/ar:resume — advance the loop
Iterate until the target is met, the budget is spent, or the run is stopped.
First action, always (MANDATORY)
Read state from disk before saying or doing anything else. Context is never the
carrier — this is what makes the loop survive compaction and session resets.
cat ./.ar/ar.jsonl 2>/dev/null | tail -50; git branch --show-current
No ar.jsonl means no run exists — say so and point at /ar:start. Otherwise
reconstruct best-so-far, run count, plateau streak and budget per
${CLAUDE_PLUGIN_ROOT}/references/resume-loop.md §1, then print the status block
before the first iteration.
Hard rules
- Never act on a branch this run did not create. If the current branch is
not
config.branch, do not switch onto it — create a fresh branch from
the present state and open a new segment (references/protocol.md §0).
- One atomic change per iteration. No compound edits. Two changes give one
number and no attribution.
- Commit before measuring, with a
Result: pending trailer; amend it with
the real result on keep.
- Revert with
git checkout -- . && git clean -fd on discard or crash.
Never -fdx — that flag deletes ./.ar/ and the run with it.
- Locked harness. Any diff touching
benchmark.sh, checks.sh,
evaluator.py or the metric-emitting code is rejected and the hypothesis
abandoned. Optimising a number while free to redefine it is not optimisation.
- Keep only above the noise floor. Improvement must exceed
noiseFloor × noiseFloorMultiple; borderline candidates get a multi-seed
re-run before the verdict, not after.
- Respect the budget —
maxRuns, maxSeconds, targetMetric.
- Defer execution. Print the measurement command; do not launch long jobs.
- Do not stop to ask permission. Once looping, keep going until the target
is met, the budget is exhausted,
/ar:stop fires, or interruption.
The iteration
Pick one hypothesis → apply one change → commit pending → measure → gate on
checks.sh → decide keep/discard/crash/checks_failed → amend or revert → append
to ar.jsonl and results.tsv, update research.md, worklog.md, ideas.md →
next. Three consecutive non-improvements switch strategy family; five propose a
paradigm shift. Full detail in references/protocol.md §2–§3.
On any stopping condition, write ./.ar/final_report.md and print — do not run —
git switch <originBranch>.
Read on demand
| Need |
File |
| Iteration steps, plateau, stopping |
${CLAUDE_PLUGIN_ROOT}/references/protocol.md |
| Keep threshold, confidence, seed re-runs |
${CLAUDE_PLUGIN_ROOT}/references/statistics.md |
| Record formats, metric contracts |
${CLAUDE_PLUGIN_ROOT}/references/state-schema.md |
| State reconstruction, overnight runs |
${CLAUDE_PLUGIN_ROOT}/references/resume-loop.md |
Read a reference when the phase that needs it is reached — not up front.
Completion status
End with a terminal status token as the last line of your reply — DONE,
DONE_WITH_CONCERNS, BLOCKED, or NEEDS_CONTEXT — per
references/completion-status.md. For /ar:resume, DONE after the iteration
is measured and kept or reverted (or a stopping condition is hit and the report
written); DONE_WITH_CONCERNS if the harness was flaky or a seed disagreed;
BLOCKED if checks.sh or benchmark.sh cannot run; NEEDS_CONTEXT if the run
state is missing or /ar:start has not run.
1---2name: resume3description: Run the autoresearch experiment loop — propose one change, measure it, keep it only if it beats the noise floor, repeat. Use to resume or continue an autoresearch run, for `/ar:resume`, for "keep iterating", "next experiment", "continue optimising", or whenever `./.ar/ar.jsonl` exists and the loop should advance. Reconstructs everything from disk, so it survives compaction, `/clear`, and a fresh session. Requires `/ar:start` to have run first.4---56# /ar:resume — advance the loop78Iterate until the target is met, the budget is spent, or the run is stopped.910## First action, always (MANDATORY)1112Read state from disk before saying or doing anything else. Context is never the13carrier — this is what makes the loop survive compaction and session resets.1415```bash16cat ./.ar/ar.jsonl 2>/dev/null | tail -50; git branch --show-current17```1819No `ar.jsonl` means no run exists — say so and point at `/ar:start`. Otherwise20reconstruct best-so-far, run count, plateau streak and budget per21`${CLAUDE_PLUGIN_ROOT}/references/resume-loop.md` §1, then print the status block22before the first iteration.2324## Hard rules25261. **Never act on a branch this run did not create.** If the current branch is27 not `config.branch`, do **not** switch onto it — create a fresh branch from28 the present state and open a new segment (`references/protocol.md` §0).292. **One atomic change per iteration.** No compound edits. Two changes give one30 number and no attribution.313. **Commit before measuring**, with a `Result: pending` trailer; amend it with32 the real result on keep.334. **Revert with `git checkout -- . && git clean -fd`** on discard or crash.34 Never `-fdx` — that flag deletes `./.ar/` and the run with it.355. **Locked harness.** Any diff touching `benchmark.sh`, `checks.sh`,36 `evaluator.py` or the metric-emitting code is rejected and the hypothesis37 abandoned. Optimising a number while free to redefine it is not optimisation.386. **Keep only above the noise floor.** Improvement must exceed39 `noiseFloor × noiseFloorMultiple`; borderline candidates get a multi-seed40 re-run _before_ the verdict, not after.417. **Respect the budget** — `maxRuns`, `maxSeconds`, `targetMetric`.428. **Defer execution.** Print the measurement command; do not launch long jobs.439. **Do not stop to ask permission.** Once looping, keep going until the target44 is met, the budget is exhausted, `/ar:stop` fires, or interruption.4546## The iteration4748Pick one hypothesis → apply one change → commit pending → measure → gate on49`checks.sh` → decide keep/discard/crash/checks_failed → amend or revert → append50to `ar.jsonl` and `results.tsv`, update `research.md`, `worklog.md`, `ideas.md` →51next. Three consecutive non-improvements switch strategy family; five propose a52paradigm shift. Full detail in `references/protocol.md` §2–§3.5354On any stopping condition, write `./.ar/final_report.md` and print — do not run —55`git switch <originBranch>`.5657## Read on demand5859| Need | File |60| ---------------------------------------- | -------------------------------------------------- |61| Iteration steps, plateau, stopping | `${CLAUDE_PLUGIN_ROOT}/references/protocol.md` |62| Keep threshold, confidence, seed re-runs | `${CLAUDE_PLUGIN_ROOT}/references/statistics.md` |63| Record formats, metric contracts | `${CLAUDE_PLUGIN_ROOT}/references/state-schema.md` |64| State reconstruction, overnight runs | `${CLAUDE_PLUGIN_ROOT}/references/resume-loop.md` |6566Read a reference when the phase that needs it is reached — not up front.6768## Completion status6970End with a terminal status token as the last line of your reply — `DONE`,71`DONE_WITH_CONCERNS`, `BLOCKED`, or `NEEDS_CONTEXT` — per72`references/completion-status.md`. For `/ar:resume`, `DONE` after the iteration73is measured and kept or reverted (or a stopping condition is hit and the report74written); `DONE_WITH_CONCERNS` if the harness was flaky or a seed disagreed;75`BLOCKED` if `checks.sh` or `benchmark.sh` cannot run; `NEEDS_CONTEXT` if the run76state is missing or `/ar:start` has not run.
emaballarin