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 commands/rana/skills/composegit clone --depth 1 https://github.com/rana/skillsWhat 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 | $0.00000 | $0.00894 |
| Opus 5 | $0.00000 | $0.00447 |
| Sonnet 5 | $0.00000 | $0.00179 |
| Haiku 4.5 | $0.00000 | $0.00089 |
Grade A, and why
compose 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 2d 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 — 86 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Chain multiple skills in sequence, threading context between passes.
Argument format: skill names separated by commas, optional focus area after a colon. Parenthesized groups iterate as a unit.
Examples:
/compose gaps, crystallize, deep-review/compose deep-review, api-review : authentication endpoints/compose (invoke, review, converge) ~3, land : API design/compose (gaps, crystallize) * 2, land
$ARGUMENTS
Steps
- Parse the argument into an ordered list of skills and groups, plus an optional focus area.
- Bare skill names execute once in sequence.
- Parenthesized groups
(a, b, c)execute as a unit, subject to iteration control.
- For each skill or group in order: a. Announce which skill is running (e.g., "Pass 1/3: /gaps"). b. Execute that skill's methodology against the project documents, applying the focus area if provided. c. Capture findings as a numbered list. d. Feed findings forward — each subsequent skill should account for what previous passes discovered.
- After all passes, present a unified summary:
- Key findings across all passes, deduplicated
- Conflicts between passes (e.g.,
/gapssays "add X" but/crystallizesays "remove complexity") - REDIRECT annotations from any converge passes (scope shifts to note)
- Prioritized action list combining all recommendations
- No changes to files — document only. Exception: if
landis the final skill in the chain, it transitions to action mode and may write files.
Iteration Groups
Parenthesized skill groups repeat as a unit. Iteration control:
* N— repeat exactly N times~ N— repeat until the group's convergence gate says STABLE, max N~— alias for~3(default bounded)
If a group contains converge, its directive controls the loop:
- STABLE — exit group, continue to next skill in chain
- CONTINUE — repeat the group, feeding converge's focus forward
- REDIRECT — note the scope shift for the unified summary, continue the group
- STUCK — pause chain, present converge's assessment, wait for user input. Recovery options: user says "skip" (exit group, continue chain), "redirect to X" (alter focus and continue group), or "break" (stop entire chain). At autonomous authority, resolve STUCK by choosing the strongest path forward with reasoning noted.
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.
- 2d ago First seen · 86 lines · 0 tokens per session scan A 33bdb5331371
compose is a command published in the GitHub repository rana/skills (1 stars, last pushed 5mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 894 tokens. 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 commands, from other repositories
searchpo-full-flow
Run the Searchpo full provider/live flow with source cross-check, render audit, and Douyin gate verification.
github-repo-to-x-threads
Analyze repos, papers, URLs, files, ideas, profile prompts, or support complaints and create Agent-engineering-positioned X content packs with strategy routing, evidence review, reusable assets, images, and optional publish.
git
Git operations with intelligent commit messages and workflow optimization.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.