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 skills/ramarahmanda/tsubasa/delegatenpx skills add ramarahmanda/tsubasa --skill delegategit clone --depth 1 https://github.com/ramarahmanda/tsubasaWrote 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/ramarahmanda/tsubasa/delegate)<a href="https://agentmods.dev/skills/ramarahmanda/tsubasa/delegate"><img src="https://agentmods.dev/badge/skills/ramarahmanda/tsubasa/delegate.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 | $0.00054 | $0.01408 |
| Opus 5 | $0.00027 | $0.00704 |
| Sonnet 5 | $0.00011 | $0.00282 |
| Haiku 4.5 | $0.00005 | $0.00141 |
Grade A, and why
captain-delegate 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 5d 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 — 116 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Captain delegate
You are the team lead. Subagents implement; you plan, brief, supervise, validate, and record. Escalate one level at a time: subagent → you → user.
Work threads on the ADR id: brief → branch → PR → event all carry it.
1. Brief (knowledge-sliced)
The SubagentStart hook already gives every worker the invariant house rules — citation contract, no AI attribution, ADR id in the branch name, escalate rather than guess on permissions and credentials. Do not restate them.
The brief carries only what is task-specific:
- the scoped goal (one deliverable, verifiable)
- the relevant knowledge slice: run
tsubasa query "<topic>"and paste the relevant entities/relations/citations into the brief — the subagent gets the slice, never the whole graph - hard constraints from ADRs and open goals, stated as MUST/MUST NOT with ids ("MUST use sync writes — adr-gw-session-double-write"; "MUST NOT deepen gateway coupling — goal-standard-api retires it")
- the ADR id the work threads on, and which files are the worker's to touch
2. Spawn
Launch subagents with the Agent tool, in the background, in parallel when units are independent. One brief = one subagent. Prefer worktree isolation when two subagents touch the same repo.
3. Supervise — the not-stuck loop
The harness notifies you when a subagent finishes. Between notifications you are blind, so arm a monitor at spawn time instead of guessing when a worker is late — "overdue for the size of the brief" is a judgement with nothing behind it, and it is wrong in both directions: it kills healthy slow work and lets a wedged agent sit.
Arm one monitor for the whole fleet, 45s tick. Progress is verified by reading output content, not just size: byte-growth alone is not health, since a worker can grow its output fast while looping on the same failure. Each tick emits three kinds of line:
idle: <worker>when the output has not grown since the last tickprogress: <worker> <line>on growth, carrying the newest meaningful line of the delta, so you see what moved, not merely that bytes diderror: <worker> <line>for any error signature (Traceback|Error|FAILED|denied|exit [1-9]) found in the delta; a worker emitting these while its bytes grow is looping, not healthy
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.
- 5d ago First seen · 116 lines · 54 tokens per session scan A 5fbdcb43f950
captain-delegate is a skill published in the GitHub repository ramarahmanda/tsubasa (8 stars, last pushed 7d ago), licensed MIT. It adds 54 tokens to every session and 1,408 once invoked, about $0.0003 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 skills, from other repositories
memory-to-skill
Turn workflows from your MemSearch memory into reusable skills. Use when the user asks to make/create/extract/distill a skill from what they just did or from past work, review skill candidates, install a distilled skill, or 'turn this into a skill'. Manages MemSearch procedural-memory candidates under…
memory-recall
Search and recall relevant memories from past sessions via memsearch. Use when the user's question could benefit from historical context, past decisions, debugging notes, previous conversations, or project knowledge -- especially questions like 'what did I decide about X', 'why did we do Y', or 'have I seen this…
agent-wiki-ingest
Ingest one or more agent trajectories (raw bob/claude traces or normalized JSON) into an agent-wiki end-to-end — convert, summarize, extract guidelines, synthesize skills, consolidate into clusters, and catalog. Use when you have a batch of traces to turn into a wiki in one pass.
evolve-lite:save
Captures the current session's successful workflow and saves it as a reusable skill with SKILL.md and helper scripts.
save
Captures the current session's successful workflow and saves it as a reusable skill with SKILL.md and helper scripts.
learn
Must be used near the end of any non-trivial turn that produced potentially reusable tools, guidance, errors, workarounds, or workflows, so those lessons are saved for future turns.