Questlock v0.13.0

Local workflows should explain why they stopped, know which worker they reached, and catch an answer that is longer than requested.

What changed

• Budget stops now carry safe structured evidence: the limit, count or estimate, count basis, and whether generation was called. MCP exposes budget_exceeded directly. Conservative estimates stay conservative; there is no hidden cap increase or retry.
• questlock_status reports the connected worker's runtime, declared request root, profile, supported schemas/checks and local/paid policy without reading task files or contacting a broker or model. Doctor distinguishes a custom profile from an uninitialized managed profile.
• Word, character and paragraph checks support inclusive bounds and observed counts. Checks without a complete valid answer are not_evaluated; incomplete, over-budget and nondurable results do not get ordinary quality scores.
• writing init|plan|record|show builds a local-only prose workflow from an approved reference plus current evidence. Private config pins the reference hash. Changed facts produce a different snapshot; an unchanged approved task can still reuse its valid result.
• A bounded historical index records reviewed answer metadata and source versions. It verifies new persisted request provenance before indexing, preserves existing entries, and does not extend inference TTL or claim human acceptance.

Validation

Full Go race tests, vet, 96 Python checks, workflow lint, source/history and secret checks passed before publication. Independent release review covered the concrete changes. Platform CI exercises the packaged CLI, both plugins, native Desktop bundle contents, cold retention and reuse, the source container, and a disposable real Langfuse instance.

Isolated native Codex and Claude Code checks discovered the packaged skill and MCP tools without model calls or active settings changes. Packaged writing plans retain the same identity for unchanged input and a different identity for changed evidence. Checked revision/reuse and cold-history recovery used zero model tokens. These are integration/fixture checks, not a claim of general writing quality or whole-chat credit savings. Paid fallback remained disabled and provider API spend was $0.

Questlock was used during development for 34 checked literal packaging/version edits, plus an exact repeat, with zero model calls. Investigation, implementation and final judgment remained with the development agents.

Upgrade and boundaries

Upgrade the worker runtime, then restart or reconnect its MCP process to expose questlock_status and the new checks. A working custom profile does not need to be replaced with managed defaults. Existing fingerprints, published releases and legacy inference reads remain supported. Historical jobs lacking the new request-provenance field can still be reused under their original inference rules; the writing index refuses to attest them.

The writing index is bounded to 128 metadata references and 128 KiB per reference ID. A full index stops new entries without evicting history. Human acceptance, later rewrites and effort remain explicit outcome feedback. A historical reference is not an expired cache hit; the inference reuse maximum remains 24 hours. Broker/provider readiness in worker status remains not_checked.

Writing walkthrough · Count checks and history · Discovery and setup · Budget diagnostics

Fixes #32, #33 and #34 through #35. Download the matching native archive, offline plugin ZIP, or macOS Desktop MCPB below and verify it against checksums.txt.

Questlock v0.12.0

Questlock 0.12.0 — keep the work, lighten the hot store

Experimental release. Older completed work can move to private local cold
storage and still be reused exactly. Pending claims, budgets, retry receipts and
sync history stay intact. Missing or corrupt archive bytes stop the read rather
than starting another model call.

• Choose a complete policy: 30, 90 or 180 days, recent versions, database/cold
  capacity and bounded batches. CPU and memory signals can scale or defer work.
• Preview a batch, apply it, verify history, restore payloads or explicitly
  compact SQLite. Both manual and managed brokers honor a persisted policy.
• Optional Langfuse storage spans report counts, timing and error categories
  without exporting task content, paths, credentials or generation usage.

Archival retains history. It does not delete old records, provide a separate
backup or impose a total filesystem quota. Metadata still needs space, and WAL,
first migration and compaction require additional disk headroom.
Setup and recovery.

Validation and limits · Release checks

Questlock v0.11.0

Questlock 0.11.0 — install it where you work

Experimental release. Codex/ChatGPT desktop and Claude Code now have thin
plugins around the same Go runtime. Add the Git marketplace and enable the
plugin, or use the platform ZIP for an offline runtime. The first checked task
and its exact repeat need no model or provider account.

• Native plugin managers own installation and enablement. The launcher does
  not rewrite client configuration or install hooks.
• Git startup fetches a pinned, checksum-verified runtime into a private cache;
  four platform ZIPs include both plugins and their native executables.
• questlock plugin try, doctor and stop use a separate shared plugin
  profile. Paid fallback remains disabled; stopping retains the data.
• Concurrent first starts publish a complete fixture atomically, preserving
  existing operator files. Existing guided setup and Desktop MCPB remain.

Isolated native Codex and Claude Code installs loaded the skill and connected
the MCP server. Direct packaged-launcher tests completed the checked revision
and exact reuse with zero generation. These checks do not claim a native chat
model turn or free hosted account usage. Official public-directory listing is
separate from the published Git marketplace and requires vendor eligibility
and review. Install · Evidence and limits.

Questlock 0.10.0 — plug it into your Go app

Experimental release, published and verified. Four native archives, two macOS
MCPB bundles, checksums and both Linux container architectures passed their gates.
Validation records the exact source and anonymous asset checks.
Immutable archives retain their source-era candidate wording. Existing broker
and inference APIs remain; execution wires the pieces together for Go apps.

• One process-local coordinator combines bounded admission, shared task/result
  reuse and optional content-free Langfuse export. Telemetry failure leaves the
  task result alone.
• Optional shared scope/UTC-day input, output and total token quotas reserve
  declared task caps before generation, settle known usage and retain unknown
  reservations. Existing per-call/day dollar controls stay separate.
• Strict task JSON loading and declarative examples cover local Ollama, pinned
  standalone SGLang, resource settings and explicit cost/token budgets.
• CPU-only runtimes use the same API. Device execution, GPU batching and KV
  storage stay with the inference engine. Router-backed serving still needs a
  qualified provider contract.

Live local validation now covers fresh generation and exact reuse through the
published Go example: 57 input + 20 output tokens once, then 0 new tokens and
no second generation. The quota settled correctly with paid fallback disabled
and $0 provider API spend. Pressure had returned to normal; admission defaults
were unchanged. Evidence and limits.

Read the Go guide, complete example
and measured overhead and reuse.
Resource admission is per process; quota accounting coordinates trusted workers
using one broker scope and matching policy. It does not meter outside calls or
turn the broker into a GPU scheduler. Provider-token cost and hardware cost are
separate; synthetic control-plane tests are not GPU or model-quality benchmarks.

Questlock 0.9.1 — local quality gates

Experimental release, published and verified. Four native archives, two macOS
MCPB bundles, checksums and both Linux container architectures passed their
release gates. Validation records the exact source, pipelines,
anonymous asset checks and guarded local upgrade. Immutable archives retain
their source-era candidate wording. Published v0.9.0 and older assets remain
unchanged.

Local team quality controls now include opt-in canonical framing, JSON output,
manual primitive types and semantic field roles, bounded nonstreaming reasoning,
explicit sampling and HTTP deadlines, and optional model release. Every original
semantic check and reviewer approval remains required. Nil options preserve
existing identities, wire bytes and defaults; paid fallback stays unavailable.

macOS admission now includes bounded reusable memory, retaining headroom and
pressure limits. Fresh roles can wait for admission before claiming; generation
is not retried. Command arrays, active client settings and the permanent brand
image remain unchanged.

The six original bounded task contracts completed with local masters and both
reviews against the same saved 65,800-token Codex baseline: zero hosted tokens
and $0 provider API spend in the selected local workflows. The final invocation
reused five completed cases and generated three resource roles. That case uses
extra operator-declared types/semantics. All deliberate failed policies remain
in the report; this is tuned workflow evidence, not model parity or general
coding quality. Separate frozen validation completed 1 of 2 cases; one risk reviewer
exhausted its output cap. Whole-build credit and invoice savings remain unknown.

The archives now include a completed single-task request, complete reasoning
admission settings, the exact matched suite and synthetic historical receipts.
team benchmark can inspect the saved comparison offline without models or
account access. New experiments keep private task data in ignored directories;
the publication guard also rejects forced/deleted private demo data in history.

Measurements and failures ·
Team guide · Offline comparison ·
Architecture · Editable draw.io

---

Questlock 0.9.0 — local roles, shared gates

Published experimental release.
Release and downloads ·
Verified publication

Four native archives, two macOS MCPB extensions and both Linux container
architectures are published and verified. Immutable archives retain their
source-era candidate wording; the web guides record publication. Older releases
and the permanent brand image stay unchanged.

A bounded local master and one to three independent local reviewers now share
claims and checked results through the existing broker. team init creates an
inspectable task; team run executes selected roles; team status reads without
generation. Installed Ollama and isolated native Codex OSS are supported. The
team schema permits no hosted role or paid fallback. Native compatibility is
currently Codex CLI 0.163.0 and its prereleases.

Workers can hand roles across HTTPS with separate scoped credentials and an
explicit private CA. Pending/uncertain claims never expire into automatic
regeneration. Rejected answers withhold final text. Exact checked repeats need
no model, native executable, or inference admission. These text roles cannot
edit files or replace an existing chat backend.

team quest exercises three identities and six cases using a real private
HTTPS broker and scripted providers, with zero model calls. team suite,
team prompt, team baseline-record and team benchmark make a matched
comparison explicit and keep failed/missing cases in the denominator.

The 99% target is not established. Six deliberate Codex subscription baseline
answers passed; the tested local model and memory headroom did not complete the
team suite. $0 paid API charges are not evidence of equivalent accepted work.
Measurements, failures and limits include the
65,800 hosted baseline tokens, local usage, pressure refusals and unknowns.
No active client config or permanent brand image changed.

Team guide · Six-case walkthrough ·
Architecture · Editable draw.io

---

Questlock 0.8.0 — local work, shared results, your policy

Published experimental release.
Release and downloads ·
Verified publication

Four native archives, two macOS MCPB extensions and both Linux container
architectures are published and verified against the tagged source. Older
releases stay unchanged. Immutable archives retain their source-era candidate
wording; these web guides record publication.

• SecondGate: a private durable edge outbox, pinned writes to one authoritative
  broker, and atomic paginated snapshot/change consumption. Offline intent is
  not acceptance. Exact uncertain retries retain their operation ID; conflicts
  require an explicit caller decision. A dedicated pinned route makes older
  brokers reject the operation before mutation.
• Local decisions: optional TypeSafe System One Adapter integration with an
  installed Ollama model. Go returns an advisory choice or abstains. This is a
  local alternative, not Jev weights; scores are uncalibrated and authorize no
  action. Larger local reasoning remains an explicit separate request.
• Adaptive admission: opt-in CPU/memory-aware queues across named local lanes,
  with protected broker headroom, hard bounds, immediate pressure pause and
  gradual recovery. One controller coordinates one process. Reservations are
  estimates; OS/container resource enforcement stays separate.
• Your policy: copyable local-only worker and personal-instruction examples.
  Provider, model, limits, checks and reuse remain explicit configuration. Personal
  maintainer preferences stay outside public repository instructions.

questlock edge quest and questlock resources quest exercise real temporary
HTTP/SQLite workflows without a model. Exi...

Questlock v0.10.0

Questlock 0.10.0 — plug it into your Go app

Experimental release, published and verified. Four native archives, two macOS
MCPB bundles, checksums and both Linux container architectures passed their gates.
Validation records the exact source and anonymous asset checks.
Immutable archives retain their source-era candidate wording. Existing broker
and inference APIs remain; execution wires the pieces together for Go apps.

• One process-local coordinator combines bounded admission, shared task/result
  reuse and optional content-free Langfuse export. Telemetry failure leaves the
  task result alone.
• Optional shared scope/UTC-day input, output and total token quotas reserve
  declared task caps before generation, settle known usage and retain unknown
  reservations. Existing per-call/day dollar controls stay separate.
• Strict task JSON loading and declarative examples cover local Ollama, pinned
  standalone SGLang, resource settings and explicit cost/token budgets.
• CPU-only runtimes use the same API. Device execution, GPU batching and KV
  storage stay with the inference engine. Router-backed serving still needs a
  qualified provider contract.

Live local validation now covers fresh generation and exact reuse through the
published Go example: 57 input + 20 output tokens once, then 0 new tokens and
no second generation. The quota settled correctly with paid fallback disabled
and $0 provider API spend. Pressure had returned to normal; admission defaults
were unchanged. Evidence and limits.

Read the Go guide, complete example
and measured overhead and reuse.
Resource admission is per process; quota accounting coordinates trusted workers
using one broker scope and matching policy. It does not meter outside calls or
turn the broker into a GPU scheduler. Provider-token cost and hardware cost are
separate; synthetic control-plane tests are not GPU or model-quality benchmarks.

Questlock v0.9.1 — checked local teams

Questlock 0.9.1 — local quality gates

Experimental release, published and verified. Four native archives, two macOS
MCPB bundles, checksums and both Linux container architectures passed their
release gates. Validation records the exact source, pipelines,
anonymous asset checks and guarded local upgrade. Immutable archives retain
their source-era candidate wording. Published v0.9.0 and older assets remain
unchanged.

Local team quality controls now include opt-in canonical framing, JSON output,
manual primitive types and semantic field roles, bounded nonstreaming reasoning,
explicit sampling and HTTP deadlines, and optional model release. Every original
semantic check and reviewer approval remains required. Nil options preserve
existing identities, wire bytes and defaults; paid fallback stays unavailable.

macOS admission now includes bounded reusable memory, retaining headroom and
pressure limits. Fresh roles can wait for admission before claiming; generation
is not retried. Command arrays, active client settings and the permanent brand
image remain unchanged.

The six original bounded task contracts completed with local masters and both
reviews against the same saved 65,800-token Codex baseline: zero hosted tokens
and $0 provider API spend in the selected local workflows. The final invocation
reused five completed cases and generated three resource roles. That case uses
extra operator-declared types/semantics. All deliberate failed policies remain
in the report; this is tuned workflow evidence, not model parity or general
coding quality. Separate frozen validation completed 1 of 2 cases; one risk reviewer
exhausted its output cap. Whole-build credit and invoice savings remain unknown.

The archives now include a completed single-task request, complete reasoning
admission settings, the exact matched suite and synthetic historical receipts.
team benchmark can inspect the saved comparison offline without models or
account access. New experiments keep private task data in ignored directories;
the publication guard also rejects forced/deleted private demo data in history.

Measurements and failures ·
Team guide · Offline comparison ·
Architecture · Editable draw.io

Container: ghcr.io/spfuzzylink/questlock:v0.9.1 (linux/amd64, linux/arm64).

Questlock v0.9.0

Questlock 0.9.0 — local roles, shared gates

Experimental release candidate. Publication is pending required CI gates.

A bounded local master and one to three independent local reviewers now share
claims and checked results through the existing broker. team init creates an
inspectable task; team run executes selected roles; team status reads without
generation. Installed Ollama and isolated native Codex OSS are supported. The
team schema permits no hosted role or paid fallback. Native compatibility is
currently Codex CLI 0.163.0 and its prereleases.

Workers can hand roles across HTTPS with separate scoped credentials and an
explicit private CA. Pending/uncertain claims never expire into automatic
regeneration. Rejected answers withhold final text. Exact checked repeats need
no model, native executable, or inference admission. These text roles cannot
edit files or replace an existing chat backend.

team quest exercises three identities and six cases using a real private
HTTPS broker and scripted providers, with zero model calls. team suite,
team prompt, team baseline-record and team benchmark make a matched
comparison explicit and keep failed/missing cases in the denominator.

The 99% target is not established. Six deliberate Codex subscription baseline
answers passed; the tested local model and memory headroom did not complete the
team suite. $0 paid API charges are not evidence of equivalent accepted work.
Measurements, failures and limits include the
65,800 hosted baseline tokens, local usage, pressure refusals and unknowns.
No active client config or permanent brand image changed.

Team guide · Six-case walkthrough ·
Architecture · Editable draw.io

---

Questlock 0.8.0 — local work, shared results, your policy

Published experimental release.
Release and downloads ·
Verified publication

Four native archives, two macOS MCPB extensions and both Linux container
architectures are published and verified against the tagged source. Older
releases stay unchanged. Immutable archives retain their source-era candidate
wording; these web guides record publication.

• SecondGate: a private durable edge outbox, pinned writes to one authoritative
  broker, and atomic paginated snapshot/change consumption. Offline intent is
  not acceptance. Exact uncertain retries retain their operation ID; conflicts
  require an explicit caller decision. A dedicated pinned route makes older
  brokers reject the operation before mutation.
• Local decisions: optional TypeSafe System One Adapter integration with an
  installed Ollama model. Go returns an advisory choice or abstains. This is a
  local alternative, not Jev weights; scores are uncalibrated and authorize no
  action. Larger local reasoning remains an explicit separate request.
• Adaptive admission: opt-in CPU/memory-aware queues across named local lanes,
  with protected broker headroom, hard bounds, immediate pressure pause and
  gradual recovery. One controller coordinates one process. Reservations are
  estimates; OS/container resource enforcement stays separate.
• Your policy: copyable local-only worker and personal-instruction examples.
  Provider, model, limits, checks and reuse remain explicit configuration. Personal
  maintainer preferences stay outside public repository instructions.

questlock edge quest and questlock resources quest exercise real temporary
HTTP/SQLite workflows without a model. Existing setup, manual publishing, worker,
revision, outcome and runtime commands remain supported; existing MCP single-active
admission is unchanged. Ordinary Go commands require no Python. The optional
System One bridge has its own pinned environment and setup inputs.

Local smoke tests retain a malformed four-choice abstention alongside passing
two-choice calls. CPU/memory pressure refusals are retained too. These establish
protocol and safety behavior, not quality parity, invoice savings, faster inference
or a whole-chat hosted-usage percentage. Native Claude Desktop connection and
real cross-host/TLS deployment qualification remain unverified.

One local control loop · Edge coordination ·
Local decisions · Resource policy ·
Personal policy examples

---

Questlock 0.7.0 — a first task before a first token

Published experimental release.
Release and downloads ·
Verified publication

The four native archives and two macOS MCPB extensions are published. Native
Claude Desktop import/connection remains pending; package and protocol checks
alone do not qualify it. Immutable archives retain their source-era candidate
wording; the web documents record publication without replacing old assets.

Run questlock setup, choose Codex, Claude Code or Claude Desktop, and finish a
checked task without a model call or provider account. Codex/Claude Code setup
validates complete isolated candidates, preserves unrelated settings and uses
an independent rollback watchdog before committing a live native read.

New commands: setup, doctor, try, local start|status|stop, and
worker local-serve. local-agent optionally runs the main native Codex CLI
loop against an installed local Ollama model in an isolated home. The managed broker provisions private credentials and
shares one local SQLite owner across clients. Paid escalation stays disabled.
Claude Desktop uses a platform-specific binary MCPB extension and its native
installer; native Desktop import/connection qualification remains pending.

The repository now uses one permanent versionless brand image and a shorter
README. Historical releases and artwork remain available. Existing manual
broker, model, revision, outcome and runtime commands remain supported.

Start here · What we measured, including failures ·
Architecture

---

Questlock 0.6.0 — checked edits without another generation

Published experimental release.
Release and downloads ·
Verified artifacts and installed-worker evidence

Published v0.5.0 and older releases/assets remain available unchanged. Four native
archives, both container architectures and the installed v0.6.0 worker are verified.
The immutable archives retain their source-era candidate wording; these web docs
record subsequent publication without replacing those assets.

New in 0.6.0

• A Go revision library and revise plan, run, and quest commands apply
  exact counted literal replacements to a SHA-256-pinned source. Required
  source/output checks bound what can be returned; no model is called.
• Immutable content-addressed results compose with the existing scoped broker
  and exact parent key/version/digest provenance. Failed source/count/checks and
  uncertain publication withhold text without a paid or local model fallback.
• questlock_revise shares the bounded MCP workspace, admission and cancellation
  boundary with questlock_delegate; neither gains a filesystem-edit tool.
• Opt-in revision accounting records known zero new usage/provider charge with
  deterministic_edit or exact-repeat reuse_zero. Human acceptance remains
  explicit. Revision outcome cohorts require v0.6.0+ readers/writers; keep them
  separate from older generation cohorts during upgrades.
• Three simple and three complex offline scenarios, complete JSON examples,
  editable architecture, and Codex/Claude skill guidance for preferring exact
  edits before new generation.

Read what, why and how, runnable examples,
and release evidence.
A formatting development run and exact repeat made zero new model calls; their
single observed latencies establish no benchmark or whole-session savings.
New facts, open-ended rewrites and reasoning changes still need reviewed tasks.

Go vet/race, packaging/public-source checks, all five offline quests, native
Darwin/Linux ARM64/AMD64 archive CI, source-container execution and disposable
Langfuse readback passed before publication. Anonymous registry verification
checked both image configs and all layers. Published binary/skills activation
used an independent rollback watchdog and a private complete known-good backup;
a fresh native MCP process discovered both tools and returned a checked exact
revision reuse with zero new generation. Installation changed no client settings.
The development workstation separately completed its guarded global local-Codex
instruction rollout; supported opt-in setup
is documented. This preference is not enabled by the library installer. Broader
local comparison evidence now records all six deterministic edits and repeats
passing with zero generation, alongside one passing local-model baseline and
five withheld failures. See the validation record for all measured usage, the
single passing timing pair and the comparison's limits.

---

Questlock 0.5.0 — explicit provider routes and runtime cache evidence

Published experimental release.
Release and downloads · Verified artifacts
Repeated tasks and repeated prefix computation are different costs. Version
0.5.0 adds explicit OpenRouter and SGLang contracts while keeping runt...

Questlock v0.8.0 — local work, shared results, your policy

Local work, shared results, your policy

• SecondGate: durable edge intent, pinned publication to one authoritative broker, and atomic snapshot/change pulls. Uncertain writes retain exact retry identity.
• Adaptive admission: opt-in bounded CPU/memory-aware queues, protected broker headroom, pressure pause and gradual recovery within one process.
• Local decisions: the official TypeSafe System One Adapter with installed Ollama weights returns an advisory choice or abstains. This is a local alternative, not Jev weights.
• Configuration over convention: copyable local-only worker policy and private personal-preference examples. No paid fallback is enabled by guided setup.

questlock setup, try, doctor and guarded recovery remain available. edge quest and resources quest add offline checks without a model account.

Four native archives, two macOS MCPB extensions and both native Linux container architectures were verified against the tagged source. All seven required CI gates passed, including disposable real Langfuse readback. Older releases are preserved.

Local testing retained a malformed-choice abstention and resource-pressure refusals. Local model calls have no provider token charge; the surrounding hosted chat still uses its normal allowance. Native Claude Desktop connection, real cross-host/TLS deployment and GPU performance remain unqualified.

Get started · Your policy · Design · Validation · Container

An independent companion for OpenRouter: OpenRouter routes the model; Questlock coordinates the work. Credit to OpenRouter, Ollama, TypeSafe, SGLang and Langfuse.

Questlock 0.7.0 — a first task before a first token

A first task before a first token.

Run questlock setup, choose Codex, Claude Code or Claude Desktop, and complete a checked first task without a model call or provider account. Setup for Codex/Claude Code validates a complete isolated candidate and uses an independent rollback watchdog. doctor reports what is qualified; local start|status|stop manages one private local broker.

local-agent optionally runs the native Codex CLI with an installed Ollama model in an isolated home. It does not reroute the hosted desktop chat or its allowance. One local tool exercise succeeded; a separate wrapper trial produced no final answer despite exit 0. Both are retained in the evidence.

Downloads: four native macOS/Linux ARM64/AMD64 archives and two native macOS MCPB extensions. All seven actual assets and both public container architectures were verified. Claude Desktop native import/connection remains unverified; bundle and standalone MCP checks are separate evidence.

Start here · Verified publication · Measured development report, including failures · Public container

Two exact local-result repeats avoided 261 local tokens; recorded development provider charges were $0. Whole-build hosted-token savings, human rework cost and the 99% hosted reduction target remain unmeasured. Historical tags/assets/artwork remain intact, and the canonical versionless brand is frozen.

Questlock 0.6.0 — checked revisions without new generation

Questlock 0.6.0 — checked edits without another generation

Published experimental release.
Release and downloads ·
Verified artifacts and installed-worker evidence

Published v0.5.0 and older releases/assets remain available unchanged. Four native
archives, both container architectures and the installed v0.6.0 worker are verified.
The immutable archives retain their source-era candidate wording; these web docs
record subsequent publication without replacing those assets.

New in 0.6.0

• A Go revision library and revise plan, run, and quest commands apply
  exact counted literal replacements to a SHA-256-pinned source. Required
  source/output checks bound what can be returned; no model is called.
• Immutable content-addressed results compose with the existing scoped broker
  and exact parent key/version/digest provenance. Failed source/count/checks and
  uncertain publication withhold text without a paid or local model fallback.
• questlock_revise shares the bounded MCP workspace, admission and cancellation
  boundary with questlock_delegate; neither gains a filesystem-edit tool.
• Opt-in revision accounting records known zero new usage/provider charge with
  deterministic_edit or exact-repeat reuse_zero. Human acceptance remains
  explicit. Revision outcome cohorts require v0.6.0+ readers/writers; keep them
  separate from older generation cohorts during upgrades.
• Three simple and three complex offline scenarios, complete JSON examples,
  editable architecture, and Codex/Claude skill guidance for preferring exact
  edits before new generation.

Read what, why and how, runnable examples,
and release evidence.
A formatting development run and exact repeat made zero new model calls; their
single observed latencies establish no benchmark or whole-session savings.
New facts, open-ended rewrites and reasoning changes still need reviewed tasks.

Go vet/race, packaging/public-source checks, all five offline quests, native
Darwin/Linux ARM64/AMD64 archive CI, source-container execution and disposable
Langfuse readback passed before publication. Anonymous registry verification
checked both image configs and all layers. Published binary/skills activation
used an independent rollback watchdog and a private complete known-good backup;
a fresh native MCP process discovered both tools and returned a checked exact
revision reuse with zero new generation. Installation changed no client settings.
The development workstation separately completed its guarded global local-Codex
instruction rollout; supported opt-in setup
is documented. This preference is not enabled by the library installer. Broader
local comparison evidence now records all six deterministic edits and repeats
passing with zero generation, alongside one passing local-model baseline and
five withheld failures. See the validation record for all measured usage, the
single passing timing pair and the comparison's limits.

Questlock 0.5.0 — runtime cache evidence

Repeated answers and repeated prefix computation waste different resources. Questlock 0.5.0 adds explicit provider/runtime contracts around shared task coordination.

• A dedicated OpenRouter companion adapter pins one model/provider, disables requested fallback, reserves an operator-declared input bound and validates native usage.
• A native SGLang v0.5.20 adapter supports standalone dense/HiCache generation with exact tokenized input IDs and runtime identity checks before reuse.
• Runtime KV evidence stays separate from API billing cache buckets. Missing counters remain unknown; exact answer reuse reports zero new generation.
• Six offline HTTP/SQLite runtime scenarios, reproducible GPU qualification instructions, editable draw.io architecture and matching branding are packaged with the release.

Validation passed on native macOS/Linux ARM64/AMD64, the source container, and disposable real Langfuse metadata readback. All four published archives and both GHCR architectures were verified, including every image layer. Existing release tags and assets are preserved.

HiSparse and all direct prefill/decode instances are inspection only pending a validated ingress. Paid OpenRouter billing and real GPU acceleration are not live-qualified. Fixtures and CPU simulation instructions make no GPU performance claim. Local delegation does not change the surrounding hosted chat's allowance; hardware and operations still cost resources.

Runtime guide · GPU setup · Tagged CI · Container CI