Design Documents
This page is automatically generated from the design_docs/implemented directory. Each design document describes a feature, bug fix, or architectural decision that has been implemented.
v1.1.0
- M-DX-PACKAGE-DOGFOODING: DX Issues Found Building AILANG Packages in AILANG
- Sprint Plan — M-TAKE-FLATMAP-PEAK-MEMORY
- M-TAKE-FLATMAP-PEAK-MEMORY:
take(n, flatMap(f, xs))Cannot Bound Peak Memory — Expose the Fused Primitive That Already Exists, Fix Its Latent Type Bug, and Teach the Trap
v1.0.0
- Sprint Plan — M-CHECK-STRICT-FALLBACKS (Static "Ok contains default-valued literal" detection)
- M-CHECK-STRICT-FALLBACKS — Static detection of "Ok contains default-valued literal" anti-pattern
- Sprint Plan: Cost-Per-Verified-Success KPI — M4a (Cohort Freeze Mechanism)
- Sprint Plan: Cost-Per-Verified-Success KPI (M1–M3)
- Cost-Per-Verified-Success KPI
- Sprint Plan: M-EFFECT-ROW-VAR-UNIFICATION
- M-EFFECT-ROW-VAR-UNIFICATION validation
- M-EFFECT-ROW-VAR-UNIFICATION — Discharge effect-row variables in the effect-checking pass
- Sprint Plan — M-MCP-EXACT-TOOL-SURFACE-LANE-B
- M-MCP Exact Tool Surface Lane B: Embeddable Session-Owned Protocol Surface
- Sprint Plan: M-NIGHTLY-FLAKE-GUARD
- M-NIGHTLY-FLAKE-GUARD: Variance Guard for the Nightly Eval Regression Detector
- Sprint Plan: M-PLANNER-CODEX-LANE
- M-PLANNER-CODEX-LANE: Route the Mission Planner Through the Hardened codex Spawn Recipe
- Sprint Plan: M-V1-MEMORY-FOOTPRINT
- M-V1-MEMORY-FOOTPRINT: Memory efficiency audit and fixes for v1.0.0
- Sprint Plan: M-VM-STACK-LIMIT-PARITY
- M-VM-STACK-LIMIT-PARITY implementation verification
- M-VM-STACK-LIMIT-PARITY — the bytecode VM's 1000-frame cap diverges from the evaluator's 10,000 recursion limit: evaluator-legal deep recursion silently truncates service handlers under
--bytecode
v0.54.0
- Sprint Plan: M-SERVEAPI-RESPONSE-HEADERS
- serve-api Response Headers: Security Headers on
--static+ Hyphenated Route Headers
v0.53.0
- Sprint Plan: M-IFC-AUTHORITY-SCOPING
- M-IFC-AUTHORITY-SCOPING — Narrowing IFC Authority Primitives (Scoped Declassify + Call-Site Positive-Label Enforcement)
- Sprint Plan: Inline-test expected values
- Inline-Test Expected Values: lists and ADTs evaluate;
checkrejects whattestcannot run - Sprint Plan: M-RUN-POLICY-RESULT-UNFORGEABLE
- M-RUN-POLICY-RESULT-UNFORGEABLE: the confined program cannot forge the supervisor's
policy-result:line (#1548) - Sprint Plan: M-STD-YAML-ENCODE
- M-STD-YAML-ENCODE —
std/yamlencode: theJsonADT → idiomatic block YAML - Sprint Plan: M-STD-YAML-MAPPING-KEY-ORDER
- M-STD-YAML-MAPPING-KEY-ORDER:
std/yamlpreserves YAML mapping key order - Sprint Plan: M-WITH-RECV-RESERVED-WORDS
- M-WITH-RECV-RESERVED-WORDS: Keep
with/recvReserved, Fix the Diagnostic Floor They Stand On
v0.52.6
- Sprint Plan: M-SPRINT-LOC-NULL-SENTINEL
- Sprint-state
estimated_locsentinel migration:0→null - Sprint Plan: M-TASK-INPUTS — Typed cloud task inputs
- M-TASK-INPUTS: Typed Task Inputs — Fetch a Named
{repo, ref, path}Into the Workspace Before the Agent Runs
v0.52.0
- M-BYTECODE-NESTED-PATTERN-LOWERING — pattern lowering is flat-only: nested patterns (
a :: b :: rest) become evaluator-only, and literal/nested sub-patterns match too permissively on the strict VM (silent wrong results) - M-CROSS-ARCH-FLOAT-DETERMINISM: std/math transcendental results (exp, log, sin, …) differ by 1 ulp between arm64 and x86_64 — all three backends delegate to host-Go
math, which is not bit-stable across GOARCH - M-CTOR-PATTERN-ALIAS-AND-SCOPE: constructor import aliases are never bound (patterns silently never match; expressions fail "undefined variable"), and a constructor pattern naming an unknown constructor compiles clean and never matches
- M-ELABORATOR-LEXICAL-SCOPE: The elaborator resolves identifiers with no lexical scope — imports, builtins, and ADT constructors pre-empt every local binder
- Sprint Plan: M-EVAL-TAIL-CALLS
- M-EVAL-TAIL-CALLS: tail-call elimination in the tree-walking evaluator
- M-FLOAT-ORD-ONE-SEMANTICS — one answer to ordered float comparisons with NaN, on every path
- M-FOLDL-CONS-COST-MODEL: the documented-safe list-building path —
foldlwith a consing step, and "prepend then reverse" — is O(n²); std/list's cost claims must say so, the language needs one linear stateful-accumulate primitive, and the VM runs the shape ~2.4× slower than the interpreter - M-FULLWIDTH-INT-LITERALS-AND-LOGICAL-SHIFT: Hex/binary/octal int literals above 2^63−1 do not parse (u64 hash constants unwritable), pattern literals silently misparse, and there is no logical right shift
- Sprint Plan: M-INT-DIV-ZERO-ERROR
- M-INT-DIV-ZERO-ERROR: integer division and modulo by zero raise RT001, not a Go panic
- M-INTERPRETER-TAIL-CALL-ELIMINATION: The Tree-Walking Interpreter Diverges From the VM on Tail Calls (RT_REC_003 at 10k Where the VM Runs in Constant Stack)
- M-ITERATIVE-LIST-REMAINING: complete M-ITERATIVE-LIST —
any/findIndex(O(n²) on the VM, RT_REC_003 on the interpreter) andfoldr+ themaximum/minimumfamily (O(n²) + frame caps) delegate to iterative builtins; the Go-codegen engine already runs them iteratively - M-JSON-NUMBER-ROUNDTRIP: std/json float numbers do not round-trip — −0.0 sign loss, ≥1e21 integer-digit expansion, ≥2^63 decode saturation, silent ±Inf on out-of-range literals
- M-NAMEDTEST-TEMPFILE-LEAK — Interrupted
ailang testleaves_namedtest_body_*.ailin the package; pkg quality/publish/git ship it - Sprint plan: M-NET-SCOPE-PUBLIC (#1522)
- M-NET-SCOPE-PUBLIC: redirect-hop containment +
Net[scope=public](#1522, S0) - M-OPENROUTER-IMAGE-OUTPUT: Route AI Image Generation Through OpenRouter (Chat-API Modalities + Per-Call Model Choice)
- M-PARSER-NULLARY-SINGLE-CTOR-CURSOR: Leading-Pipe Nullary Constructor Swallows the Next Declaration's First Token
- M-TEST-HARNESS-MODULE-SCOPED-ENVS: Module-scoped private names in the test harness
- M-VM-ADT-TAG-CHECK-LOWERING — mixed-match ADT tag checks lower as record field access (
_record_geton an ADT): strict VM crash on constructor sub-patterns (e.g.TText(t) :: r), broken Go from--emit-go-v2 - M-VM-IFCHAIN-TAG-GUARD-LOWERING — ADT sub-pattern conds crash the strict VM (
_record_geton an ADT) and guards evaluate before pattern bindings (unbound variable) — two if-chain lowering gaps outside the nested-pattern and var-pattern sibling scopes - M-VM-MATCH-LOWERING — one recursive match lowering for the bytecode VM (nested patterns, constructor sub-patterns, guards over bindings, catch-all arms)
- M-VM-VAR-PATTERN-DEFAULT-ARM — variable patterns in match arms are evaluator-only on the strict VM (unbound variable), and misplaced/guarded variable arms give silent wrong results on the VM and in generated Go
v0.51.0
- Sprint Plan: M-VM-PURE-BUILTIN-COVERAGE
- M-VM-PURE-BUILTIN-COVERAGE: pure builtins run on the strict VM, and the gap is measured
v0.48.0
- M-MOTOKO-DST-MIGRATION: Adopt Arni's phase-core/DST refactor and re-prove our fork's improvements
- Motoko fork disposition — iteration 1
v0.47.0
- Sprint plan: M-NUMERICS-VEC-ARRAY-INGEST
- M-NUMERICS-VEC-ARRAY-INGEST: Unboxed float arrays, bulk update, binary numeric ingest
v0.46.0
- Sprint Plan: M-EFFECT-CEILING-REACHABLE
- M-EFFECT-CEILING-REACHABLE: compute the package effect ceiling over what the package reaches
- Sprint plan: M-SERVEAPI-OPERATOR-SURFACE
- M-SERVEAPI-OPERATOR-SURFACE: what serve-api exposes is the operator's choice
- M-SMT-NULLARY-CALLEE — sprint plan (no design doc)
v0.44.0
- Sprint Plan: M-SERVEAPI-BIND-HOST-CORS
- M-SERVEAPI-BIND-HOST-CORS: serve-api binds loopback by default, CORS is opt-in and allowlistable
- Sprint Plan: M-SERVEAPI-WS-BRIDGE
- M-SERVEAPI-WS-BRIDGE: a serve-api WebSocket route with a per-frame AILANG verdict
- Sprint Plan: M-STD-AUDIO
- M-STD-AUDIO:
std/audio, PCM to WAV in pure AILANG plus a pure-Go Ogg Opus encoder
v0.43.2
- M-COMPILE-CACHE-DIRTY-BUILD-KEY — a rebuilt compiler must not be served its predecessor's verdicts
- M-FLOAT-EQ-ONE-SEMANTICS — one answer to
NaN == NaN, on every path
v0.42.0
- M-EQ-DERIVE-CONTAINERS — Sprint Plan (re-land)
- M-EQ-DERIVE-CONTAINERS — make
==work for records, Options and lists when the parts already have Eq
v0.41.0
- Sprint Plan: M-EXECUTOR-POLICY-HARDENING
- M-EXECUTOR-POLICY-HARDENING: Make AILANG's Execution Policy an Enforced Boundary
v0.40.1
- M-AI-DECIDE-SYSTEM-ONE — shadow lane-router report
- Sprint Plan: M-AI-DECIDE-SYSTEM-ONE (Phase 1)
- M-AI-DECIDE-SYSTEM-ONE: Typed Decisions as an AILANG Primitive (TypeSafe Jev via OpenRouter)
- M-PKG-BIN-ENTRYPOINTS:
[bin]— install a CLI from an AILANG package
v0.40.0
v0.39.3
- Sprint Plan: M-DANEEL-AILANG-EXECUTOR
- M-DANEEL-AILANG-EXECUTOR — Daneel gets its own executor on the
ailang_onlylane
v0.39.0
- Sprint Plan: M-AGENT-AILANG-ONLY-EXECUTION
- M-AGENT-AILANG-ONLY-EXECUTION — AILANG as the agent's only route to acting, fleet-wide
v0.38.7
v0.38.6
- Sprint Evaluation — m-debugcacheforms-flaky-on-macos-ci (round 1)
- Sprint Evaluation — m-debugcacheforms-flaky-on-macos-ci (round 2)
- Mutation-Audit Record — M-DEBUGCACHEFORMS-FLAKY-ON-MACOS-CI
- Sprint Plan: M-DEBUGCACHEFORMS-FLAKY-ON-MACOS-CI — drain the capture pipe before closing its read end
- M-DEBUGCACHEFORMS-FLAKY-ON-MACOS-CI — drain the capture pipe before closing its read end
- Sprint evaluation — v1_iter354_pin_age (m-pin-drift-blind-under-sha-pin)
- Sprint evaluation — ROUND 2 (delta review) — v1_iter354_pin_age
- Mutation audit — M-PIN-DRIFT-BLIND-UNDER-SHA-PIN (sprint
v1_iter354_pin_age) - Sprint Plan: M-PIN-DRIFT-BLIND-UNDER-SHA-PIN — PIN_AGE, a stale deployment pin reports its distance from origin/dev
- M-PIN-DRIFT-BLIND-UNDER-SHA-PIN: PIN_AGE — a stale deployment pin reports its distance from origin/dev
v0.38.0
- Sprint Plan: M-PROCESS-SUBCMD — Subcommand Allowlists for the Process Effect
- M-PROCESS-SUBCMD: Subcommand Allowlists for Process Effect
- M-RIG-LOCK-YIELD: Cooperative Yield Requests for
rig-lock.sh
v0.37.3
- Sprint Plan: M-DEBUG-SINK-STRUCTURED-LINES
- M-DEBUG-SINK-STRUCTURED-LINES: Structured Debug.log lines survive every host sink
v0.36.0
- Sprint Plan — M-SMT-INTERP-SHOW
- M-SMT-INTERP-SHOW: Type-directed
shownormalization — unblock Z3 verification of string-building functions
v0.35.4
v0.35.3
- Sprint Plan — M-COORDINATOR-TEST-PARALLELISM (executor-ready)
- M-COORDINATOR-TEST-PARALLELISM: Attack the Four Timer-Bound Coordinator Tests by Injecting Their Timers
- Evaluation — M-DAEMON-TASK-EXEC-RUN-UNTESTED (v1 mission iteration 352, round 1)
- Mutation audit — M-DAEMON-TASK-EXEC-RUN-UNTESTED (V1 mission iteration 352)
- Sprint plan — M-DAEMON-TASK-EXEC-RUN-UNTESTED (V1 mission iteration 352)
- M-DAEMON-TASK-EXEC-RUN-UNTESTED — make
executeTaskexecutable under test
v0.35.2
- Sprint Plan: M-COMPILE-CACHE-UNVERIFIED-ARTIFACTS
- M-COMPILE-CACHE-UNVERIFIED-ARTIFACTS — bind executable cache artifacts to their authorizing key
- Sprint Plan: M-PI-EVALUATOR-SESSION-HANDSHAKE
- M-PI-EVALUATOR-SESSION-HANDSHAKE — satisfy the judge's local protocol before review
v0.35.1
- Sprint plan: M-MOTOKO-SUITE-BOUND-DERIVATION
- M-MOTOKO-SUITE-BOUND-DERIVATION: derive the suite's wall-clock and node-ceiling bounds from a stimulus measured in-test
v0.35.0
- Sprint Plan: M-DX-PI-HARNESS — pi dev-harness extensions (Streams A+B)
- Sprint Plan: M-DX-PI-HARNESS-P2 — Distribution v2: ailang pi install
- M-DX-PI-HARNESS — pi as a first-class development harness for AILANG (two streams)
- M-EVAL-W8: harness errors must not be scored as capability failures
- Sprint Plan — M-FMT-PRINTER-LINE-WIDTH-LIMIT
- M-FMT-PRINTER-LINE-WIDTH-LIMIT: A Width Predicate for the Printer's Existing Multi-Line Layouts, Then the Second Corpus Reformat
- Sprint Plan: M-MESSAGE-PLANE-FAIL-LOUD
- M-MESSAGE-PLANE-FAIL-LOUD: the remaining silent seams
- Sprint plan — M-MOTOKO-CONNECTION-PROBE-RUN-LANE-HARNESS
- M-MOTOKO-CONNECTION-PROBE-RUN-LANE-HARNESS: behavioral pin for production lane process-group kill
- Sprint Plan: M-MOTOKO-DISCOVERY-ARM-DISCRIMINATING-REFUSAL
- M-MOTOKO-DISCOVERY-ARM-DISCRIMINATING-REFUSAL: make the real wall-clock discovery refusal observable and reachable
- Sprint Plan: M-PI-HARNESS-UPGRADE
- M-PI-HARNESS-UPGRADE — move the fleet off abandoned pi 0.73.1, and make the boundary visible in the data
- Sprint Plan: M-PIPELINE-RECONCILIATION
- M-PIPELINE-RECONCILIATION: one pipeline definition, two execution lanes
- Sprint Plan — M-SPAWN-PIN-ENFORCEMENT
- M-SPAWN-PIN-ENFORCEMENT — Deterministic role-spawn path enforcement for mission-control
v0.34.0
- LC-1
m-list-repr-spike— M6 report: full matrix, kill-criterion arithmetic, and the programme verdict - M-OLLAMA-CLOUD-PROVIDER: Ollama Cloud as a Second Route for Open-Weight Models
v0.33.4
- Sprint Plan: Persistent Browser Auth Profiles
- M-BROWSER-AUTH-PROFILES — persistent authenticated identity without password exposure
- Sprint Plan: M-STDLIB-FREEZE-GATE — Retire the Rotted Freeze Duplicate, Alias the Accepted Name to the Live Gate
- M-STDLIB-FREEZE-GATE-PATH-MISMATCH: Retire the Rotted Freeze Duplicate, Alias the Accepted Name to the Live Gate
v0.33.3
- Sprint Plan: Remote Browser Session Providers
- M-REMOTE-BROWSER-SESSION-PROVIDERS — provider-neutral browser sessions for local and cloud agents
v0.33.2
- HANDOVER — M-MISSION-LOOP-UNIFIED-TELEMETRY
- Sprint Plan: M-MISSION-LOOP-UNIFIED-TELEMETRY
- M-MISSION-LOOP-UNIFIED-TELEMETRY
v0.33.1
- Sprint Plan: M-CI-FLAKE-SYSTEMIC-FIX
- M-CI-FLAKE-SYSTEMIC-FIX: One Gating Convention, Bounded Waits, a Default-Deny Egress Boundary, and a Known-Offender Lint for the Go Test Suite
- Sprint Plan — M-OLLAMA-V1-STREAMING-IDLE-TIMEOUT
- M-OLLAMA-V1-STREAMING-IDLE-TIMEOUT: Stream the ollama /v1 tool-calling path with a read-level idle timeout
- Sprint Plan: M-OPENROUTER-BROADCAST-INGEST
- M-OPENROUTER-BROADCAST-INGEST
- M-PARSER-RESERVED-KEYWORD-DIAGNOSTICS: catch
as-as-identifier before it cascades
v0.32.0
- Sprint Plan: M-EVAL-STANDARD-CONFIDENCE-GATING
- M-EVAL-STANDARD-CONFIDENCE-GATING: ELO Confidence Gating for the Cloud Standard Release Baseline
- M-FMT-DIALECT-ALIGNMENT
- PRE-READ: m-recorded-stream-api (ailang#546) — attended evaluation of the offered implementation
- Sprint Plan — M-RECORDED-STREAM-API (sprint 1 of 2)
- M-RECORDED-STREAM-API — Productionize the offered recorded
std/aistream
v0.31.0
- Sprint Plan — M-AI-REASONING-EFFORT
- M-AI-REASONING-EFFORT — Cross-Provider Request-Side Reasoning Control
- Sprint Plan: M-ANTHROPIC-CACHE-HIT-RATE
- M-ANTHROPIC-CACHE-HIT-RATE: Close the prompt-cache gap on the
Generatepath - Sprint Plan: M-EFFECT-ROW-SHOW-INTERP (#386)
- M-EFFECT-ROW-SHOW-INTERP: Preserve Effect Rows Across Pure Nested Calls
- M-EVAL-FMT-WEAKMODEL-AB — M3 Analysis Results
- M-EVAL-FMT-WEAKMODEL-AB — M4 Verdict (final)
- M-EVAL-FMT-WEAKMODEL-AB — M5 (hard-set re-run) preregistration
- M-EVAL-FMT-WEAKMODEL-AB — M5 (hard-set re-run) results
- PREREGISTRATION — M-EVAL-FMT-WEAKMODEL-AB
- Sprint Plan: M-EVAL-FMT-WEAKMODEL-AB
- M-EVAL-FMT-WEAKMODEL-AB: measure whether the fmt hook helps a weak model author correct AILANG
- Sprint Plan: M-EVAL-MEASUREMENT-CONTRACT
- M-EVAL-MEASUREMENT-CONTRACT: Distinguish "Measured Badly" from "Failed to Measure"
- M-EVAL-TOKEN-HEADROOM: Equal Headroom, Visible Truncation, Labelled Thinking State
- Sprint Plan: M-MCP-EXACT-TOOL-SURFACE — Lane A
- Sprint Plan — M-PROPERTY-GENERATOR-COVERAGE Lane A only
- Sprint Plan — M-PROPERTY-GENERATOR-COVERAGE Lane B1 only
- M-PROPERTY-GENERATOR-COVERAGE: Vacuous-Pass Honesty + Structural Generator Derivation for Contract Property Tests
- Sprint Plan: M-PROPERTY-TEST-TRUST (#547 + #535)
- M-PROPERTY-TEST-TRUST: Deterministic and Sound Contract Verdicts
- Sprint Plan: M-Z3-ADT-RECORD-SORT — Declaration Closure Through Records + Honest
ai-checkExit - M-Z3-ADT-RECORD-SORT: Declare Reachable ADTs Through Records and Make
ai-checkHonest - Sprint Plan: M-Z3-HARD-TIMEOUT
v0.30.0
- Sprint Plan: M-AILANG-FMT-ADOPTION
- M-AILANG-FMT-ADOPTION:
ailang fmtAdoption — Discoverability + Opt-in Harness Hooks - Sprint Plan — M-AILANG-FMT-INLINE-INTERIOR
- M-AILANG-FMT-INLINE-INTERIOR: Stable Multi-Line Let-Chain Comment Boundaries
- Sprint Plan: M-AILANG-FMT-PHASE2 — Lossless Comment Preservation for
ailang fmt - M-AILANG-FMT-PHASE2: Phase 2 — Lossless Comment Preservation for
ailang fmt - M-AILANG-FMT:
ailang fmt— Canonical AILANG Source Formatter - Sprint Plan: M-ARCH-BOUNDARIES (Phases 1–3)
- M-ARCH-BOUNDARIES: Formalize Dashboard/Core Separation
- Arity-Style Diagnostic (m-arity-style-diagnostic)
- Sprint Plan: M-DOCS-GATE-NOT-REQUIRED
- M-DX-AI-DISCOVERY — Sprint Plan (Executor Handoff)
- M-DX-AI-DISCOVERY: One-Shot Stdlib Discovery for AI Agents (RE-SCOPED)
- M-DX-EXAMPLES-COVERAGE: Close the Examples Gap — Coverage, a Real CI Gate, and a Working
--examplesFlag - M-DX-EXPECTED-FAIL-FIXES: Fix Remaining Expected-Fail Examples
- M-DX-JSON-BOOL: JSON Boolean Coercion and Firestore Encoding Consistency
- M-DX-RECORD-CONS: Record Literal + :: Cons Pattern Bug
- Sprint Plan: M-DX-SPLIT-ARG — Compile-Time Warning for Reversed
splitArguments - M-DX-SPLIT-ARG: Compile-Time Warning for Reversed
splitArguments - M-DX-TAPP-TRECORD: Type Inference Bug with Nested [[RecordType]]
- M-EFFECT-MODE-VALIDATION: Enforce the Closed Mode Set for Parameterised Effects
- Sprint Plan: M-EFFECT-REPLAY-CONTRACTS (Rand pilot)
- M-EFFECT-REPLAY-CONTRACTS: Replay Contract Registry + Mode-Aware Runtime Dispatch (Rand pilot)
- M-EFFECT-REPLAY-SUBSUMPTION: Validate-Path Effect-Mode Subsumption
- M-EVAL-ELO-PRIORITY-ROTATION: ELO-Driven Benchmark Prioritization for the OS Rotation Filler
- M-EVAL-LOCAL-CLOUD-UNIFY: local models in the same tables + rolling trends as cloud
- Sprint Plan — M-FMT-PROPERTIES-PRINTER-ROUNDTRIP
- m-fmt-properties-printer-roundtrip — Contract-Clause Printer Round-Trip Fix
- M-GEMINI-EVALUATOR-DIFF-BRIDGE: inject a sprint diff bundle into a sandboxed gemini evaluator
- M-GEMINI-EXEC-PROJECT-PLUMBING: Thread GCP Project/Location Env into
ailang execTask Construction - M-GEMINI-REPO-MOUNT Phase 2 — Sprint Plan (Clone-over-egress capability)
- M-GEMINI-REPO-MOUNT — Managed Agents repository and inline-source mounts
- Sprint Plan: M-LAMBDA-OPEN-RECORD-PATTERN
- M-LAMBDA-OPEN-RECORD-PATTERN — Lambda +
{field, ...}pattern doesn't propagate row polymorphism - Sprint Plan — M-MATCH-XCHECK-ERROR-QUALITY
- M-MATCH-XCHECK-ERROR-QUALITY —
MatchForeignConstructorErrorshows empty constructor list for non-imported ADTs - M-MISSION-ADAPTIVE-MULTIPROVIDER-ROUTING: the heterogeneous model fleet — quota-aware selection, design quorums, cross-provider executors, local-GPU lane
- Sprint Plan — M-MISSION-AGENTIC-ROUTING (remaining slice: M1b · M2 · M3)
- M-MISSION-AGENTIC-PROVIDER-ROUTING: enforce per-role AGENTIC cross-provider routing + right-size the roles
- Sprint Plan — M-MISSION-COST-CHAINS
- M-MISSION-COST-CHAINS: make
ailang chainsthe mission's cost tracker — fix $0 attribution + ingest loop activity - Sprint Plan — M-MISSION-FLEET-AB (Phases A+B of m-mission-adaptive-multiprovider-routing)
- M-MISSION-PORTABILITY: extract the mission loop into a portable template — bootstrap kit for the Ailang World mission
- M-MISSION-QUORUM-AGENTIC-VERIFY: quorum reviewers that VERIFY and HONE (agentic, repo-armed), not just reason
- Sprint Plan — M-MODLESS-FAIL-LOUD (MOD014)
- M-MODULE-LESS-RUN-FAIL-LOUD:
ailang run/checkmust not silently succeed on a module-less file - Sprint Plan: M-MODULE-LET-FUNC-RESOLUTION
- M-MODULE-LET-FUNC-RESOLUTION: Module-level lets cannot reference module funcs (false "undefined variable"; hint cites closed #327)
- Sprint Plan — M-PRELUDE-OPTION-RESULT (mission iter 27)
- M-PRELUDE-OPTION-RESULT: Add Option/Result to the prelude (structural AI-DX fix)
- Sprint Plan: M-PROMPT-FOOTGUNS — Prompt-Teaching Footguns → Compiler Diagnostics
- M-PROMPT-FOOTGUNS-TO-DIAGNOSTICS: Route Prompt-Teaching Footguns to Compiler Diagnostics
- Sprint Plan: M-PUBLIC-FEEDBACK-DELIVERY-AUDIT
- M-PUBLIC-FEEDBACK-DELIVERY-AUDIT: external feedback isn't reaching Discord (or the inbox)
- M-SERVEAPI-RAW-HANDLER-MCP:
@nomcpexclusion + opt-in@rawhandlers over MCP - Sprint Plan: M-SMT-CALLEE-SORT-GATE
- M-SMT-CALLEE-SORT-GATE: Reject Unencodable Callee Signature Sorts (No Z3 Crash)
- M-STD-AI-VISION-INPUT — Sprint Plan
- M-STD-AI-VISION-INPUT — Image (vision) input on std/ai Message
- Sprint Plan: M-STD-YAML —
std/yamlYAML→JSON Bridge - M-STD-YAML:
std/yaml— YAML Ingestion via JSON Bridge - Sprint Plan — m-stdlib-regex (Linear-Time RE2 Regex Builtin)
- m-stdlib-regex — Linear-Time (RE2) Regex Builtin
- m-stdlib-url-parse — RFC-3986 URL Parsing via Go net/url
- Sprint Plan: M-SYNTAX-AI-FORGIVING — Forgiving statement-separator syntax
- M-SYNTAX-AI-FORGIVING: Forgiving statement syntax — accept the AI's newline-separator prior
- Sprint Plan: M-XMOD-ALIAS-POLY — Parameterized Type-Alias Substitution
- m xmod alias poly
- Sprint Plan: M-EFFECT-MODE-VALIDATION
v0.29.0
- Ghost reconciliation — 2026-07-12 (v1.0 backlog triage)
- m budget scoping bug
- M-CLOUD-OBSERVATORY: Observatory Firestore Backend for Cloud Run
- M-COORDINATOR-INBOX-WILDCARDS: glob-based inbox routing in the agent registry
- M-DASHBOARD-PUBSUB-EVENTS: Dashboard Pub/Sub Event Subscriber
- M-DIAGNOSTIC-COVERAGE: The footgun coverage table — error-time teaching with CI-enforced fix-carrying diagnostics (R1.1)
- M-EFFECT-ROW-POLY-PARAMS — Effect-row polymorphism on higher-order arguments
- Sprint Plan: M-EVAL-BOUNDED-PIPELINE
- M-EVAL-BOUNDED-PIPELINE: Fused Bounded Combinators + Memory Ceiling
- M-EVAL-FRONTIER-TIER — Saturation Demotion Audit
- M-EVAL-FRONTIER-TIER: A harder benchmark tier + curation to de-saturate the suite
- M-EVAL-LOCAL-OLLAMA — Optimize eval-suite for local Ollama models
- M-EVAL-REGRESSION-DETECTOR-CONTRACT: Specify trial grouping, flaky-vs-persistent classification, and infra-noise policy
- M-EVAL-RIG-RELIABILITY — post-mortem + remediation plan
- M-FEEDBACK-GATE-CLOUD-ADAPTER: Firestore stores + classifier wiring for the feedback gate
- M-FEEDBACK-TRIAGE-GATE: Cost & abuse gate for the public feedback → agent pipeline
- Sprint Plan: M-FILE-HANDLING
- M-FILE-HANDLING: fileData/fileUri Support & serve-api POST Param Fix
- Improve Module Loading Error Messages for Standalone/Scratch Files
- M-NAMED-TEST-BLOCKS: Execute
test "name" { ... }blocks — and stop reporting skipped suites as passing - Sprint Plan: M-PKG-FEEDBACK-LOOP (Test + activate per-package feedback + macOS notifier)
- M-PKG-FEEDBACK-LOOP: Validate the per-package feedback loop end-to-end
- M-RECORD-UPDATE-LOCAL-RESOLUTION: Local functions in record-update fields resolve as "undefined variable" (#327)
- Sprint Plan: M-RIG-RELIABILITY — make the rig catch breaks + measure docx
- M-TYPEENV-SUB: TypeEnv Substitution Gap — ADT Return Types Lost in Cross-Module Exports
- Sprint Plan: M-TYPEENV-SUB — TypeEnv Substitution Gap (ADT return types lost in exports)
- Sprint Plan: M-V1-STABILITY-PROMISE — The 1.x Stable-Surface Promise
- M-V1-STABILITY-PROMISE: The 1.x Stable-Surface Promise
- m verify stdlib stale path
- M-WASM-TYPECHECK-LIMITS — depth-budget guard + clear error for WASM type-checker overflow
- Sprint Plan: M-FEEDBACK-GATE-CLOUD-ADAPTER
- Sprint Plan: M-FEEDBACK-TRIAGE-GATE
v0.28.1
v0.28.0
v0.27.0
- AILANG v0.26.1 — Programmatic Gap Analysis
- m onboarding ergonomics
- Sprint Plan: M-SNAKE-FEEDBACK — Terminal I/O + Onboarding fixes
- m terminal io
v0.26.1
- M-AGENT-ERGONOMICS: One-Cycle Self-Correction for AILANG Agents
- M-AGENT-STUCK-FIXES: deterministic fixes for the errors that loop the agent
- M-EVAL-REIMPLEMENT-BENCH: Multi-file Reimplement Tasks as Tracked Benchmarks
- M-EVAL-RELIABLE-GRADING: Grade the Model's Actual Workspace, Not an Isolated File
- M-EVAL-VERSION-BANKING: per-release eval banking so the loop can see its own progress
v0.26.0
- Sprint Plan: M-EMBED-TASK-PREFIX
- M-EMBED-TASK-PREFIX: Model- and Role-Aware Embedding Task Prefixes
- Sprint Plan — M-EVAL-BENCHMARK-UI-CONSOLIDATION
- M-EVAL-BENCHMARK-UI-CONSOLIDATION: A coherent benchmark/dashboard information architecture
- M-EVAL-DASHBOARD-REDESIGN: Surface ELO ratings, difficulty bands, and regraded scores on the website
- Sprint Plan — M-EVAL-ELO-PERSIST (M-EVAL-RATING-EFFICIENCY part 2)
- M-EVAL-OS-CONTINUOUS-ROTATION: Background, incremental OS cross-language/harness evals between releases
- M-EVAL-OS-LONGITUDINAL Phases 2–5 — Sprint Plan
- M-EVAL-OS-LONGITUDINAL: Per-Release Local-Model Eval Trend
- M-EVAL-OUTPUT-NORMALIZE: Boolean-case + numeric-format grading parity (sibling of M-EVAL-JSON-COMPARE)
- M-EVAL-RATING-EFFICIENCY: ELO-Style Benchmark Difficulty + Targeted Reruns + Tier Saturation
- M-MOTOKO-AGENT-SYSTEM-PROMPT — give motoko a lean agentic system prompt
- M-OLLAMA-PER-MODEL-MAX-TOKENS — flow the registry's declared max_output_tokens to motoko
- M-OLLAMA-TEMPERATURE-KNOB: Configurable Temperature on the Agentic Ollama Path
- M-SECRET-EFFECT: Gated
SecretEffect with 1Password-Backed Remote Human Approval - M-SECRET-EFFECT — Sprint Plan (Phase 1 + 2)
- M-SECRET-REMOTE-APPROVAL-WIRING — Sprint Plan
- M-SECRET-REMOTE-APPROVAL-WIRING: Make Push-to-Phone Secret Approval Operational
- M-THREE-CAMPS Re-run — Fair Cross-Language Benchmark on the Local Rig
- M-WEBSITE-ANTIGRAVITY-DOCS-SYNC: Retire Gemini CLI from the Website, Document Managed Agents (Antigravity)
v0.25.1
- Sprint Plan: M-IFACE-COMPACT-ADT-FIELDS
- M-IFACE-COMPACT-ADT-FIELDS: Compact Interface Carries ADT Constructor Fields
v0.25.0
- M-AILANG-ERROR-QUALITY: Error Messages as the Lever for Small-Model Success
- M-BUG-LETREC-SINGLE-CALL: Letrec with Single Recursive Call Fails
- M-MOTOKO-FAILURE-MODE-OBSERVABILITY — segment agent failures by mode (disengage vs grind-wrong)
- M-MOTOKO-OBS-TOOLCALLS: Surface
agent_tool_callsin Eval Result JSON - M-MOTOKO-OBS-TRANSCRIPT: Retain Motoko Tool-Call Transcript
- M-MOTOKO-OLLAMA-LOOP-CONVERGENCE: Make the Motoko Agent Loop Terminate Against Local Ollama
- M-OLLAMA-V1-TOOLCALLING: Route Ollama Tool-Calling via OpenAI-Compat /v1
- M-RIG-LOCK-ENFORCE: Enforce the Rig Lock in
ailang eval-suite - Sprint Plan: M-TYPE-LIST-ELEMENT-SOUNDNESS
- M-TYPE-LIST-ELEMENT-SOUNDNESS: List-Literal Element Types Escape the Type Checker
v0.24.1
v0.24.0
- M-EVAL-NETWORK-MOCK-FIXTURE: Deterministic local HTTP mock for network benchmarks
- M-EVAL-PROMPT-DELIVERY: How to Deliver the Teaching Prompt to Local-Model Agents
- M-LETREC-BLOCK-SCOPE — Block-form
letrecdropped its binding - M-MOTOKO-LOCAL-OLLAMA: Motoko Agent Loop Against Local Ollama on the Eval Rig
- M-STDLIB-XML-LENIENT — Sprint Plan
- m stdlib xml lenient
v0.23.x
- Weekend Iteration Report — Iter 6 N=3 Variance Addendum (2026-05-24)
- Weekend Autonomous Iteration Report — 2026-05-23 / 24
v0.23.0
- M-AI-CALL-JSON-SIMPLE-RESULT: Result-returning variant of callJsonSimple
- Sprint Plan: M-COORD-TAG-ROUTING-LASTMILE
- M-COORD-TAG-ROUTING-LASTMILE: Close the last-mile gaps preventing tag-routed dispatch
- Sprint Plan: M-DOCPARSE-RESILIENCE-FIXES
- M-MCP-UNIT-PARAM-BINDING: Reject omitted MCP tool params instead of binding Unit
- M-MSG-AUTO-TRIAGE-PIPELINE: Autonomous Inbound Triage + Central Notification Bus
- Sprint Plan: M-MSG-TRIAGE-ROUTER (Build Order Step 1)
- M-NOTIFY-CHANNELS-FRAMEWORK: Outbound Notification Channels (Adapter Framework)
- Sprint Plan: M-NOTIFY-CHANNELS (Build Order Step 2)
v0.22.0
- M-MANAGED-AGENTS: Retire Gemini CLI, Adopt Vertex Managed Agents API
- Sprint Plan: M-COORD-CODEX (Codex Executor for Coordinator + Evals)
- M-COORD-CODEX: Codex Executor for Coordinator + Evals
- Sprint Plan: M-COORD-MULTI-HOST-WORKERS
- M-COORD-MULTI-HOST-WORKERS: Bare-metal Host Workers in the Coordinator System
- M-EVAL-LOCAL-OBSERVABILITY-FOLLOWUP — Sprint Plan
- M-EVAL-LOCAL-OBSERVABILITY-FOLLOWUP — Wire chain_id/stage_id into eval-suite OTEL resource attrs
- M-EVAL-LOCAL-OBSERVABILITY — Sprint Plan
- M-EVAL-LOCAL-OBSERVABILITY — Reliable local eval observability
- M-EVAL-METRICS-TAXONOMY: Beyond Pass/Fail — Metrics for Continuous LLM Eval
- Sprint Plan: M-MANAGED-AGENTS — Retire Gemini CLI, Adopt Vertex Managed Agents API
- M-MATCH-ADT-XCHECK-REGRESSION — Foreign-constructor patterns silently accepted again
- Sprint Plan: M-MOTOKO-AILANG-RECONCILE
- M-MOTOKO-AILANG-RECONCILE: Restore working motoko_agent build on current AILANG
- Sprint Plan: M-MOTOKO-V021-EFFECT-ROW-MIGRATION
- M-MOTOKO-V021-EFFECT-ROW-MIGRATION: Finish motoko_agent's AILANG v0.21+ migration
- Sprint Plan — M-PATTERN-AND-INVOCATION-REPAIR
- M-PROMPT-STDLIB-COVERAGE
- Sprint Plan — M-SCHEME-IMPORT-PRESERVE-ADT-HEAD
- M-SCHEME-IMPORT-PRESERVE-ADT-HEAD — Imported function schemes lose ADT head constructors
- M-THREE-CAMPS-LANGUAGE-SURVEY: Gap Analysis Across 16 AI-Designed Languages
- M-THREE-CAMPS Sprint Plan (Gap-Analysis-First)
- Sprint Plan: M-TRANSITIVE-ALIAS-ENV-IMPORT
- M-TRANSITIVE-ALIAS-ENV-IMPORT — propagate transitively-imported type aliases into the unifier's aliasEnv
- Sprint Plan: M-WASM-TYPECHECK-FLOAT-DIVERGENCE
- M-WASM-TYPECHECK-FLOAT-DIVERGENCE — native CLI accepts code that WASM type-checker rejects with "float vs int"
- M-ZERO-ARG-INVOCATION-SURFACES — Unify zero-arg export invocation across all surfaces
v0.21.0
- M-BYTES-TOINTS-BYTEAT: Add byte-to-int extraction primitives to std/bytes
- Sprint Plan: M-COG-RUNTIME-BROWSER
- M-COG-RUNTIME-BROWSER: Cognitive OS Browser Substrate
- Sprint Plan: M-COG-RUNTIME
- M-COG-RUNTIME: Cognition Runtime — Effects, Fabric, Event Log, Scheduler, DOM Replay
- M-DX26: Property Test "Empty Program" Bug
- M-SERVEAPI-GET-QUERY-SHADOW — GET query args silently shadowed by zero-value padding
- M-VERA-BENCH-INTEGRATION: Add AILANG as a target language in VeraBench
- M-VERA-BENCH-INTEGRATION Sprint Plan
- M-WASM-REFLECTIVE-RUNTIME: Cognitive OS (Umbrella)
v0.20.1
v0.20.0
- M-AILANG-LSP-FOR-AI Sprint Plan
- M-AILANG-LSP-FOR-AI — AILANG LSP server targeting AI coding agents (and humans)
- M-DX26 Phase 5:
ensuresResult Binding — Sprint Plan - M-PARSER-BLOCK-TRAILING-RECORD: Sprint Plan
- M-PARSER-BLOCK-TRAILING-RECORD: Block Expression Greedily Consumes Trailing
{as New Block Instead of Record Literal - M-SERVEAPI-SURFACE-DROPS Sprint Plan
- M-SERVEAPI-SURFACE-DROPS — Refuse to start when declared
@routemodules are silently dropped - Sprint Plan: M-TYPECHECK-NO-AUTO-UNWRAP-RESULT (v0.20.0)
- M-TYPECHECK-NO-AUTO-UNWRAP-RESULT: Reject
.fieldaccess on tagged unions withoutmatch
v0.19.2
- M-STDLIB-URL-ENCODE: URL form-encoding in std/net
- m stdlib xml walk perf sprint plan
- m stdlib xml walk perf
v0.19.1
- Sprint Plan: M-STDLIB-HTML —
std/htmlHTML5 Parser Module - M-STDLIB-HTML:
std/html— HTML5 Parser Module
v0.19.0
- Sprint Plan: M-EVAL-SWEET-SPOT
- Sprint Plan: M-EVAL-SWEET-SPOT-WEBSITE-INTEGRATION
- M-EVAL-SWEET-SPOT-WEBSITE-INTEGRATION: Surface sweet-spot data on the public benchmark dashboard
- M-EVAL-SWEET-SPOT: Eval Sweet-Spot Reporting & Failure-Category Distinguishability
- M-FS-SANDBOX-DIAGNOSTICS: Sandbox path rejection diagnostics
- M-NET-BINARY-BODIES — Sprint Plan
- M-NET-BINARY-BODIES: binary request and response bodies for
std/net - M-PKG-INTRA-PACKAGE-IMPORTS — Sprint Plan
- M-PKG-INTRA-PACKAGE-IMPORTS: intra-package imports for multi-module packages
- Sprint Plan: M-WASM-AI-STEP-BYO-KEY (v0.19.0)
- M-WASM-AI-STEP-BYO-KEY: WASM
ai.stepvia direct BYO-key provider fetch
v0.18.11
- M-CI-SETUP-AILANG-ACTION: official setup-ailang GitHub Action + install discoverability
- Sprint Plan: M-EXT-PORTABILITY-GATE (v0.18.11)
- M-EXT-PORTABILITY-GATE: pre-publish durability gate + asset bundling for extension packages
- M-PARSER-ROW-POLY-EFFECTS: row-extension
|sugar for effect annotations
v0.18.10
- Sprint Plan: M-MATCH-ADT-XCHECK (v0.18.10)
- M-MATCH-ADT-XCHECK: typechecker rejects match arms with constructors from a foreign ADT
v0.18.9
- M-AI-OPENROUTER-APP-ATTRIBUTION: per-request app attribution overrides for OpenRouter
- M-AI-OPENROUTER-REASONING-FIELD: surface reasoning from OR-routed thinking models
v0.18.8
v0.18.7
- Sprint Plan: M-AI-STEP-STREAMING (v0.18.7)
- M-AI-STEP-STREAMING: Typed-StepResult streaming via per-chunk callback
v0.18.6
v0.18.5
v0.18.4
- Sprint Plan: M-AI-PROMPT-CACHING (v0.18.4)
- M-AI-PROMPT-CACHING: Model-independent prompt-cache hints across all AI providers
- Sprint Plan: M-MOTOKO-EXECUTOR-ADAPTER
- M-MOTOKO-EXECUTOR-ADAPTER: motoko_agent as a CLI-Subprocess Executor
- Sprint Plan: M-PKG-AUTO-UPDATE-DX
v0.18.3
v0.18.2
- Sprint Plan: M-MOTOKO-PARALLEL-EXECUTION-ISOLATION
- M-MOTOKO-PARALLEL-EXECUTION-ISOLATION: Per-task isolation so
--agent-parallel ≥ 2works
v0.18.1
- Sprint Plan: M-MOTOKO-EVAL-HARNESS-HARDENING
- M-MOTOKO-EVAL-HARNESS-HARDENING: Close all 10 integration gaps surfaced by live smoke
v0.18.0
v0.17.x
v0.17.1
- Sprint Plan: M-AILANG-EXT-REGISTRY-GEN
- M-AILANG-EXT-REGISTRY-GEN:
ailang generate-extension-registrycommand - M-MOTOKO-EXTENSION-INTEGRATION Sprint Plan
v0.17.0
- M-AI-OPENAI-LOCAL-ENDPOINT-RELAX — Allow empty OPENAI_API_KEY when OPENAI_BASE_URL is custom
- M-EXTERNAL-CONSUMER-DX Sprint Plan
- M-EXTERNAL-CONSUMER-DX: Diagnostics & Artifacts for External AILANG Projects
v0.16.4
v0.16.3
- Sprint Plan: M-MOTOKO-INLINE-TEST-HARNESS
- M-MOTOKO-INLINE-TEST-HARNESS: Fix inline test harness for import-dependent helpers and aliased import collisions
v0.16.2
v0.16.0
- M-AI-OPENROUTER: OpenRouter Provider with Routable Model Resolution
- Sprint Plan: M-AI-OPENROUTER
- Sprint Plan — M-AILANG-FS-RESULT
- M-AILANG-FS-RESULT — Result-returning std/fs builtins
- Sprint Plan: M-PKG-AUTONOMOUS-CASCADE-SAFE
- M-PKG-AUTONOMOUS-CASCADE-SAFE: Validate + Harden the Autonomous Package Cascade
- M-PKG-CASCADE-DETERMINISTIC-FIRST
- Sprint Plan: M-STD-DEFLATE-ZLIB
- std/deflate: raw zlib/deflate primitives for PDF FlateDecode and wire protocols
- M-TAINT-TYPES — Sprint Plan
- M-TAINT-TYPES: Value-Labelled Information-Flow Types
v0.15.x
- Sprint Plan: M-AI-EFFECT-MODES
- M-AI-EFFECT-MODES: AI Effect Modes (mode=fixed|routeable|replay-only, scope=byok)
- Sprint Plan: M-CI-BUILD-SPEED (Cut multivac CI from ~25min to ~5min)
- M-CI-BUILD-SPEED: Cut multivac build time from ~25min to ~5min
- Sprint Plan: M-EFFECT-REFINEMENT-PHASE1
- M-EFFECT-REFINEMENT Phase 1: Parameterised Effect Syntax + Row Algebra (Rand pilot)
- M-MAC-NOTIFY-DAEMON — Sprint Plan
- M-MAC-NOTIFY-DAEMON: Mac Notification Daemon for Cloud Events
- Sprint Plan: M-SMT-CROSS-MODULE-FUNCTIONS
- M-SMT-CROSS-MODULE-FUNCTIONS: Z3 verification of imported function calls
v0.15.1
- M-AI-CALL-STREAM-HELPER Sprint Plan
- M-AI-CALL-STREAM-HELPER: Synchronous accumulator wrapper for AI token streaming
- M-BRAIN-BOOTSTRAP:
ailang micro-rag bootstrapfor fresh installs - M-EVAL-COST-AND-SPEED-BUDGETS — Sprint Plan
- M-EVAL-COST-AND-SPEED-BUDGETS
v0.15.0
- Sprint Plan: M-AGENT-MCP-ONBOARDING (Fix broken indexes + install/onboarding tools)
- M-AGENT-MCP-ONBOARDING: Fix broken indexes + add install/onboarding tools
- Sprint Plan: M-AGENT-MCP (Remote MCP Server for ailang.sunholo.com)
- M-AGENT-MCP: Remote MCP Server for ailang.sunholo.com
- M-AI-PROVIDER-CONFIG: Config-Driven AI Providers via Package Manifests
- M-AI-STREAMING-HELPER Sprint Plan
- M-AI-STREAMING-HELPER: AI Token Streaming Helper + Cross-Domain Discovery
- Sprint Plan: M-BRAIN-MICRORAG
- M-BRAIN-MICRORAG: Micro-RAG — Just-in-Time Knowledge Injection on Tool Calls
- M-CRYPTORAND: Cryptographic Randomness as a First-Class Effect
- M-EVAL-CROSS-HARNESS: Cross-Harness Benchmark Comparison
- M-EVAL-CROSS-HARNESS Sprint Plan
- Sprint Plan: M-EVAL-LANG-JSGO
- Sprint Plan: M-EVAL-LANGREG (Language Registry)
- Sprint Plan: M-EXEC-EXPAND (Codex + opencode Executors)
- M-EXEC-EXPAND: Codex + opencode Executors (Uniform CLI-Subprocess Shape)
- M-EXECUTOR-VARIANTS Sprint Plan
- M-EXECUTOR-VARIANTS: Per-Agent Docker Image Variants
- M-MCP-EDGE-THROTTLE: Per-IP rate limiting for the public MCP endpoint
- Sprint Plan: M-MICRORAG-EXPAND (μRAG hook coverage expansion)
- M-MICRORAG-HOOK-EXPANSION: μRAG hook coverage across harnesses + UserPromptSubmit
- M-OLLAMA-LOCAL-EVAL Sprint Plan
- M-OLLAMA-LOCAL-EVAL: Local Model Eval Harness Adaptation
- CII Best Practices badge — submission scratch
- CII Silver tier — feasibility triage
- M-SUPPLY-CHAIN-HARDENING-2 — Sprint Plan
- M-SUPPLY-CHAIN-HARDENING-2 — second pass on Scorecard findings
v0.14.3
v0.14.2
- M-BENCHMARK-SECTION: Multi-Page Benchmark Section for AILANG Website
- Sprint Plan: M-CLI-ARGS-WIN (v0.14.2)
- M-CLI:
-args-file, Stdin Input, and Windows Runtime CI - Sprint Plan: M-EXEC-PI (Pi Coding Harness Executor)
- M-EXEC-PI: Pi Coding Harness Executor
v0.14.1
- M-SONAR-GATE-CLEANUP: Drive SonarCloud Quality Gate to PASS
- M-SUPPLY-CHAIN-HARDENING — OpenSSF Scorecard response
v0.14.0
- M-BENCHMARK-SUITE-TIERS: Restructure the Eval Suite into Smoke / Core / Stretch Tiers
- M-EVAL-CATEGORY-ANALYSIS: Per-Category Benchmark Analysis & Curation
- Sprint Plan: M-EVAL-SUITE-PREP
v0.13.0
- M-CONCAT-DISAMBIG Sprint Handover
- Sprint Plan: M-CONCAT-DISAMBIG — Eliminate
++Ambiguity - M-CONCAT-DISAMBIG: Eliminate
++Operator Ambiguity Once and For All
v0.12.0
- Sprint Plan — M-OBS-TRACE-TRIAGE
- Observatory Trace Triage — Make Function-Level Spans Opt-In
- M-PERF6 Phase 4 Sprint Plan
- M-PERF6 Sprint Plan: Runtime Performance Hotspots
- M-PERF6: Runtime Performance Hotspots
- M-STDLIB-TAR-GZIP Sprint Plan
- M-STDLIB-TAR-GZIP: std/tar + std/gzip Native Archive Support
v0.11.4
- M-POLY-ORD-DEFAULTING: Polymorphic
Ord/Eqlambdas silently default toInt - Sprint Plan: M-POLY-ORD (Cache Fix + Ord/Eq Defaulting Regression)
- Specializer & Cloner Coverage Gaps (DEBUG_STRICT=1)
v0.11.3
- M-EVAL-SHORT-CIRCUIT-BOOL: Short-Circuit Evaluation for
&&and|| - M-PARSEFOLD-EARLY-TERMINATION: Short-Circuit Support for
parseFold/zipXmlScanFold - Sprint Plan: v0.11.3 Hotfix — Short-Circuit + parseFold Early Termination
v0.11.2
- Sprint Plan: M-INCREMENTAL-TYPECHECK
- M-INCREMENTAL-TYPECHECK: Compilation Skip via Artifact Serialization
- Sprint Plan: M-LAT-BUDGET — Latency Budgets & Dividend Ledger (Phase 0)
- M-LAT-BUDGET: Latency Budgets & Dividend Ledger
v0.11.1
v0.11.0
- M-BYTECODE-2D Benchmarks: Bytecode VM vs Tree-Walking Evaluator
- M-BYTECODE-2D M6 Parity Report
- M-BYTECODE-HOF-BUILTINS Sprint Plan
- M-BYTECODE-HOF-BUILTINS: VM Callback Mechanism for Higher-Order Builtins
- M-BYTECODE-LAMBDA-RESOLUTION Sprint Plan
- M-BYTECODE-LAMBDA-RESOLUTION: Fix Lambda Body Name Resolution in Bytecode Compiler
- M-BYTECODE-MULTIMODULE — Sprint Plan
- Sprint Plan: M-BYTECODE-PHASE2E
- M-BYTECODE-PURE-EFFECTS Sprint Plan
- M-BYTECODE-PURE-EFFECTS: Wire Pure-in-Practice Builtins to VM
- M-BYTECODE-REGALLOC-FIX Sprint Plan
- M-BYTECODE-REGALLOC-FIX: Fix Register Allocation Bugs
- Sprint Plan: M-BYTECODE-STDLIB-BUILTINS
- M-BYTECODE-VM: Bytecode Virtual Machine from Statement IR
- M-BYTECODE-XML-BUILTINS Sprint Plan
- M-BYTECODE-XML-BUILTINS: Wire XML Builtins to VM
- Sprint Plan: M-CODEGEN-STRATEGIC-REVIEW — Statement IR + Clean-Room Go Emitter
- M-CODEGEN-STRATEGIC-REVIEW: Honest Assessment of Compile-to-Go and the Path Forward
- M-DX-MODULE-PREFIX: module_prefix Ignored During Package File Resolution
- Sprint Plan: M-DX-SERVE-API-COERCION
- M-DX-SERVE-API-COERCION: Fix serve-api type coercion in cross-package function calls
- Sprint Plan: M-EVAL-XLANG — Cross-Language AI Code Generation Benchmark (Revised)
- M-EVAL-XLANG: Cross-Language AI Code Generation Benchmark
- M-LOWER-FIX: Statement IR Lowering Completeness for Bytecode VM
- M-MCP-QUALITY: MCP Server Quality & @route Header Access
- Sprint Plan: M-MCP-QUALITY
- Sprint Plan: M-PERF-DOCPARSE
- M-PERF-DOCPARSE: Close the Python Performance Gap for Document Parsing
- Sprint Plan: M-PERF-GOROUTINE-ID
- M-PERF-GOROUTINE-ID: Eliminate runtime.Stack() Bottleneck in Builtin Dispatch
- Sprint Plan: M-PHASE2A-BENCH — Evaluator Performance Benchmarks
- M-SERVEAPI-ROUTE-NOT-FOUND Sprint Plan
- M-SERVEAPI-ROUTE-NOT-FOUND: Typed ROUTE_NOT_FOUND error for unmatched serve-api paths
- Sprint Plan: M-STD-MAP
- M-STD-MAP: HashMap Type + Array Gaps
- Sprint Plan: M-STD-STRING-PERF
- M-STD-STRING-PERF: O(n) String Processing Primitives
- Sprint Plan: M-STREAMING-ZIP-XML
- M-STREAMING-ZIP-XML: Streaming ZIP Entry + XML Fold
- Phase 2A Benchmark Results: Evaluator vs Native Go
v0.10.12
v0.10.11
- M-MCP-CASCADE: MCP Tool Names → Module Path Collisions → Startup Perf — Postmortem & Forward Plan
- Sprint Plan: M-MCP-CASCADE
v0.10.2
v0.10.1
- M-BITWISE-OPS: Bitwise Operators and Integer Builtins
- Sprint Plan: Cons (
::) as Expression Operator - Cons (
::) as Expression Operator - Sprint Plan: M-MULTIPART-BYTES
- M-MULTIPART-BYTES: Multipart Upload BytesValue Handling in serve-api
- M-SERVE-API-TRANSITIVE-IMPORTS: Fix Transitive Import Resolution in serve-api
v0.10.0
- Sprint Plan: M-AI-IMAGE
- M-AI-IMAGE: AI Image Generation via std/ai
- Sprint Plan: M-BRAIN-CONTEXT
- M-BRAIN-CONTEXT: Contextual Brain Injection During Active Sessions
- M-DX-XPKG-RESOLVE: Cross-Package Stdlib Function Resolution Bug
- M-DX-PKG-CHECK: Package-Level Type Checking
- M-DX-PKG-TEST: Package-Level Test Runner
- M-DX-RELIMPORT: Local Import Syntax for Intra-Package Modules
- Sprint Plan: M-DX-ROUTE-CTX — HTTP Request Context for @route Handlers
- M-DX-ROUTE-CTX: HTTP Request Context for @route Handlers
- M-DX-UTF8: UTF-8 String Operation Correctness + replace() Builtin
- Sprint Plan: M-DX-XPKG-RESOLVE — Cross-Package Stdlib Function Resolution Bug
- Sprint Plan: M-EXIT-CODE
- M-EXIT-CODE: Process Exit with Explicit Exit Code
- Sprint Plan: M-OBS-RETENTION
- M-OBS-RETENTION: Observatory Database Retention & Growth Control
- Sprint Plan: M-PKG-AUTONOMOUS-UPDATES
- M-PKG-AUTONOMOUS-UPDATES: Message-Driven Autonomous Package Updates
- M-PKG-CI-PUBLISH: Auto-Publish Packages via Cloud Build
- Package Ecosystem Status — Post-Publication Audit
- Sprint Plan: M-PKG-EXPLORER — Package Explorer Website
- M-PKG-EXPLORER: Package Explorer Website
- M-PKG-INSTALL-LATEST: Support @latest in ailang install
- Sprint Plan: M-PKG-INTERREF
- M-PKG-INTERREF: Package Inter-Function References Fail When Loaded as Dependency
- Sprint Plan: M-PKG-LOCK-PORTABILITY
- M-PKG-LOCK-PORTABILITY: Make ailang.lock Portable Across Machines and Containers
- Sprint Plan: M-PKG-METADATA-URLS
- M-PKG-METADATA-URLS: Add Repository, Homepage, and License URL to Package Metadata
- M-PKG-RESOLVER-DIRECT-WINS: Direct Dependencies Override Transitive Versions
- Fix: ailang lock Transitive Registry Dependency Resolution
- M-PKG-UPGRADE-CHAIN-TOOLING: Package Upgrade Chain Diagnostics and Automation
- M-ROUTE-COLLISION: Route Collision Guard & Optional A2A
- M-ROUTE-NOWRAP: @nowrap annotation for unwrapped @route responses
- Sprint Plan: M-SERVE-API-AGENT-ENHANCEMENTS
- M-SERVE-API-AGENT-ENHANCEMENTS: Named Parameter Binding & Response Headers for Agent-Friendly APIs
- Sprint Plan: M-SERVE-API-ENDPOINT-FILTERING
- M-SERVE-API-ENDPOINT-FILTERING: Endpoint Filtering & JSON Auto-Unwrap for serve-api
- M-SERVE-API-GET-ARGS: GET Route Function Arguments via Query Parameters
- Sprint Plan: M-STDLIB-CRYPTO-JWT
- M-STDLIB-CRYPTO-JWT: RSA Signature Verification & JWT Support
- Sprint Plan: M-STDLIB-CRYPTO
- M-STDLIB-CRYPTO: Cryptographic Hashing for AILANG
- Sprint Plan: M-TYPE-ALIAS — Cross-Package Record Type Alias Unification
- M-TYPE-ALIAS: Cross-Package Record Type Alias Unification
- Sprint Plan: M-TYPE-V2-MIGRATION
- M-TYPE-V2-MIGRATION: Complete TFunc → TFunc2 Type System Migration
- M-VALIDATOR-PATTERN-BINDINGS: Validator Rejects Pattern Match Bindings
v0.9.11
- M-DX-APP: Application Package Adoption — Module Prefix Mismatch
- Sprint Plan: M-DX-PKG-CHECK + M-DX-PKG-TEST
- M-PKG-RESOLVER-DIRECT-WINS: Version Conflict Detection in Dependency Resolver
v0.9.10
v0.9.9
- M-PKG-MSG: Package Messaging Graph for Multi-Agent Coordination
- Sprint Plan: M-PKG-MSG — Package Messaging Graph
v0.9.8
- Sprint Plan: M-COST2 Dashboard Firestore Read Optimization
- M-COST2: Dashboard Firestore Read Optimization
v0.9.7
- Sprint Plan: M-PKG-REGISTRY — Client-Side Implementation
- M-PKG-REGISTRY: AILANG Package Registry on GCP
v0.9.6
v0.9.5
- Sprint Plan: M-PKG Phase 1 — Local Package Management
- M-PKG: AILANG Package System & Multi-Agent Coordination
- Sprint Plan: M-PKG Phase 1.5 — Tightening
- Sprint Plan: M-SERVE-API-ZERO-VALUE-PADDING
- M-SERVE-API-ZERO-VALUE-PADDING: Zero-Value Padding for Missing Named Parameters
v0.9.4
- M-SERVE-API-CONCURRENCY: Per-Request Evaluator for Thread-Safe serve-api
- Sprint Plan: M-SERVE-API-DX
- M-SERVE-API-DX: serve-api Developer Experience — Custom Routes, File Upload, Binary Response, Auth
- M-SERVE-API-JSON-DECODE: Fix TaggedValue arg decoding in serve-api
- M-STDLIB-UUID4: Add uuid4() to std/rand
- Sprint Plan: M-STDLIB-XML — XmlNode Constructors, writeEntryBytes, escapeXml Tests
- M-STDLIB-XML: XmlNode Constructors, writeEntryBytes, escapeXml
- M-TYPE-EFFECT-ROW: Fix Effect Row Unification Regression
v0.9.3
v0.9.2
- Sprint Plan: M-BRAIN — AILANG Persistent Semantic Cache
- Sprint Plan: M-BRAIN-VECTORS
- M-BRAIN-VECTORS: Vector-Native Brain with Embedding Storage and Cosine Search
- M-BRAIN: AILANG Persistent Semantic Cache ("The Brain")
- M-BUG-INLINE-TEST-EXTRACTION: Fix ExtractFunctionBinding for Module-less Files and Absolute Paths
- M-CLOUD-DUAL-AUTH: OAuth vs API Key Cloud Run Jobs
- Sprint Plan: M-CLOUD-JOB-RELIABILITY
- M-CLOUD-JOB-RELIABILITY: Silent Failure Detection & Recovery
- M-CLOUD-PLUGIN-SKILLS: Cross-Project Skill Distribution via Plugin System
- Sprint Plan: M-CLOUD-PROGRESS-TRACKING
- M-CLOUD-PROGRESS-TRACKING: Cloud Execution Visibility & Cost Controls
- M-CODEGEN-REGISTRY-ONLY: Delete Legacy Stdlib, Registry-Only Runtime
- M-CONTRACT-OPLOWERING-FIX: Complete OpLowering for Contract Expressions
- M-CONTRACT-PIPELINE-DX: Contract Pipeline Coverage & Developer Experience
- M-DEBUG-ERASURE: Implement Debug Ghost Effect Erasure in Release Mode
- Sprint Plan: M-DEBUG-ERASURE — Debug Ghost Effect Erasure
- Sprint Plan: M-DOCPARSE-DX Stdlib & DX Improvements
- M-DOCPARSE-DX: Stdlib & DX Improvements from DocParse Feedback
- M-GIT-GUARDRAILS: Per-Agent Git Mode for Cloud Coordinator
- M-HARNESS-COMMIT-CONTRACT: Website Builder Commit Message & Payload Contract
- Sprint Plan: M-HASH-COLLECTIONS Phase 1
- M-HASH-COLLECTIONS: Hash-Based Collections & Deterministic Equality
- Sprint Plan: M-ITERATIVE-LIST — Iterative List Builtins
- M-ITERATIVE-LIST: Iterative List Builtins for Large Collections
- Sprint Plan: M-PERF5 Data-Intensive Workload Performance
- M-PERF5: Data-Intensive Workload Performance
- Sprint Plan: M-PERF6 Compilation & Runtime Performance
- M-PERF6: Compilation & Runtime Performance (Phase 1)
- M-PERF7 Sprint Plan: DocParse Production Pipeline
- M-PERF7: DocParse Production Pipeline Optimization
- M-REST-INGESTION: REST API for Client Message Ingestion & Retrieval
- Sprint Plan: M-WASM-DICTIONARY-DISPATCH
- M-WASM-DICTIONARY-DISPATCH: Fix WASM Dictionary Dispatch Bugs
v0.9.1
- Sprint Plan: M-SMT-FRAGMENT-EXPANSION-V2 — Z3 Verification Phase 2
- M-SMT-FRAGMENT-EXPANSION-V2: Z3 Verification Phase 2
v0.9.0
- Sprint Plan: M-ASYNC-IO Phase 2 — asyncExecProcess
- Sprint Plan: M-ASYNC-IO Phase 3 — Subprocess Stdin Writing (ProcessHandle)
- M-ASYNC-IO Phase 3: Subprocess Stdin Writing (ProcessHandle)
- M-ASYNC-IO Phase 2: Subprocess Stdout as StreamSource (asyncExecProcess)
- Sprint Plan: M-ASYNC-IO Phase 1 — Multi-Source Event Multiplexer
- M-ASYNC-IO: Multi-Source Event Multiplexing for std/stream
- Sprint Plan: M-CLOUD-DISPATCH
- M-CLOUD-DISPATCH: Cloud Run Job Dispatch & Cloud Logging
- M-CLOUD-E2E-FIXES: Cloud End-to-End Wiring Fixes
- M-CLOUD-E2E: Cloud End-to-End Message Flow
- M-CLOUD-ENDPOINT-AUTH: Secure Coordinator Status Endpoints
- Sprint Plan: M-CLOUD-HEALTH — Cloud Run HTTP Health & Status Endpoints
- M-CLOUD-HEALTH: Cloud Run HTTP Health & Status Endpoints
- M-CLOUD-INFRA: GCP Cloud Infrastructure for AILANG Services
- Sprint Plan: M-HTTP-HOOKS-CLOUD-TELEMETRY
- M-HTTP-HOOKS-CLOUD-TELEMETRY: Claude Code HTTP Hooks for Cloud-Native Telemetry
- Sprint Plan: M-MODULE-SCOPE — Per-module Environment Isolation
- M-MODULE-SCOPE: Non-exported function name collision across modules
- M-PERF-OBSERVATORY: Tiered Loading for Chain Performance
- M-PROTOCOL-SUPPORT: OpenAPI, MCP, and A2A for
ailang serve-api - Sprint Plan: M-PUBSUB — Pub/Sub Cloud Messaging Transport
- M-PUBSUB: Pub/Sub Cloud Messaging Transport
v0.8.1
- M-CHAIN-TRANSCRIPT: Session Transcript Viewing via ailang chains
- M-CLOUD-STORAGE: Cloud Database Migration
- Sprint Plan: M-CONTRACT-EVAL — Contract-Guided Evaluation Harness
- M-CONTRACT-EVAL: Contract-Guided Evaluation Harness
- M-CONTRACTS-ASSERT: Contract-Based Preconditions
- M-DEV-TOOLS-AI-DOC: AI-Consumable Dev Tools Documentation
- M-DX-TEST-HARNESS-NIL: Inline Tests Fail with Stdlib Imports
- M-EVAL-CHAINS: Chains as Single Source of Truth for Agent Evals
- M-EXECUTOR-IDLE-TIMEOUT: Activity-Based Idle Timeout for Executors
- M-PARSER-EXPORT: Single-Constructor ADT Cursor Bug Drops
exportFlag - M-PERF3: Performance Quick Wins for 1.0.0
- M-PROCESS: std/process Module for External Command Execution
- Sprint Plan: M-PROCESS — std/process Module
- M-SEMANTIC-ENVELOPE: Multi-Aspect Semantic Embeddings for Agent Messaging
- M-STREAM-BIDI: Generic Bidirectional Streaming Primitives
- Sprint Plan: M-STREAM-BIDI Phase 1 — Core WebSocket Streaming
- M-STREAM-DX: Streaming Demo DX Improvements
- Sprint Plan: M-STREAM-DX
- Sprint Plan: M-STREAM-PHASE2-DX — Stream Typed ADTs & DX Improvements
- M-STREAM-PHASE2-DX: Stream Typed ADTs & DX Improvements
- M-VERIFY-FLOAT-CMP: Fix Panic on Float Comparisons in Contract Verification
- M-VERIFY: SMT Verification & Redundant Generation (Phase 1-3)
- Sprint Plan: M-WASM-CLOSURE-ENV
- M-WASM-CLOSURE-ENV: Fix Closure Environment Resolution in WASM Bridge
- Sprint Plan: M-WASM-STREAM-BRIDGE
- M-WASM-STREAM-BRIDGE: WASM Stream Effect Handler Bridge
v0.8.0
- M-CONTRACTS-OPLOWERING: Apply OpLowering to Contract Expressions
- M-GEMINI-TRACE: Gemini CLI Trace Investigation
- M-SEM-KERNEL: AILANG as a Symbolic Reasoning Kernel
- Sprint Plan: M-SMT-BACKEND-MVP
- M-SMT-BACKEND-MVP: SMT Verification Backend (Phase 1)
- M-SMT-BOUNDED-RECURSION: Bounded Unrolling for Recursive Function Verification
- Sprint Plan: M-SMT-FRAGMENT-EXPANSION Phase A — Cross-Function Calls
- M-SMT-FRAGMENT-EXPANSION: Expanding the SMT Decidable Fragment
- Sprint Plan: M-SMT-LISTS — List Verification (Phase D)
- Sprint Plan: M-SMT-RECORD-DISCOVERY
- M-SMT-RECORD-DISCOVERY: SMT Record Type Discovery from Bodies and Return Types
- Sprint Plan: M-SMT-RECORDS — Record Verification via SMT-LIB Datatypes
- Sprint Plan: M-SMT-STRINGS — String Verification (Phase B)
- M-TELEMETRY-HOOKS: Worktree Session Enrichment - Handoff Document
- M-TRACE-EXPORT Phase 2-4: OTEL Integration, Replay, and Training Export
- M-TRACE-EXPORT Phase 5: Instrumentation Gaps & Polish
- M-TRACE-EXPORT: Execution Trace Export for AI Training
- Sprint Plan: M-TRACE-EXPORT Phase 2 — OTEL Span Emission
- Sprint Plan: M-TRACE-EXPORT Phase 3 — Trace Replay
v0.7.4
- M-CHAINS-SOURCE-OF-TRUTH Phase 3: Chain Metric Rollup
- M-CHAINS-SOURCE-OF-TRUTH Phase 3: Chain Metric Rollup - Sprint Plan
- [Feature Name]
- Fix: Inline Tests Fail with Complex Imports
- Sprint Plan: Fix Inline Tests with Complex Imports
- M-EVAL-GAP: AILANG vs Python Parity Analysis
- M-EVAL-GAP-FIXES: Address Language Gaps from v0.7.0 Eval Analysis
- M-PIPELINE-STANDALONE-EXPR: Standalone Expression Evaluation
- Sprint Plan: M-STDLIB-GAPS
- Design Doc: M-STDLIB-GAPS - Standard Library Gap Analysis
v0.7.3
- AILANG Demo: Invoice Processor with WebAssembly
- Sprint Plan: M-EFFECTFUL-LIST-COMBINATORS
- M-EFFECTFUL-LIST-COMBINATORS: Effectful map/filter/fold for std/list
- Sprint Plan: M-STDLIB-XML — XML Parsing Standard Library
- M-STDLIB-XML: XML Parsing Standard Library
- Sprint Plan: M-STDLIB-ZIP — ZIP Archive Standard Library
- M-STDLIB-ZIP: ZIP Archive Standard Library
- Sprint Plan: M-STRUCTURED-AI-OUTPUT
- M-STRUCTURED-AI-OUTPUT: Structured output in std/ai
v0.7.2
- M-CHAINS-SOURCE-OF-TRUTH: Make
ailang chainsthe Canonical Data CLI - M-WASM-STDLIB: Embed Standard Library in WASM Binary
v0.7.1
- E2E Test: Coordinator Handoff Chains - Sprint Plan
- Hello World Feature
- M-CHAT-HISTORY: Claude Code Conversation History Integration
- M-CHAT-HISTORY Sprint Plan
- M-DETERMINISTIC-CHAT-LINKING: Deterministic Task-to-Chat Message Linking
- Inline Tests Documentation & Examples (M-DX23)
- M-DX24: Developer Experience Improvements - BigQuery Connector Feedback
- Sprint Plan: M-DX25 Scoped Budgets with Dual Counters & Budget Reporting
- M-DX25: Scoped Budgets with Dual Counters & Budget Reporting
- M-HOT-RELOAD: Hot Reload for serve-api
- Sprint Plan: M-HOT-RELOAD - Hot Reload for serve-api
- M-POLY-ADT: Polymorphic ADT Type Inference with Option
- M-POLY-ADT Sprint Plan: Fix Polymorphic ADT Constructor Types
- M-SERVE-API: Auto-Serve AILANG Exports as REST API
- Sprint Plan: M-TRACE-BRIDGE
- M-TRACE-BRIDGE: Bridge OTEL Spans to TraceRegistry
- M-VERIFY-CONTRACTS: Runtime Contract Enforcement (Language-Wide)
- M-WASM-REPL: Browser Module Loading for WASM REPL
- M-WASM-REPL Sprint Plan: Browser Module Loading
- Sprint Plan: M-WORKSPACE-ACCESS
- M-WORKSPACE-ACCESS: Multi-Workspace Access Control with Public/Private Visibility
- Sprint Plan: Test Chain Environment Propagation
- Test Chain Environment Propagation
- Sprint Plan: M-TRACE-TEST - Trace Testing Framework
- Trace Test
v0.7.0
- AILANG Dogfooding: Gaps Discovered
- M-EVAL-AGENT-QUEUE: Queue-Based Agent Evaluation Architecture
- Retrospective: Observability Dashboard Development
- Concatenation Operator (
++) Type Inference Bug - Evaluation Dashboard Reliability
- M-AI-OLLAMA Sprint Plan
- M-AUTH Sprint Plan: Firebase Authentication & Dashboard Authorization
- M-AUTH: Firebase Authentication & Role-Based Authorization for Dashboard
- M-BUILTIN-SAFETY: Defensive Type Checking in Builtin Implementations
- M-CODEGEN-LIST-TYPE-DEFINITION: Fix Undefined List Type in Generated Code
- M-CODEGEN-LIST-TYPE: Sprint Plan
- M-COLLAB-PROVIDER-STATS Sprint Plan
- M-COLLAB-PROVIDER-STATS: Provider & Workspace Tags and Statistics for Collaboration Hub
- M-CONTROL-PLANE-INTERACTIVE-FILTERING
- Sprint Plan: M-CONTROL-PLANE-V4-INTEGRATION
- M-CONTROL-PLANE-V4-INTEGRATION: Unified Telemetry for Agent Monitoring
- M-CONTROL-PLANE-V4: AILANG Operations Control Plane
- Sprint Plan: M-COORD-ARTIFACT-DISCOVERY
- M-COORD-ARTIFACT-DISCOVERY: Fix Coordinator Skill Invocation and Artifact Discovery
- M-COORD-AUTO-REVISION: Automatic Agent Revision on GitHub Feedback
- Sprint Plan: M-COORD-GENERIC-WORKFLOWS
- M-COORD-GENERIC-WORKFLOWS: Configurable Agent Workflows
- M-COORD-HUMAN-LOOP Sprint Plan
- M-COORD-HUMAN-LOOP: Human-in-the-Loop Coordinator Workflow
- Sprint Plan: M-COORD-TASKCHAIN-TESTS
- M-COORD-TASKCHAIN-TESTS: Add TaskChain Test Coverage
- M-COORD-UI-APPROVALS: Dashboard Coordinator Approval Workflow
- M-COORDINATOR: Always-On Autonomous Development Daemon
- M-COORDINATOR-TEST-WORKFLOW: Simple Coordinator Test File Creation
- Sprint Plan: M-DASHBOARD-APPROVAL-INTEGRATION
- M-DASHBOARD-APPROVAL-INTEGRATION: Complete Dashboard Coordinator Approval Workflow
- M-DASHBOARD-CLI: Dashboard API CLI Commands
- M-DASHBOARD-DOGFOODING: AILANG Dashboard Integration
- M-DB-CLEANUP: Database Schema Cleanup and Consolidation
- Sprint Plan: M-DEPRECATE-AILANG-AGENT
- M-DEPRECATE-AILANG-AGENT: Deprecate Standalone Agent Binary
- Sprint Plan: M-DX24 - Developer Experience Improvements from BigQuery Connector Feedback
- M-DX24 Sprint Plan: Developer Experience Improvements
- M-EMBED: Generic AILANG Embedding API
- M-AI-OLLAMA: Unified Ollama Provider for Local Models
- M-EXECUTOR-TEST: Comprehensive Test Coverage for Multi-Executor System
- M-GAP1: Fix Teaching Prompt Lambda Syntax
- M-GAP2: Fix Final Expression Lambda Arity Bug
- Sprint Plan: M-GAP2 Fix Path-Dependent Lambda Arity Bug
- Sprint Plan: M-GAP2 Final Expression Lambda Arity Bug
- Sprint Plan: M-GAP4 - Record Width Subtyping via Row Polymorphism
- M-GAP4: Record Width Subtyping via Row Polymorphism
- Sprint Plan: M-GAP5 - Add
repeatto std/string - M-GAP5: Add
repeatto std/string - Sprint Plan: M-GAP6 - Add maximum/minimum to std/list
- M-GAP6: Add
maximumandminimumto std/list - M-GENERIC-PIPELINE: Config-Driven Agent Pipelines
- Sprint Plan: M-JSON-HELPERS
- M-JSON-HELPERS: Convenience Functions for JSON Access
- Sprint Plan: M-MSG-ROUTING - Message Routing DX Improvements
- Sprint Plan: M-OPENAI-RESPONSES-API Completion
- Sprint Plan: M-OPENAI-RESPONSES-API
- M-OPENAI-RESPONSES-API: OpenAI Responses API & Agent Module
- M-OTEL-CROSS-PROCESS: Cross-Process Trace Linking
- Sprint Plan: M-OTEL-CROSS-PROCESS
- M-OTEL-DASHBOARD Implementation Report
- Sprint Plan: M-OTEL-ENHANCED-TRACING-DX
- M-OTEL-ENHANCED-TRACING-DX: Human-Friendly Tracing for Debugging
- M-OTEL-EXTENDED: Extended OpenTelemetry Instrumentation
- M-OTEL-EXTENDED Sprint Plan
- Parser Infinite Loop Guard
- Sprint Plan: M-PARSER-LOOP - Parser Infinite Loop Guard
- Sprint Plan: M-POLY-ARITH - Fix Polymorphic Arithmetic Operators
- M-POLY-ARITH: Fix Polymorphic Arithmetic Operators in Lambdas
- M-SCRIPT-INVOKE Sprint Plan
- M-SCRIPT-INVOKE: Deterministic Script Execution for Agent Workflows
- M-SESSION-WORKSPACE: Claude Code Rich Telemetry via Hooks
- Sprint Plan: M-STDLIB-DATETIME
- M-STDLIB-DATETIME: Date and Time Standard Library
- Sprint Plan: String Reverse Builtin (
_string_reverse) - String Reverse Builtin (
_string_reverse) - M-TASK-HIERARCHY Follow-ups
- M-TASK-HIERARCHY: Task-to-Trace Linking
- M-TASK-HIERARCHY Sprint Plan
- Sprint Plan: M-TRANSCRIPT
- M-TRANSCRIPT: Unified Conversation History & Feedback Loop
- Observatory Architecture
- AILANG v0.6.3 Bug Fixes
v0.6.2
- Sprint Plan: DX-17 Phase 2 - Normalize TList to TApp
- DX-17 Phase 2: Normalize TList to TApp at Parse Time
- M-BUG-EFFECT-CHECKER-CONFLATION: Effect Checker Incorrectly Requires IO for Pure Functions
- Sprint Plan: M-BUG-EFFECT-CHECKER-CONFLATION
- Sprint Plan: M-CAPABILITY-BUDGETS
- M-CAPABILITY-BUDGETS: Resource-Bounded Effects
- M-CODEGEN-BOOL-ASSERTIONS: Type assertions for boolean contexts in Go codegen
- M-CODEGEN-DICTIONARIES Sprint Plan
- M-CODEGEN-DICTIONARIES: Generate Type Class Dictionary Implementations for Go
- M-CODEGEN-VALUE-TYPES: Size-Based Pointer vs Value Strategy
- Sprint Plan: M-COORD-APPROVALWATCHER-OBSERVABILITY
- M-COORD-APPROVALWATCHER-OBSERVABILITY: ApprovalWatcher Debug and Observability
- M-COORD-GITHUB-AUTO-ROUTING Sprint Plan
- M-COORD-GITHUB-AUTO-ROUTING: GitHub-Driven Autonomous Workflow
- Sprint Plan: M-COORD-GITHUB-AUTO-ROUTING Completion
- Sprint Plan: M-COORD-STABLE
- M-COORD-STABLE: Coordinator Daemon Stability & Agent Collaboration Foundation
- M-COORD-FEEDBACK Sprint Plan
- M-COORD-FEEDBACK: Coordinator Executor Feedback Loop
- M-COORDINATOR-QUESTION-MODE: Read-Only Execution for Questions
- M-DX11 Phase 2: Type Checker Debug Event Emission
- M-DX11 Phase 2 Sprint Plan: Debug Event Emission
- M-DX17-LIST-NORMALIZATION-BUG: List[T] Not Normalized to Lowercase
- M-DX19: Auto-Derive Eq for ADT Types
- M-DX19 Sprint Plan: Auto-Derive Eq for ADT Types
- M-DX22 Sprint Plan: ADT Constructor Resolution
- M-DX22: Go Codegen - ADT Constructor Resolution Ambiguity
- M-GITHUB-USER-OVERRIDE: GitHub Account Override Flag
- M-OTEL: OpenTelemetry OTLP Integration
- M-OTEL Sprint Plan: OpenTelemetry OTLP Integration
- Sprint Plan: M-PATTERN-GUARDS
- M-PATTERN-GUARDS: Pattern Guard Evaluation in Codegen
- M-RECORD-PATTERNS Sprint Plan
- M-RECORD-PATTERNS: Record Pattern Matching
v0.6.1
- M-BUG: List Length Returns Sum Instead of Count
- Sprint Plan: M-BUG-LIST-LENGTH - Resolution Documentation
- Sprint Plan: M-CODEGEN-ADT-DOUBLE-PAREN
- M-CODEGEN-ADT-DOUBLE-PAREN: Fix Empty Double-Paren ADT Constructor Calls
- Sprint Plan: M-DX Check Directory Support
- M-DX: Add Directory Support to
ailang check - M-DX16: Inline Record Literals in Match Arms
- M-DX16 Sprint Plan: Inline Record Literals in Match Arms
- M-DX17: Go Codegen - ConcatList Undefined + Closure Variable Scoping
- M-DX18: Go Codegen - Non-Exported Function Namespacing
- M-DX20: Wildcard Pattern Type Inference in List Patterns
- M-DX21: Show Stdlib Version Warning Once Per Session
- M-ENV-TYPE Sprint Plan: Fix EnvError Type Parameter Bug
- M-ENV-TYPE: Fix std/env EnvError Type Parameter Bug
- M-EXEC-GEMINI Sprint Plan: Gemini CLI Executor Implementation
- M-EXEC: Multi-Executor Support for AI Coding Agents
- M-LETREC-SCOPING: Fix Letrec Recursive Binding Scope Regression
- Sprint Plan: M-LETREC-SCOPING Fix
- M-VERIFY: Runtime Contract Checks (Phase 0 + 0.5)
- Sprint Plan: M-VERIFY Phase 0 + 0.5 - Contract Infrastructure & Runtime Checks
v0.6.0
- DX-15: Shared Memory + Near-Duplicate Caching MVP
- DX-16: SharedIndex — Deterministic Semantic Retrieval for SharedMem
- Sprint Completion Report: DX-16 SharedIndex — Deterministic Semantic Retrieval
- DX-17: TList/TApp Type Unification - Sprint Plan
- DX-17: TList/TApp Type Unification Bug (Phase 1)
- M-CODEGEN-UNIFIED-SLICE-CONVERTERS: Complete Typed Slice Conversion
- M-DOC-SEM: Lazy Embeddings for Neural Doc Search
- Sprint Plan: M-DOC-SEM Lazy Embeddings for Neural Doc Search
- M-DOCS-ACCURACY: Website Documentation Accuracy Update
- Sprint Plan: M-DX11-DEBUG-TYPES-CLI (Type Inference Debug CLI)
- M-DX11-DEBUG-TYPES-CLI: Type Inference Debug CLI
- Sprint Plan: M-DX11-TYPE-PROVENANCE
- M-DX11-TYPE-PROVENANCE: Type Origin Tracking
- M-DX27: Fix Invalid Bool Type Assertion in Nested Match
- M-DX29: Fix Nested ADT Type in Option Pattern Match
- M-DX30: Add Missing EqString Runtime Helper
- M-JSON-ACCESSORS: Enable JSON Accessor Functions
- M-MSG-SEMANTIC: Semantic Search + Safe Deduplication for AILANG Messages
- M-MSG-SEMANTIC Sprint Plan: Semantic Search for AILANG Messages
- M-RT1: List Pattern Match Return Value Bug
- Sprint Plan: M-TYPENAME-NESTED-PROPAGATION
- M-TYPENAME-NESTED-PROPAGATION: TypeName Propagation to Nested Record Literals
- Semantic Caching: Implementation Status (v0.5.11)
v0.5.11
v0.5.10
- Codegen Bug Pattern Analysis: v0.5.10
- M-CODEGEN-ADT-TYPE-ASSERT: Fix Invalid Type Assertions on ADT Constructor Results
- M-CODEGEN-CROSS-MODULE: Cross-Module Function Calls Use _impl
- M-CODEGEN-LIST: Flatten List Literals and Bool Matches
- Sprint Plan: M-CODEGEN-LIST - Flatten List Literals and Bool Matches
- M-CODEGEN-NESTED-RECORD-TYPE: Fix Nested Record Type Selection
- M-CODEGEN-RECORD-TYPENAME-PRESERVATION: Fix TypeName Loss During Substitution
- M-CODEGEN-TUPLE: Tuple Pattern Matching in Go Codegen
- M-CROSS-MODULE-RECORD-UNIFICATION: Fix Type Unification for Cross-Module Nested Record Fields
- Sprint Plan: M-DX11-CYCLES (Type Graph Cycle Detection)
- M-DX11-CYCLES: Type Graph Cycle Detection Command
- M-DX11: Type Inference Debugging Tools
- Sprint Plan: M-DX11-TYPE-REPORT (Type Inference Debug Primitive)
- M-FIX-FLOAT-OP: Float Operator Dispatch in Pure Functions
- Fix Type Application in Return Type Annotations (Option[T])
- String Conversion Functions (floatToStr, intToStr)
- M-UNIFIED-AI-PROVIDERS: Unified AI Provider Architecture
- Sprint Plan: M-UNIFIED-AI-PROVIDERS
v0.5.9
- M-CODEGEN-FLAT-IF-ELSE: Flatten Nested Closure Chains in If-Else Codegen
- M-CODEGEN: Array Runtime Typed Slice Support
- M-CODEGEN-OPTION-RUNTIME-HELPERS: Fix GetOpt to Return *Option Instead of Map
- M-CODEGEN-POINTER-RETURN-TYPES: Fix Type Assertion Failures for Record Returns
- M-CODEGEN-STDLIB-MATH: Go Codegen Missing std/math Function Definitions
- M-CODEGEN-V2: Flat Go Code Generation
- Sprint Plan: M-CODEGEN-V2 - Flat Go Code Generation
- M-DX11: Cyclic Type Diagnostics & Hang Prevention
- Sprint Plan: M-DX11 Cyclic Type Diagnostics
- Sprint Plan: M-DX11-TRAVERSE - Safe Type Traversal Library
- M-DX11-TRAVERSE: Safe Type Traversal Library
- M-FIX-FLOAT-OP: Complete Fix Summary
- M-FIX-IF-ELSE-LET: Better Error for Let in If-Else Branches
- M-FIX-NESTED-RECORD-LIST: Nested Record List Type Inference Bug
- M-MSG-GITHUB: GitHub Issues Backend for Messaging System
- Sprint Plan: M-MSG-GITHUB - GitHub Issues Backend
v0.5.8
- Sprint Plan: M-BUGFIX v0.5.8 Bug Fix Sprint
- M-CODEGEN-BLANK: Fix Blank Identifier and Return Type Codegen Bugs
- M-CODEGEN-BOOL-SLICE: Add Bool Slice Converter
- M-CODEGEN-LIST-CONCAT: Add List Concatenation Runtime Helper
- M-CODEGEN-TYPE-ASSERTIONS: Fix _impl Functions Calling Typed Exports
- M-CODEGEN-TYPED-PARAMS: Fix Typed Parameters Becoming interface
- M-CODEGEN-ZERO-ARG: Fix Zero-Arg Function Code Generation
- M-PERF1: Effect Checker Performance for Large Array Literals
- M-PERF2: Cyclic Type Traversal Hang - Post-Mortem Analysis
- M-PERF2: Operator Lowering Hang on Complex Modules
- M-TYPE-ALIAS: Type Alias Expansion in ADT Variant Parameters
v0.5.7
- M-CODEGEN-TYPED-SLICES: Runtime List Functions for Typed Slices
- M-DX11: Better Stdlib Discovery for AI Agents
- typesIdentical Performance Bug (Large Records)
v0.5.6
- M-EVAL-GUARD: Eval Process Guardrails & Orphan Detection
- M-MSG: Agent Messaging System Improvements
- M-MSG Sprint Plan: Agent Messaging System Rewrite
- M-TYPE1: Array/TArray Type Unification Bug
- M-TYPE1 Sprint Plan: Array/TArray Type Unification Bug Fix
v0.5.5
- M-DX23: Typed Function Signatures
- M-DX24: Typed Function Bodies
- M-DX25: Typed Let Bindings
- M-DX26: Typed Wrapper Architecture
v0.5.4
- M-DX17: Integer Literals Generated as int Instead of int64
- M-DX18: RecordAccess Uses Map Syntax on Typed Structs
- M-DX22: Auto-Generate ADT Slice Converters
v0.5.3
- M-DX11: Named ADT Constructor Fields for Go Codegen
- Sprint Plan: M-DX11 Relaxed Module Matching
- Sprint Plan: M-DX12 Typed World Boundary Marshalling
- M-DX12: Typed World Boundary Marshalling
v0.5.2
- M-DX11: Relaxed Module Matching
- M-GAME-B Phase 2: Complete Go Codegen for Game Modules
- M-GAME-B Phase 2 Sprint Plan
v0.5.0
- M-AGENT-PROTOCOL: Agent-to-Agent Communication Protocol
- M-EVAL-AGENT: Multi-Agent Eval Benchmark Suite
- M-GAME-A: Go Codegen Foundation
- M-GAME-B: Effects for Games
- M-GAME-C: Compiler UX & Extern Functions
- M-GAME-D: Sim Example & Integration
- M-GAME-E1: Debug Effect (Foundational Trace Substrate)
- M-GAME-E2: AI Effect (General-Purpose AI Oracle)
- Sprint Plan: M-GAME-ENGINE - Game Support Enablement
- AILANG Consumer Contract v0.5.x
- Consumer Contract Gap Analysis
- gameengine
- Sprint Plan: M-BUG-FIXES (TVar2 List Pattern + ADT Test Harness)
- M-BUG-NESTED-RECORD-ANF: ANF Completion for Let RHS
v0.4.8
- Sprint Plan: M-BUG-LOCAL-IMPORTS - Fix Local Module Import Resolution
- M-BUG-LOCAL-IMPORTS: Fix Local Module Import Resolution
- Sprint Plan: M-BUG-RECORD-UPDATE-INFERENCE
- M-BUG-RECORD-UPDATE-INFERENCE: Fix Record Update Type Inference
- Sprint Plan: M-BUG-RECURSION-DEPTH
- M-BUG-RECURSION-DEPTH: Fix Unexpected Recursion Depth Overflow
- M-IMPORT-ALIASING: Module and Symbol Aliasing
v0.4.7
v0.4.6
- String split() Implementation Summary (v0.4.6)
- M-LANG-CLI-ARGS Sprint Completion Report
- Sprint Plan: M-LANG-CLI-ARGS (v0.4.6)
- M-LANG-CLI-ARGS: Command-Line Arguments Access
- M-DX10: Complete S-CALL0 and Unit-Argument Model
- M-DX11: String Split Builtin
- UI Collaboration Hub - Human-AI Orchestration Interface
v0.4.5
- Sprint Plan: M-BUG-NULLARY - Nullary Constructor Pattern Matching Fix
- Agent Execution Integration
- Nullary Constructor Pattern Matching Bug
v0.4.3
- M-SOUNDNESS Effect Checking Implementation - Completion Report
- Sprint Plan: M-SOUNDNESS Effect Checking Implementation
- M-SOUNDNESS: Effect Checking for Entry Module Prelude
v0.4.0
v0.3.25
v0.3.23
v0.3.22
v0.3.20
- M-CLAUDE-CODE-HEADLESS Sprint Plan
- M-CLAUDE-CODE-INTEGRATION: Hooks & Inbox (Phases 1-2)
- M-CLAUDE-CODE-INTEGRATION: Document Split
v0.3.18
- M-DX4 Sprint Plan: CoreTypeInfo Completeness (REFINED)
- Sprint Plan: M-DX4 CoreTypeInfo Population Gaps
- M-DX4: CoreTypeInfo Completeness & Type-Guided Lowering
v0.3.17
- AILANG: Compilation Failures
- Sprint Plan: M-COMPILE-ERROR - Enhanced Error Messages for AI Code Generation
v0.3.16
- M-DX2: Operator Development Experience Improvements - COMPLETE ✅
- M-DX2 Milestone 1: Type-Guided Lowering - COMPLETE ✅
- M-DX2 Milestone 2: Core IR Helpers - COMPLETE ✅
- M-DX2 Milestone 3: Debug CLI - COMPLETE ✅
- M-DX2 Milestone 4: Better Runtime Errors - COMPLETE ✅
- Lambda Expressions Example Refactor
- List Concatenation Operator Fix
- M-EVAL Round-Robin: Better Parallel Distribution Across Model Providers
v0.3.15
v0.3.14
- 🛡️ Bulletproof Regression Guard System - COMPLETE
- M-LANG: JSON Decode Implementation (v0.3.14)
- Agent Inbox - Quick Reference
- Agent Inbox Message Acknowledgment System
- SessionStart Hook for Agent Inbox
- List Pattern Spread Syntax
- Tier 0 Completion Summary - v0.3.14
v0.3.10
- M-DASH: Eval Dashboard Reliability
- M-DX1: Developer Experience Improvements
- M-EVAL-LOOP: Self-Improving AI Feedback Loop
- 1. Rebuild (every time)
- M-DX1 Day 3: Polish & Migration (v0.3.10+)
- M-DX1 Future Polish (v0.3.15+)
- M-LANG: Implement
show()Builtin Function
v0.3.9
v0.3.6
- Design Doc Deduplication: Implementation Summary
- Enhancement: Design Doc Deduplication & Merging
- AI Eval → Design Doc Generation: Implementation Summary
- Prompt suggested:
v0.3.5
- AI Usability Assessment: Is AILANG "The Language AIs Prefer"?
- v0.3.5: Functional Completeness Sprint
v0.3.2
v0.3.0
- M-R4: Recursion Support
- M-R5: Records & Row Polymorphism
- M-R7: Type System Fixes (Integral & Float Comparison)
- M-R8: Block Expressions
- M-EVAL-CAPS: Capability Specification in Benchmark YAML
- M-EVAL: AI Usability Benchmarking Harness
- v0.3.0 Implementation Plan
v0.3
- AILANG v0.3.3: Float Equality OpLowering Fix
- Add
letrecSyntax for Recursive Lambdas - AILANG v0.3.4+ Strategic Direction - Implementation Audit
- Audit Report: 20251013_next_steps_audit.md Status
- Numeric Coercion for Mixed-Type Arithmetic
- v0.3.11 Critical Regression Fix Summary
- Float Equality Bug - Investigation & Resolution Report
- M-R5b: Record Extension & Update Syntax
- M-R6: Clock & Net Effects
- M-REPL0: REPL Basic Stabilization
- Dashboard Workflow Improvements (v0.3.12)
- PYTHON: Runtime Errors [RESOLVED]
v0.2.0
- M-P1: Parser Baseline - COMPLETED ✅
- M-P2 Implementation Progress Report
- M-P2: ADT Syntax + Tuples - 5-Day Plan
- M-P3: Pattern Matching Foundation (Phase 3.0)
- M-P4: Effect System (Type-Level Only) - 4 Days
- M-P5: Pattern Matching in Function Bodies - Sprint Plan
- M-S1B: Fix Export System & Builtin Visibility
- M-S1 Polish: Ship v0.1.0 in 72 Hours
- M-S1: Stdlib Implementation Sprint Plan
- M-S1 Current Status - Stdlib Implementation
- M-S1: Stdlib Implementation Status
- Parser Next Steps: Pattern Matching in Function Bodies
- AILANG v4.0: The AI-Enhanced Design
- emits main.airun (signed), main.sbom.json, main.ledger.min.json
- M-R1: Module Execution Runtime (v0.2.0)
- M-R2: Minimal Effect Runtime
- M-R3: Pattern Matching Polish
- M-R3: Pattern Matching Polish
- milestonea2
- self improvement goal
- Sprint Plan: M-R1 Phase 5 - Function Invocation & stdlib Support
- Sprint Plan: M-R1 Phase 5 - Function Invocation & stdlib Support (REFINED)
- AILANG v0.1.0 MVP Roadmap
- AILANG v0.2.0 Implementation Plan
- when to switch from go
v0.0.12
v0.0.4
v0.0.3
- V0 0 3 IMPLEMENTATION REPORT
- gpt 5 appendix
- gpt5 reference code
- AILANG: The AI-First Programming Language
Generated at build time. 1182 design documents across 152 versions.