Two halves, and a person between them
Not an engine. Each stage is a skill with its own file; load that file and follow it, the
way the generated slash command already says. There is no state file, no new verb and
nothing here that decides anything a stage would not decide on its own.
What it produces
Half one ends with the brief page a person decides on — docs/solution-intent.html,
regenerated by ai-eng report intent --html. Half two ends with the green gate output,
shown in the conversation, or with a page that says exactly why it is not green.
First half — up to the person
/ai-cycle <what you want>
The order of the halves is data, not prose here: read policy/skill-sequence.toml and
follow it. First half — research, then spec, then challenge, then council, then a brief
published as a page: what is proposed, what was refuted, what is missing, and what it
costs. Then stop.
Stopping is the whole of this half. ai-eng report blocked already lists a drafted
specification nobody has approved, so the halt exists and fires; this hands the person the
thing they need to decide instead of a status line. The approval is a record with the
specification's exact digest in it, and until that record exists the second half has
nothing to run against — the task envelope refuses outright when the bytes moved.
Second half — after them
/ai-cycle build <NNN>
The same map's second half, in order, once an approval record carrying the
specification's exact digest exists — the map's [gate] section is the line between the
halves. Inside ai-build, run one task at a time through ai-eng spec show <NNN> --task <n>: two kilobytes, not the whole plan. Where the plan's order allows two tasks
at once and the host can start two agents, start two; where it cannot, do them in turn.
The critics — review, verify, security — run each in its own context, then ai-eng audit verify, then ai-ship.
When the gate goes red
It does not stop at the first red and it does not get to cheat.
just stops at the first failing recipe, so a red gate has exactly one guilty recipe.
An attempt is counted per task and failing recipe, never per message and never per
gate run: counting messages would give one mistake that broke six tests twelve attempts,
and counting runs lets somebody else's failure eat this task's budget.
- Before spending an attempt, run the same command again, unchanged, the way the gate runs
it — in parallel, because a test that only fails in parallel fails where it matters.
Green on the repeat spends nothing and is written down as non-determinism. Two of those
notes on the same test in one build is not repair any more: escalate.
- Two attempts. Only a whole green gate resets the count. A recipe that went green and
goes red again keeps its count — measured: one review went red three times in a row
before passing, and resetting would have spent six attempts and never escalated.
- Never repair by silencing: no
--no-verify, no suppression comment, no skip mark, no
loosened bound, no raised ceiling. If the honest fix is a raised ceiling, that is the
escalation, with the arithmetic.
- If the count cannot be read, treat the attempts as spent and escalate.
- Escalating means a page a person can act on: what failed, what was tried, what it would
cost to fix properly. Never a terminal paragraph and never a silent stop.
What this is not
- "The stages are ceremony; the person wants the result" — the order is data, not prose, and the second half refuses to run until an approval record carrying the specification's exact digest exists.
Done when
Half one ends with a brief and no approval written by anybody here. Half two ends with a
green gate whose output is shown, or with a page saying exactly why it is not green.
1---2name: ai-cycle3description: >-4 Walks one request through the governed cycle by loading each stage's own skill body and5 following it — research, spec, challenge, council, then a brief a person reads. After6 that person says go, `ai-cycle build <NNN>` runs the second half: build, review, verify,7 security, audit, ship. Trigger for "run the whole cycle on this", "/ai-cycle build 021".8 Not for one stage on its own — call that stage directly. Not for approving anything: it9 stops at the brief and has no field in which an approval could be written.10license: Apache-2.011compatibility: needs git12disable-model-invocation: true13metadata:14 source: https://skillmd.com/skills/arcasilesgroup/ai-cycle15---1617# Two halves, and a person between them1819Not an engine. Each stage is a skill with its own file; load that file and follow it, the20way the generated slash command already says. There is no state file, no new verb and21nothing here that decides anything a stage would not decide on its own.2223## What it produces2425Half one ends with the brief page a person decides on — `docs/solution-intent.html`,26regenerated by `ai-eng report intent --html`. Half two ends with the green gate output,27shown in the conversation, or with a page that says exactly why it is not green.2829## First half — up to the person3031`/ai-cycle <what you want>`3233The order of the halves is data, not prose here: read `policy/skill-sequence.toml` and34follow it. First half — research, then spec, then challenge, then council, then a brief35published as a page: what is proposed, what was refuted, what is missing, and what it36costs. Then **stop**.3738Stopping is the whole of this half. `ai-eng report blocked` already lists a drafted39specification nobody has approved, so the halt exists and fires; this hands the person the40thing they need to decide instead of a status line. The approval is a record with the41specification's exact digest in it, and until that record exists the second half has42nothing to run against — the task envelope refuses outright when the bytes moved.4344## Second half — after them4546`/ai-cycle build <NNN>`4748The same map's second half, in order, once an approval record carrying the49specification's exact digest exists — the map's `[gate]` section is the line between the50halves. Inside `ai-build`, run one task at a time through `ai-eng spec show <NNN>51--task <n>`: two kilobytes, not the whole plan. Where the plan's order allows two tasks52at once and the host can start two agents, start two; where it cannot, do them in turn.53The critics — review, verify, security — run each in its own context, then `ai-eng audit54verify`, then `ai-ship`.5556## When the gate goes red5758It does not stop at the first red and it does not get to cheat.5960- `just` stops at the first failing recipe, so a red gate has exactly one guilty recipe.61 An attempt is counted per **task and failing recipe**, never per message and never per62 gate run: counting messages would give one mistake that broke six tests twelve attempts,63 and counting runs lets somebody else's failure eat this task's budget.64- Before spending an attempt, run the same command again, unchanged, the way the gate runs65 it — in parallel, because a test that only fails in parallel fails where it matters.66 Green on the repeat spends nothing and is written down as non-determinism. Two of those67 notes on the same test in one build is not repair any more: escalate.68- **Two attempts.** Only a whole green gate resets the count. A recipe that went green and69 goes red again keeps its count — measured: one review went red three times in a row70 before passing, and resetting would have spent six attempts and never escalated.71- Never repair by silencing: no `--no-verify`, no suppression comment, no skip mark, no72 loosened bound, no raised ceiling. If the honest fix is a raised ceiling, that is the73 escalation, with the arithmetic.74- If the count cannot be read, treat the attempts as spent and escalate.75- Escalating means a page a person can act on: what failed, what was tried, what it would76 cost to fix properly. Never a terminal paragraph and never a silent stop.7778## What this is not7980- "The stages are ceremony; the person wants the result" — the order is data, not prose, and the second half refuses to run until an approval record carrying the specification's exact digest exists.8182## Done when8384Half one ends with a brief and no approval written by anybody here. Half two ends with a85green gate whose output is shown, or with a page saying exactly why it is not green.