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/addyosmani/factory/factorygit clone --depth 1 https://github.com/addyosmani/factoryWhat 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 | $0.00022 | $0.00668 |
| Opus 5 | $0.00011 | $0.00334 |
| Sonnet 5 | $0.00004 | $0.00134 |
| Haiku 4.5 | $0.00002 | $0.00067 |
Grade A, and why
factory 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 — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the factory control room. The human wants to know what is going on and what needs them, quickly.
Read docs/factory/CONTRACT.md and docs/factory/CHARTER.md. Query open issues with
factory:* labels and open factory PRs, then read the newest records under
docs/factory/runs/. Treat QUEUE.md and STATE.md as snapshots only.
If the user passed an argument, handle it:
status(or no argument) - the report belownext- the single highest-value thing for a human to do right now, and whyqueue- the full queue grouped by dispositionstuck- only items blocked, stale, or twice-rejected<issue-number>- everything about that one item
The status report
Lead with what needs a human. Nothing else is urgent.
FACTORY - <repo> - tier: <tier>
NEEDS YOU (<n>)
PR #<n> <title>
<why it needs you: load-bearing / tests modified / gate skipped / verifier reservations>
FQ-<n> needs-info: <the actual question>
REVIEW QUEUE: <n> / <charter limit>
<if at or over the limit, say plainly: the factory should stop taking new work
until this drains. The constraint is not how many agents can run, it is how
many decisions are pending your judgment.>
RUNNING
<issues labeled factory:in-progress and visible routine runs, with links>
QUEUE
ready-to-implement <n>
ready-to-spec <n>
needs-info <n>
wait-to-implement <n>
awaiting-review <n>
HEALTH
gates on main: GREEN | RED (<failing>) | MISCONFIGURED (<missing>)
skipped gates: <list, or none>
charter gaps: <n> <- unreviewed decisions the factory could not make
last monitor: <date> <- if over a week, say the loop is not closed
FLOW (last 30 days, when records are complete)
verifier rejection rate: <n>%
median review wait: <duration>
escaped defects: <n>
SUGGESTED NEXT
<one thing, with the reason>
Rules
- Never merge, approve, or close anything from this command. It reports.
- If the review queue is at the charter limit, lead with that above everything else. A full review queue is the binding constraint on the whole factory and everything else is noise until it drains.
- If gates are red on the default branch, lead with that instead: every verdict since it broke was measured against a broken baseline.
- Be terse. This is a dashboard. Long prose defeats the purpose.
- If no run record has landed in over a week, say so. Stale evidence reads as calm.
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 · 73 lines · 22 tokens per session scan A a92aef3eaba7
factory is a command published in the GitHub repository addyosmani/factory (168 stars, last pushed 11d ago), licensed MIT. It adds 22 tokens to every session and 668 once invoked, about $0.0001 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 2026-08-30.
Other commands, from other repositories
implement
Read the Linear issue $ARGUMENTS using the Linear MCP tool.
create-tickets
Read the spec file at $ARGUMENTS.
ox-session-list
belongs in the ox CLI JSON output (guidance field), not here. Skills are agent-specific wrappers; ox serves all agents (Codex, etc.). --> List recent sessions from the project ledger and offer to view one.
session-list
List recent sessions from the project ledger and offer to view one.
ox-status
Check SageOx project status including authentication, sync, and daemon health.
cook
Execute the entire todo/in-progress backlog sequentially via ccf-implementer, then batch-verify (review + code-review in parallel, simplify, re-gate, updatespec).