Skip to content

Releases

v0.88.10-alpha

Choose a tag to compare

@calionauta calionauta released this 16 Nov 18:54
0055d09

What changed

Added

  • A host without a command-prefix key can still win the bookkeeping. opencode's
    "shell" only names a binary, so the prefix is a two-line wrapper that exports
    BASH_ENV, plus a bashenv that sets pipefail and puts the tool directory on
    PATH. The same three checks apply, and they were run through the host
    (opencode run returned exit 1 for a failing pipeline) rather than read off the
    config schema. The "what a host must implement" table gains the row.

Changed

  • The harness reference now states the part prompt text cannot buy: a session that
    carried a "one call per question" rule made 85 consecutive direct calls and zero
    scripted ones. Bookkeeping belongs to the harness.

v0.88.9-alpha

Choose a tag to compare

@calionauta calionauta released this 16 Nov 18:47
aa748f9

Changed

  • The shell rule is an outcome, not a tool. The previous entry told the worker to
    put "two or more commands" into a codemode script. Measured over a 2 000-call
    session: 99% of calls already carried more than one command (4.2 on average), and
    codemode was written 0 times in 85 calls after the rule was added. A rule that
    prescribes an instrument is inert when the behaviour it wants already happens — so
    the entry now states the invariant ("one call per question, filter inside the call,
    one screen out") and names codemode only for what shell does badly: tool
    orchestration and JSON the shell would mangle.
  • harness.md puts the bookkeeping on the harness. ~30% of 8 265 shell segments
    were cd (2 005), export PATH (487) and timeout (810) — a stateless shell making
    every call restate where it is and what it can run. The fix is a shell prefix and a
    persistent cwd, plus set -o pipefail: a pipeline ending in | tail -2 reports the
    exit status of tail, which is how this project once tagged a release from a red
    suite.

v0.88.8-alpha

Choose a tag to compare

@calionauta calionauta released this 16 Nov 14:00
e72470f

Changed

  • harness.md says what codemode.mode: "only" actually bought, measured.
    Four runs on pi 1.5.2, one host, one model: on a task that wanted enumeration,
    "only" produced the right answer at 44.8k tokens against 35.9k and the same
    four calls; on a task that wanted one edit, three calls against five. It does
    not reduce round trips — the claim it is sold on — so the entry keeps "on"
    as the default and reserves "only" for volume jobs, with a note to measure
    before believing it.

v0.88.7-alpha

Choose a tag to compare

@calionauta calionauta released this 16 Nov 13:39
12eee23

Fixed

  • harness.md no longer recommends -bash, because it amputates instead of
    enforcing.
    A codemode script can call only active direct tools plus
    codemode/deferred tools, so removing bash from defaultTools removes it
    from the session entirely: measured on pi 1.4.0 and 1.5.2, tools.bash does not exist, ALL_TOOLS lists no bash, and the agent is left with no shell. The
    entry now enables codemode and removes nothing, names codemode.mode: "only"
    as the one hard enforcement that keeps capability, and says what exposure: "codemode" would cost for a built-in (re-registering it, truncation and hooks
    included).
  • harness.md says what PTC buys: roughly 4.3M → 149k tokens on a measured
    50-item triage, break-even on a single small task, with the two failure modes
    worth naming (a script that silently covers half the items, and a wrong tool made
    cheaper to call).

v0.88.6-alpha

Choose a tag to compare

@calionauta calionauta released this 16 Nov 13:22
e74d2b8

Changed

  • The human-facing question stops naming a retired ladder. The run-knobs
    migration replaced the appetite, but ask-patterns.md still told the worker to
    ask about "Pattern 7 (Appetite)" and plan-critique to say the estimate was
    "bigger than the appetite" — against a config that no longer exposes one. The
    question now names the declared quality and exploration knobs and the surface
    that would come off, and the legacy appetite: line stays readable for
    projects that still carry it. Merged from the branch opened on 2026-10-27, after
    rebasing onto main.

v0.88.5-alpha

Choose a tag to compare

@calionauta calionauta released this 16 Nov 13:07
e680045

Changed

  • harness.md names the surgical enforcement, not the blunt one.
    {"defaultTools": ["+codemode", "-bash"]} changes the inherited selection, so
    read/edit/write stay declared and the shell becomes reachable only from a
    script; codemode.mode: "only" hides everything, including the exact-match edit
    that a script does worse. Verified on pi 0.88: a fresh session lists
    read, edit, write, codemode, no bash.
  • opencode 2.3.0 needs no flag — Code Mode is built in (experimental.codemode_vm
    is deprecated and ignored, "Code Mode always runs on the VM") and MCP servers are
    exposed through it by default. A per-tool deny needs a plugin plus a policies
    entry.

Fixed

  • AGENTS.md no longer states the navigation ladder as a reflex. ripwire and
    cymbal are conditional, with the situation that pays for each and the measured
    reason (up to 28% more cost, arXiv 2609.30725).

v0.88.4-alpha

Choose a tag to compare

@calionauta calionauta released this 16 Nov 12:56
26b3a9f

Added

  • cli-tools/harness.md: the three harness capabilities a stage cannot ask
    for, with the setting that enforces each.
    A prompt can request economy; only
    the harness can make it true. Programmatic tool calling is the big one —
    {"defaultTools": ["+codemode"], "codemode": {"mode": "only"}} hides the
    built-in tools and makes the model reach them through scripts, where the
    default "on" only appends a note to each description. The file also covers
    the two-tool MCP surface (search_tools + execute), the workdir-script
    fallback when a host has neither, tool-result pruning, why a gate must name a
    workspace member rather than a role, and the two measurements to repeat on the
    next card.

Changed

  • code-map.md says when a navigation rung costs more than it saves. A map
    says where; the edit still needs the code, so a rung pays only when it
    replaces a search. Named symbol → ffgrep + read --offset: two calls, the
    second returns the code. ripwire --for for an unfamiliar repo touching three
    or more files. Structure-aware retrieval raised agent cost by up to 28% on
    SWE-bench Verified (arXiv 2609.30725) — the
    extra hop is the mechanism, so the ladder no longer chains ripwire into cymbal
    by reflex.

Fixed

  • npm run build failed on types/stages.ts — StageSkipCondition's index
    signature did not admit undefined, which its optional any_of requires. CI
    runs typecheck, not build, so the documented command was broken without a
    red pipeline.

v0.88.3-alpha

Choose a tag to compare

@calionauta calionauta released this 15 Nov 23:33
5ad7e39

Added

  • doctor checks that a recipe declaration has a file. stages.yaml names
    a recipe per stage and a runner reads a 404 as "this stage has no recipe", so
    a declaration with nothing behind it was silent. doctor now resolves every
    execution.recipe: against recipes/<id>.yaml and reports recipe-missing
    (error) with stage, recipe and the path it looked in, recipes-dir-missing
    when the directory itself is gone, and stages-missing /
    stages-unreadable when the declaration file cannot be read. A recipe: the
    reader cannot bind — outside an execution: mapping, or inside an
    execution: written inline on one line — is stages-unparsed (warn) naming
    the line, never a finding charged to the stage above it. STELOW_STAGES and
    STELOW_RECIPES override the two paths (documented in stelow doctor
    schema output).
  • A check that could not run doesn't read as a clean bill. Absent stages
    file is stages-unchecked (info), an unreadable one is an error, and
    doctor: ok (no drift) prints only when there are no findings at all. Two
    warn classes were also undocumented: orphan-pid and transitions-missing.

v0.88.2-alpha

Choose a tag to compare

@calionauta calionauta released this 15 Nov 17:49
9fadd4c
  • Number the promises. The spec mints a stable id per promise (S1, S2), and design, tech, build and audit cite the id instead of restating the promise. On the Multica smoke run the audit had to translate between its own findings and the spec's promises, because both sides described them in prose: a requirement with two names is a requirement nobody can trace.

v0.88.1-alpha

Choose a tag to compare

@calionauta calionauta released this 15 Nov 16:22
9282f7e

The English pass in 0.88.0 was checked with an accent scan, and an accent scan
cannot see Mesmo ator, Etapa, Nenhum or compras corporativas. Those
survived into the rendered contracts, which is how a host ends up enforcing a
vocabulary the methodology claims it retired.

Fixed

  • The Portuguese that carried no accent. 05-job-actors.md kept Mesmo ator
    in five places (including a criterion another host reads as data),
    09-emotional-social-jobs.md kept a whole table header and the Nenhum
    status, 11-switch-interview.md kept Nenhuma, and two workflow references
    (plan-critique-context.md, step-3-feature.md) kept Portuguese instructions
    and step names. All English now.

  • docs/study-goal-vs-plannotator.md was a Portuguese study note in a
    repository that says documents are English.

Added

  • scripts/check-language.sh, wired into verify:generated. It looks for
    Portuguese words and phrases, not for accents, skips the frozen archive, the
    changelog and URLs, and fails on a hit. A future translation pass that misses
    an unaccented word now fails in CI instead of shipping in the data.