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/binary16labs/prime-silo/phase-runnernpx skills add binary16labs/prime-silo --skill phase-runnergit clone --depth 1 https://github.com/binary16labs/prime-siloWhat 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 | $0.00082 | $0.00507 |
| Opus 5 | $0.00041 | $0.00253 |
| Sonnet 5 | $0.00016 | $0.00101 |
| Haiku 4.5 | $0.00008 | $0.00051 |
Grade A, and why
phase-runner 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.
What it actually says
Phase runner (draft — manual protocol until W1/W2 land)
- Read the protocol once:
delivery/README.md. It is binding; this skill is its checklist. - Select: open
delivery/board/BOARD.md, take the TOPMOST item in READY. Never choose. - Claim: move its line to CLAIMED (
· <agent> · <ISO date>), append aclaimedline todelivery/board/LOG.md, commitchore(delivery): claim <ID>BEFORE working. - Sandbox:
git worktree add .worktrees/<ID> -b feat/<ID> main(contractsandbox: worktree). Work only inside it. Never restart services a live run depends on. - Read ONLY your contract (
delivery/tasks/<ID>.md). Insufficient contract = logblocked: contract insufficient — <missing>and stop. Never improvise scope. - TDD, in order: failing test (run it, observe red, commit) → implement → green → refactor.
Name tests after the contract's
Scenario:lines. Respect the allowlist andbudget. - Verify: run the contract's
verifycommand until exit 0 from a clean tree. Lint/format touched files. Validate contracts changed:node scripts/gates/w0.mjs. - Hand off: move the board line to VERIFY + append
ready-for-verifyto LOG (honest deviations included), commit on main. You may NOT move your own task to DONE — a fresh clean-checkout session with a distinct identity re-runs the gate (README rule 7). - Blocked twice = stop. Append
blockedwith the exact error and what you tried.
Log format: <ISO-ts> | <task-id> | <event> | <agent> | <note> — append-only, never edit history.
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 · 26 lines · 82 tokens per session scan A 9f42908c5249
phase-runner is a skill published in the GitHub repository binary16labs/prime-silo (5 stars, last pushed 9d ago), licensed MIT. It adds 82 tokens to every session and 507 once invoked, about $0.0004 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
compose:tdd
Use when implementing any feature or bugfix, before writing implementation code.
test-driven-development
TDD: enforce RED-GREEN-REFACTOR, tests before code.
engineering-workflow
可靠工程工作流的执行细则——探索→计划→TDD→系统化调试→代码审查→完成前验证。供方法论教练在开发任务中逐步引导。.
draw-image
Generate an image from a text prompt using an OpenAI-compatible image generation API (gpt-image-1-mini or compatible). The image is uploaded to the gofile.io public file sharing service and ONLY the public download page URL is returned. Trigger when user asks to draw, paint, generate, or create an image.
compose:subagent
Use when executing implementation plans with independent tasks in the current session.
compose:worktree
Use when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback.