills as instructions;
do not invoke the current command recursively or add provider calls from a seat.
MANDATORY COMPLIANCE — DO NOT SKIP
When this skill is invoked, you MUST produce a full implementation plan following the structure below. You are PROHIBITED from:
- Skipping the plan and jumping straight to implementation
- Producing a vague outline instead of the zero-context plan format
- Deciding the task is "simple enough" to not need a plan
- Omitting file paths, complete code, test instructions, or verification steps
The user asked for a plan, not an implementation. Write the plan first.
Your first output line MUST be: 🐙 **CLAUDE OCTOPUS ACTIVATED** - Implementation Planning
Overview
Write comprehensive implementation plans assuming the engineer has zero context for the codebase and questionable taste.
Document everything: which files to touch, complete code, how to test, how to verify.
Principles: DRY. YAGNI. TDD. Frequent commits.
Decisions before tasks
Build a dependency graph for unresolved decisions before writing implementation tasks. A decision record contains its question, evidence required, dependencies, owner, resolution, and the implementation it unblocks. Use the repository's configured tracker. Beads is not an end-user requirement.
Decision states are open, claimed, resolved, superseded, and blocked,
mapped to native tracker states or labels. Resolve only from evidence or a
recorded human decision. If new evidence invalidates a decision, reopen its
dependent work and explain why. A cycle means the work is not ready; recut the
decisions rather than marking tasks ready.
Claim through the tracker's atomic operation and read ownership back before writing. If atomic claiming is unavailable, appoint one integrator. Never overwrite another claim. When the tracker fails, save an explicitly unfiled proposal in existing plan storage, stop tracker writes, and never fabricate IDs or migrate a database.
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 · +24 lines ca27357798a7
- 1mo ago First seen · 335 lines · 21 tokens per session scan A 55d17f4004fc
skill-writing-plans is a skill published in the GitHub repository nyldn/claude-octopus (4,320 stars, last pushed today), licensed MIT. It adds 21 tokens to every session and 2,192 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 2026-09-03.
Other skills, from other repositories
api-integrations
Expose external APIs to Falcon Foundry via OpenAPI specs. TRIGGER when user asks to "create an API integration", "adapt an OpenAPI spec for Foundry", "expose an API to workflows", "connect to a third-party API", or runs foundry api-integrations create. Also trigger when user has an OpenAPI/Swagger spec and wants it…
app-upgrades
Upgrade a Falcon Foundry app that was installed from a template to a newer version of that template, in place. TRIGGER when user asks to "upgrade an app", "update an app installed from a template", "pull the latest template into my app", "update a sample app installed in my CID to the latest version", or gives an app…
one-shot
This skill should be used when running the full autonomous engineering workflow from plan to merged PR.
schedule
This skill should be used when creating, listing, or deleting scheduled agent tasks via GitHub Actions cron workflows. It generates workflow YAML files that invoke Soleur skills on a recurring schedule using claude-code-action.
postmerge
This skill should be used when verifying a merged PR deployed correctly and production is healthy.
social-distribute
This skill should be used when distributing a blog article across social platforms (Discord, X/Twitter, IndieHackers, Reddit, Hacker News, LinkedIn). Writes a persistent content file for automated publishing.