Repository navigation
Added
-
agentdescent github publish|status: publish one accepted candidate as a
proposal branch, commit, pull request and optional check run, and read a PR's
state and checks back. Publishing is a plan by default (--applywrites), the
token comes only from an environment variable, the base branch is never written,
and an existing proposal PR is updated rather than duplicated. The PR body
carries the run/artifact ids and evidence hash; there is no auto-approval,
auto-merge or branch-protection change. -
agentdescent github review: read a PR's reviews, inline comments and check
runs into one packet, extract classified review constraints (test, security,
api, docs, style, behavior), and report merge readiness with blockers. It is
read-only, never approves or merges, and requires an explicitrequires_human_click
step;CHANGES_REQUESTEDand its inline comments are blocking, plain notes are
not turned into tasks, resolved ids can be supplied because REST does not
expose thread resolution, and only a reviewer's decisive review counts (a later plain
comment never lifts a block), so a change request that reviewer later
approved or had dismissed no longer blocks. -
agentdescent skills report|audit|search: a durable skill bank with an outcome
ledger (uses/successes/failures), a bank-relative lift audit that
quarantines underperforming skills (read-only unless--apply), active-only
retrieval through the artifact graph, and explicit retire/activate lifecycle
states. The store is an atomically written JSON snapshot; no model or network
call is made. -
agentdescent hgm next|report: HGM's missing evaluation policy and
expand-vs-evaluate scheduler over a JSON tree. The planner applies the
n_task_evals ** alphabudget split, Thompson sampling on clade evidence for
expansion, own evidence for evaluation, optional cool-down, and unsubmitted-task
selection; it stops at the budget or when every task is spent. Utility measures
are validated as[0, 1]; no model or network call is made. -
agentdescent harness manifest|checkandHarnessManifest: a file-level
component contract for harness evolution. Components name path globs, an RRSI/SHE
responsibility, editable/frozen state and per-component file/line budgets; the
manifest also has global budgets, allowed/forbidden paths and aFileTreekwargs
projection.checkfails closed on frozen, forbidden, unowned or over-budget
edits; no model or network call is made. -
agentdescent compile-skillandcompile_skill: turn a skill document into a
single tested function through a caller-suppliedtext -> sourcecompiler,
behind the framework AST gate, with repeatable tests and optional
reference/case equivalence. Any gate, test or equivalence failure returns
status: "text"with a reason and leaves the skill document as the fallback; no
model or network call is made by the module. -
Documented the contributor decision tree for Algorithm card vs MethodPolicy vs full
examples/port vsaggregator_factory(docs/algorithm-decision-tree.md), cross-linked from authoring / port-fidelity / aggregator-factory.plannow warns when acceptance/promotion (as well as conflict/fusion) are installed on a factory path and will be ignored — engine composition without silent drop remains #807. -
kind=skill_libraryis a first-class EvolveSpec / CLI / MCP kind:FileTree+
tree_runner(layout="skill_library"),init --kind skill_libraryseeds ≥2
*/SKILL.md, andplanechoes the resolved layout. See
examples.skill_library_evolution(#820). -
agentdescent apply --install-to PATH(and MCPinstall_to) installs an evolved
skill_libraryinto a host skills root such as~/.claude/skillswith the same
write_tobackup semantics and without pruning unrelated host skills (#823). -
agentdescent demo --algorithmruns offline for gepa, dgm, ace, skillopt and
rrsi on a text fixture (archive_selection stays on the skill_dir demo);
evoskill is refused with an explicit reason (#845). -
agentdescent demo --kind skill_libraryruns the offline multi-file
*/SKILL.mdlibrary demo from an install (no source checkout, no key).
The defaultdemosmoke note and README / quickstart-skill / usage now
point at that command;python -m examples.skill_library_evolutionstays
as the source-checkout CLI (--dry-run,--agent,--install-to) (#878). -
status,watch,show,apply,cancel,resumeanddiagnoseaccept
latest(orlast) in place of a run id and use the newest run in the store,
soagentdescent show latestworks straight after a demo; an empty store
says so instead ofno run 'latest'(#892). -
MCP Path A now has a
demotool (algorithm, optionalkind) that returns
the same payload asagentdescent demo --json, so a host can run the zero-key
smoke without a provider key. The shared skill tells hosts to trydemo(or
agentdescent demo --json) before doctor → plan → start (#903). -
The read-only run store streams a run's status and new rounds over Server-Sent
Events atGET /api/runs/<id>/events, withLast-Event-IDresume, so a host
panel or platform can follow a run without polling (#922). -
agentdescent demo --algorithm evoskill(anddemo --kind skill_library --algorithm evoskill) now runs the evoskill card offline, with no API key: the demo reflector speaks the Proposer/Generator protocol and induces a*/SKILL.mdwith frontmatter, held-out reward rises, and outcomes includeevoskill-select. Documentation no longer says evoskill is refused in the offline demo (#934). -
RunManifest.replay_eligible/replay_notesand an optionalreplayrecord
(agentdescent.replay_fidelity.ReplayFidelity) that captures what a replay needs
and the manifest's input hashes do not cover: RNG seed/state digests at
interaction boundaries, an environment fingerprint (cwd, interpreter, tool
versions, and the names of the allowlisted environment variables present --
PYTHONHASHSEED,
PYTHONPATH,LANG,LC_ALL,TZ,OMP_NUM_THREADS), and hidden-state
digests. No value material is stored (not even hashed), a caller cannot widen the
allowlist, and a credential-shaped name is refused even when allowlisted.Trajectorycarries the same record (withreplay_eligible/
replay_notesproperties),ReplayLab(manifest=...)refuses a counterfactual
withReplayNotEligiblenaming what was not reconstructed, andreplay_check()
refuses to compare two runs when either side was not captured (#939). -
A run, a commit and a decision can now name the decision procedure that produced
them, as a content hash over thePoliciesbundle, the decision-relevant
AggregatorConfigfields and the verifier's knobs:
policy_digestonDecisionRecord,MergeContext,RunManifest(build_manifest)
andEvolutionResult, plus apolicy-digest:trailer on the ledger's commit
message. A decision taken under a different policy than the previous one for the
same artifact links to it (parent_id+evidence["policy_change"]), and
agentdescent.policies.policy_digest()computes the version for a hand-built
bundle, andStableDistanceAcceptance/AdaptiveTrustRegionnow expose their
own knobs to it. A decision record that does not state a policy keeps
policy_digestempty rather than claiming the shipped default, and
run_guarded()records the digest its result carried. Additive: empty by
default, unchanged commit messages and unchangedmanifest_id(#943). -
A promotion decision now sees what reviewing the change costs a human: the new
ReviewCostrecords the diff's files, lines, blast radius, coarse risk tier and
evidence count and estimates review minutes from documented constants, and
PromotionRulesgains the opt-inmax_review_cost/max_review_minuteslimits
(both default off, so with them unset the decision and its hashes are unchanged)
that defer a disproportionate change and recommend batching it. The decision's
evidence reportsattention_per_accepted_gainwhen a cost and a positive eval
delta are present (#950).
Changed
-
Hard cards vs ports contract: Path A names wheel Algorithm cards only;
examples/research ports stay source-only CLI modules (same slug ≠ same
fidelity). Decision tree, README, plugins, SKILL.md, and port-fidelity updated;
future MethodPolicy card candidates listed as scoped follow-ons (#384). -
algorithms list/show, MCPlist_algorithms/describe_algorithm, andplan's algorithm echo now label every registry entry as an Algorithm card (surface=card), filter-friendly by hostkind, and — for ACE / GEPA / DGM / SkillOpt / EvoSkill — name the siblingexamples/<algo>/research port with an honestfidelity_noteso same name is never read as same fidelity. -
The homepage interactive demo now replays a session the way the shipped skill runs one: each host's real slash command (or plain request for DeepSeek Harness), default skill folder and MCP tool naming; doctor, spec, plan with the real call estimate, a start approval, status polls with a round/phase/held-out progress panel,
show diff=truewith a unified diff, and a separately approved apply with the real backup location. Still a browser-only sample; no model calls. -
The README is reorganised around install, a 60-second offline quickstart,
calling any named algorithm, adding one, and a host table (Claude Code, Codex,
OpenCode, DeepSeek Harness, Grok Bot).docs/testing-the-plugins.mdis merged
intodocs/testing-guide.md, and stale port counts, version references and
duplicated install text across the docs are fixed (#797). -
The MCP server now exposes Path A by default (doctor → plan → start → status/show → apply, plus cancel/resume, diagnose, and algorithm list/describe/spec). Audit, release, platform, experience and the other satellites stay callable with the same names when
agentdescent mcp --advancedorAGENTDESCENT_MCP_ADVANCED=1is set. Docs:docs/mcp-path-a.md. -
evolve/async_evolveaccept typedBudgetSettings,RuntimeSettingsandObservabilitySettingsbags; flat kwargs stay as shortcuts. Docs andEVOLVE_PARAM_ALLOWLISTfreeze the flat surface so new knobs go in Policies or a bag (#804). -
aggregator_factory=now forwards merge-sidePoliciesinto the factory and
applies them on aggregators that expose the slots; leaving them unused raises
unlessaccept_orphaned_merge_policies=True. GEPA and SkillOpt factories honour
acceptance/fusion(#807). -
ProposalPolicymay return multiple strings:evolve/async_evolveingest
each as its own evidence card, passctx.kfrompolicy.k, and no longer raise
ProposalContractErroronlen > 1. Self-verify, when enabled, runs per
proposal (#808). -
evolve(asynchronous=True)is a deprecated bridge toasync_evolve: it now
requires explicitmax_secondsandmax_rollouts(no silent 20s default, no
roundsreinterpretation) and emitsFutureWarning. Policies wiring is one
shared_POLICY_DRIVER_SUPPORTtable for both drivers (#809). -
Flagship skill demo is now
examples/skill_library_evolution.py: a multi-file Agent Skill library (layout="skill_library", offline agent must read skills off disk,--dry-run/--install-to).examples/skill_evolution.pyand README /docs/quickstart-skill.mdare retitled as instruction / playbook evolution so SingleSlot / AppendRules are not called "a skill" without qualification (#818). -
MethodPolicy's keyed string bank is renamed
KeyedSkillBank(aliasSkillLibrary
kept). Voyager / SkillWeaver docs and imports no longer call it an Agent Skill
library; disk libraries point atskill_library_evolution/ EvoSkill (#821). -
Decision tree and aggregator-factory prose match closed #807 (compose/raise, no
stale “until”); quickstart-skill titles say instruction / one skill / library;
README listspolicy_slotand points at #820 forskill_library; GEPA docs /
print paths say prompt/instruction, not skill folder (#822). -
algorithm: evoskillonkind=skill_libraryselectsSkillLibraryTreeLIBRARY
layout (skills/*/SKILL.md);skill_dirstays SINGLE_SKILL.plannotes name
the layout. Cardkindsincludeskill_library(#824). -
ACE card notes and summary state the
## generalfallback and that admit means
"cleared Jaccard dedup"; FiNER templates are labeled example-only;extra_calls
no longer saysnonewhile describing a reflection call (#842). -
Docs state that Path A reported held-out reward is the select-set score (optimistic); a locked test split remains #843.
-
evolve(test_frac=...)can carve a test partition out of the held-out tail:
accept/select keep reading the select split, the test split is scored once at
the end and reported astest_reward, so the printed reward is no longer the
score of the set that chose the winner. With no split,evolve,showand
diagnosenow say plainly that the reported reward is the select set, not an
unseen test set (#843). -
A text-only Algorithm card asked to run on a
skill_library(or any other tree
kind) now says so in the kind-mismatch error, before the list of cards that do
apply:'gepa' is text-only: it reads a single text file, not a skill_library target. The list of applicable cards and the one-file--kind textroute are
unchanged (#846). -
Algorithm decision tree notes which cards are text-only vs
skill_library-capable (#846). -
agentdescent demolabels itself a smoke trajectory and points at
examples.skill_library_evolutionfor multi-round library edits; that offline
path now edits a*/SKILL.mdbody (STATUS draft→ready) and
init --algorithm evoskilldefaultsbatch_sizeto 1 (#847). -
init --kind skill_librarytips name thearchive_selectionPath A card
(andalgorithms list --kind skill_library); quickstart-skill, decision-tree,
and algorithm-authoring kinds stay aligned (#864, Refs #846). -
An unknown algorithm name that is a near-miss typo of a registered card now ends
with "did you mean 'gepa'?" (algorithms show/run, specs, MCP), instead of
only listing every card (#874). -
agentdescent planand the plan step ofalgorithms runprint a short human
summary by default -- task and round counts, the agent-call upper bound, the
agent, the small-held-out warningevolveraises, and thenext:command --
instead of the whole plan document.--jsonis unchanged (#919). -
TrustEvidencecan now carry aRunHealthrecord (attempted,
failed_tasks,threshold): failed tasks are dropped from the paired scores
instead of being folded in as zeroes, so the reported gain is the mean over the
tasks that answered, and a run whose failure rate is above the threshold fails
its eval gate, sopromote()cannot returnpromote. A bundle built without
health data keeps today's keys, values and hash (#993).
Fixed
-
The skill-library seed (
init <target> --kind skill_libraryand
algorithms run <name> --kind skill_library) now refuses an unwritable target
as oneerror: cannot write <path>: <reason>line with exit 2, or
{"ok": false, "error": ...}under--json, instead of a traceback. Both
destinations are preflighted before anything is written, so a refused--out
leaves no seeded*/SKILL.mdbehind; a writable target still seeds its
starters and the#868one-starter note is unchanged (#1003). -
An expired or unprovisioned subscription refused by the provider -- Ark
Coding's HTTP 400InvalidSubscription, OpenAI'saccount_deactivated,
Anthropic'sbilling_error-- now fails the first call asSubscriptionInactive
instead of being retried as a transientRuntimeError. The run stops within
seconds instead of after three attempts per call, the error line names the
subscription as the cause,diagnoserepeats that next step instead of
recommendingdoctor, andresumerefuses it without--force(#1006). -
--budget/budget_usdis only checked between rounds:plan,evolve, and
MCPstartnow warn when the budget is below the seed+one-round call upper
bound (so the first round alone can overshoot), and the plugin quickstart /
plugins docs match that semantics (#764). -
ReflectiveFusion now ranks a synthesised union against its parent singles before the acceptance gate, so a merge that loses to a stronger parent cannot commit just because it still beats the seed (
trust_union=Truekeeps the older cheap unranked path). -
ReflectiveFusion's MERGE_PROMPT now receives compact parent cheap-score and rationale metadata, and instructs the model to pick one coherent parent for incompatible taxonomies/enums instead of OR-concatenating them (#767).
-
plan/evolve/ MCP start now hard-refuse a bare isolatedopencodeworker
when a gateway key (e.g.ARK_API_KEY) is set, instead of soft-warning and then
burning calls on OpenCode's unactivated Volcengine default. Init/algorithms run
warnings say fillingreflectalone is not enough — pass--model/--base-url
(or write the Coding recipe). Wired--model/--base-urlspecs stay green (#790). -
Missing-CLI plan/algorithms-run errors for directory kinds no longer suggest
openai_compatible(which those kinds refuse); they point at WorkspaceAgent
CLIs instead. kind=text / reflect remedies are unchanged (#791). -
agentdescent algorithms list --kind policy_slotnow prints the same explanation
MCPlist_algorithmsalready returns (that kind evolves a policy slot andplan
refuses an algorithm), instead of a bare "no algorithms for kind" line.--json
keeps the array shape and puts the note on stderr (#799). -
agentdescent algorithms show/run(and MCPdescribe_algorithm/
algorithm_spec) now tip that a documentedexamples/research port
(reflexion,textgrad,self-refine, …) is not a named Algorithm card —
pointing atdocs/algo-*.mdand the checkoutpython -m examples.<slug>…
path — instead of only listing the seven registered cards (#801). -
PopulationAggregatornow archives per-task held-out scores and fills
Candidate.per_task, soPolicies(selection=ParetoFrontier(...))can run
GEPA-class parent selection withoutaggregator_factory=(#806). -
doctorkeeps tipping Grok Bot operators to save the scaffold skill as a user skill whenuser_skillis missing (and appends the same follow-up on a stale/partial scaffold problem). The Bot loads the saved user skill, not the scaffold alone (#825). -
init/algorithms runwriting an Ark (*.volces.com)openai_compatibleblock now set"stream": trueby default so reasoning models likedeepseek-v4-flashdo not hit the 120s non-streaming propose timeout;planwarns when a hand-edited Ark block omits it, and the quickstart / plugins Ark examples match (#832). -
The offline
agentdescent demosample output in the plugin quickstart and
testing guide now shows the plan line the CLI actually prints (up to 68 agent calls, not 64) (#833). -
agentdescent show/show --json(and MCPshow) now surface
fusion_stats/fusion_summaryfrom reflective-merge ranking
(contested, unranked, losses) so Path A can see merge health after #766
without the ledger Python API (#834). -
kind=textnow usesSeededAppendRulesso a one-sentence reflector proposal
appends under# Learned rulesinstead of replacing the whole seed prompt
viaSingleSlot(#837). GEPA keepsSingleSlotwith a full-rewrite propose
template. -
evoskillonkind=skill_librarywrites top-level<slug>/SKILL.md(with
Agent Skills frontmatter) instead of nestedskills/<slug>/SKILL.md, so apply
into a host skills root stays loadable and edits reuse the seed path (#838). -
GEPA no longer stores unused
conflict/promotion/screenPolicies as
attributes, so #807's orphan check fails closed instead of reporting them
as active (#839). -
SkillOpt
LRSchedulerclamps the anneal fraction to[0, 1]so the edit
budget stays atmin_lraftertotal_stepsinstead of climbing back (#840). -
RRSI card defaults use
T=8, a smallnoise_band, and non-zero betas so a
short evolve run still exercises the anneal and cost gates (#841). -
DGM increments
childrenonly for archive admits; unused evoskillval_every
removed; GEPASelectionContext.headtracks the last selected parent (#844). -
algorithms runerror paths dedupe identical init/plan tips, and--help
names Path A vsalgorithms(#848). -
agentdescent demo --jsonnow emits a parseable result object (ok, run_id,
rewards, learned, plan, next) instead of ignoring the flag and printing the
human narrative (#862). -
agentdescent init <path>andagentdescent algorithms run NAME <path>now
guesskind=skill_libraryfor a directory with two or more*/SKILL.mdand
no rootSKILL.md, instead of falling through toskill_dir(#866). -
init --kind skill_library/algorithms runon a library path now name only
the starter*/SKILL.mdfolders actually written (e.g.example-alphawhen the
target already had one skill), instead of always claimingexample-alpha / example-beta(#868). -
When a GSM8K research port cannot fetch the dataset, the error now ends with
what to do next:pip install pyarrow(when that is why the parquet path
failed), setHF_ENDPOINTto a reachable mirror, or set
AGENTDESCENT_GSM8K_SAMPLE=1for the bundled offline sample (#870). -
The missing-reflector-key warning from
plan/algorithms runnow names how
long the run really lasts: "fails after one round" on a--rounds 1quick
trial instead of always "two rounds" (#872). -
Unknown algorithm parameter names now fold case and
-/_(batch-size→
batch_size) and a near-miss typo ends with "did you mean 'batch_size'?",
matching how algorithm names already resolve (#876). -
When a
kind=textcard or caller setsstrategy, plan no longer also
claimsSeededAppendRules.gepa/ace/dgm/skilloptdrop the
generic note;rrsiand a bare text spec still show it (#879). -
--kindand speckindnow fold case and-/_(skill-library→
skill_library) and a near-miss typo ends with "did you mean 'skill_library'?",
matching how algorithm and parameter names already resolve (#882). -
--agentnow folds case and-/_(claude-code→claude_code,
openai-compatible→openai_compatible), matching how algorithm names,
parameter names, and--kindalready resolve (#884). -
An unknown CLI verb no longer lists the internal
runworker among
choose fromchoices, and a near-miss (algoritms,compile_skill)
getsdid you mean …?(#886). -
Algorithm parameter values that declare
choicesnow fold case and-/_
(sigmoid-novelty→sigmoid_novelty,KEEP_ALL→keep_all) and a
near-miss typo ends with "did you mean 'novelty'?", matching how
parameter names already resolve (#888). -
uninstallno longer tells grok-bot, dsh, or OpenCode users to remove files
via a host marketplace those hosts do not have. The marketplace aside stays
only for Claude Code and Codex, which can be installed that way (#890). -
init --algorithmwrites the registered algorithm name into the spec, so
--algorithm GEPAorarchive-selectionis saved asgepa/
archive_selection(matchingalgorithms listandalgorithms run) instead
of the alias as typed (#894). -
show_payload/status_payload/apply_payloadand the MCPshow/status/apply/diagnose/cancel/resumetools acceptlatest(orlast) the same way the CLI does, so a host that asks to show the newest run no longer getsno run 'latest'(#896). -
agentdescent diagnosenow names the runnable next command with the resolved run id forsolved(show <id>,apply <id> --dry-run),cancelled(resume <id>),running(watch <id>) andpartial-improvement(show <id> --json), instead of prose such as "review and apply the artifact" (#898). -
agentdescent resume(CLI and the MCPresumetool) now refuses a run that is alreadydone [target_reward]and points atshow <id>/apply <id> --dry-run, instead of relaunching it, spending calls again and overwriting its recorded initial reward and outcomes with the no-op rerun. Runs that stopped on a bound, were cancelled or failed still resume (#900). -
agentdescent demo --algorithm NAMEnow ends with the one command that runs that algorithm on your own data (agentdescent algorithms run NAME <your-prompt.txt|your-skill-dir> <your-cases.jsonl> --model <model> ..., thenagentdescent evolve agentdescent.NAME.spec.json), in both the text footer and--jsonnext, instead of a genericinitthat dropped the algorithm.demowithout--algorithmis unchanged (#902). -
agentdescent demono longer leaves anagentdescent-demo-*directory in the
temporary path after every bare run: the default root is removed on the success
and the error path, while--dirkeeps the directory it names and--json
still reports adirthat exists. The quickstart line now says which case keeps
the files (#906). -
A run store that cannot be created (
--store,AGENTDESCENT_HOMEor the default
~/.agentdescent/runson a read-only, missing, unwritable or file path) is now a
RunStoreError: the CLI exits 2 with oneerror: cannot create run store ...
line, or{"ok": false, "error": ...}under--json, instead of a raw
FileNotFoundError/FileExistsError/PermissionErrortraceback (#907). -
The install FAQ's port taxonomy now names all four catalogue groups, Genesis
included, so the dry-run split adds up to the 23 ports the same paragraph
quotes instead of the 8 + 14 = 22 it printed (#908). -
agentdescent serve --host <non-loopback>now refuses to start, with one
error:-style line and exit 2, unless the operator passes--allow-remote: the
run panel has no authentication and lists every run in the store, which a single
--host 0.0.0.0used to publish to the network. Loopback hosts are unchanged,
and the opt-in prints the same no-authentication warningagentdescent mcp
does (#911). -
agentdescent github publishnow reads the repository'sdefault_branchand
refuses a proposal that names it, even when the proposal's own base is another
branch. An existing proposal branch is only fast-forwarded: a rewritten history
raises oneGitHubErrorand nothing is written, and no ref update is ever sent
withforce: true(#913). -
The review gate now counts a
Verdict: PASSwritten above itsTip SHA:line,
above a bareTip: <sha>line, or carrying(tip <sha>)inline, so a PASS at
the right tip is no longer read as "says nothing at this tip";What I ran (...)
counts as theRan:list. A PASS with noRan:/Not run:is still counted but
markedweak: no Ran:/Not run:(#915). -
doctornow flags an MCP launcher whose first argument is an absolute path that is no longer an executable file: the host readsmcp broken/state: partial(exit 1) instead of a silently greenmcp ok. The host line names the dead path, and the problems line tells you to re-runagentdescent install <host>. A relative or bare command such aspython3is not flagged; Grok Bot staysmcp n/a(#918). -
agentdescent resumeno longer relaunches a run whose last failure was a spent
provider quota: afailedrun whose error starts withQuotaExhaustedis
refused with exit 2 and the provider's own reset time in the message, plus
--forceto retry anyway, and the status file is left untouched (#926). Only
that case is gated -- a cancelled, bounded or ordinary provider failure still
resumes. The MCPresumetool takesforce=truefor the same opt-out. -
agentdescent demo --algorithm rrsiprinted what it learned as the internal
state mapping (state -> {'ra4bfacbc48': 'COLUMN: amount'}) in the human footer
and in the--jsonlearnedfield; both now name the rule text
(rules -> 'COLUMN: amount'), join several rules with|and count any
beyond the first two, so no internal id or Python repr reaches the reader and
the line matches whatshowrenders (#927). -
The review gate no longer treats a later
Ran:/What I ran:block as
strong just because an earlier block in the same body had aNot run:line;
hasNotRunnow resets when a later list starts, the same wayranalready
does (#931). -
agentdescent harness checkand theFileTreeit configures now share one
matcher (filetree.match_any), so a manifest cannot mean two different things
on the two surfaces. Every matching component applies -- read-only wins over
editable and a frozen catch-all can no longer be hidden behind an earlier one --
budgets are charged to each matching component, and*no longer crosses/
(**means any depth) (#932). -
github reviewno longer reportsready: falsebecause a reviewer's approval
body or a lifted (dismissed) review contains an action word such asadd,
requiredorcannot: the action-word promotion is now keyed on the review
state and applies to comments only, and a review body that is itself a
Verdict: PASSreport is never blocking. Those rows are still listed as
suggestions, andCHANGES_REQUESTED(with its inline comments) still blocks
(#933). -
agentdescent apply --install-to(and MCPapply(install_to=...)) now resolves known host names and aliases (claude-code/claude,codex,opencode/open-code,dsh/deepseek-harness) to their host skills root using the same path helpers and--home/ host env rules asinstallanddoctorwhen no matching path exists in the working directory, whilegrok-bot/grokrefuses writes, exits 2, and prints manual user-skill save paths (#935).agentdescent apply --jsonalso emits structured{"ok": false, "error": ...}payloads on errors like other CLI verbs. -
build_trust_evidence(..., artifact_changed=False)now reportsgain == 0and
noise_only is Truefor an artifact that is byte-identical before and after,
and such a bundle can no longer returnpromote: the zeroed eval delta is not a
pass, so evaluation noise is not repriced as learning (#937). -
The review skill now takes the merge lane the diff's risk tier gives it: a T0/T1
pull request is still auto-merged pinned to the reviewed tip, while a T2/T3 one
getsawaiting-approvaland waits for the maintainer's click instead of being
auto-merged by an agent (#961). -
append_provenancekeeps its records in the ledger's.git/directory instead
of the artifact work tree, sogit add -Acan no longer sweep provenance files
(and their.lock) into an artifact commit; anartifact_idcontaining a path
separator,..or an absolute prefix is rejected instead of escaping the
provenance root (#966). -
MCP
applyvalidates resolved destinations, changed tree entries and backup
paths against the operator's write allowlist before writing. The shared path
guard also resolves directory links, including Windows junctions, so an
allowed path spelling cannot redirect a write outside the configured roots (#969). -
evolve()cancels workers still waiting for a concurrency slot when their round
times out, so they cannot start new rollouts after the round or run has ended.
Calls already in flight may still finish (#970). -
Async runs now record workers that fail during initialization, task selection
or completion bookkeeping, and stop with an error when every worker has
retired. Caller contract errors propagate to the calling thread; rollout
backend retries and healthy workers keep their existing behavior (#971). -
EvalOps.run()now uses the default regression gate to setpassed, so a
single-task result no longer reports a pass thatEvalOps.gate()rejects.
Explicit gates that disable significance checks remain available. -
Use UTF-8 for CLI agent, code/plugin runner and command grader pipes regardless of the parent locale, preserving Unicode answers and JSON grading inputs. Invalid UTF-8 output raises an explicit adapter error instead of corrupting answers or scores (#973).
-
agentdescent serveSSE round events (GET /api/runs/<id>/events) now emit each row's 0-based append ordinal inrounds.jsonlas the eventid:instead ofrow["round"], keepingroundindataand ensuring resumed runs andLast-Event-IDreconnects never skip rows (#976). -
applyrefuses a symlink planted in the destination (--to) instead of
following it: the link itself, or any existing component under--to, is named
in the error, so a link atskills/SKILL.mdcan no longer redirect the write -
or the.bak-Ncopy taken before it - to a file outside--to. Containment is
checked on the leaf as well as its parent, and the destination rule now lives in
filetree.dest_problembeside the tree writer's boundary. The directory
backup slots scanbak-Nwithlexiststoo, so a dangling link there is
skipped instead of crashing the backup copy (#977). -
merge-lanereadsreview-policy.jsonandrisk-rules.jsonfrom the PR's base
commit instead of the clone it runs in, so a branch that edits either file can
no longer lend itself a PASS or a lower tier;.github/**and
.claude/skills/**are T3 by rule, and a base that cannot serve either file
prints no lane at all (#978). -
A refused artifact write —
init --out,algorithms run --out, or
apply --tofor a text file or a tree — now prints one
error: cannot write <path>: <reason>line and exits 2, or
{"ok": false, "error": ...}under--json, instead of raising the raw
OSErroras a traceback. The run store'scannot create run store ...
contract (#907) is unchanged (#990). -
The last artifact writes outside the run store now report a refused write as
oneerror: cannot write <path>: <reason>line with exit 2 instead of a
traceback:init --outover an existing spec (the.bak-Ncopy runs before the
wrappedopen), andapply --from <file>/apply --from <dir>when the
destination's parent cannot be created, including the tree branch's per-entry
writes and backups (#994).