Aleph Skill v2.1.1

Aleph Skill v2.1.1

Aleph 2.1.1 is a focused bundled-component update. It moves the internal D Research runtime from 3.2.0 to 3.2.1 while keeping Aleph workspace schema_version and formula_version at 2.0.0. Hosts still install and load one Aleph skill; D Research remains available internally as aleph-component://d-research.

Highlights

• Updates the byte-locked D Research component to 3.2.1.
• Adds production-capable semantic retrieval through optional sentence-transformers, with deterministic local-hashing fallback when that extra is unavailable.
• Improves BibTeX and CSL export for articles, books, proceedings, structured personal names, and literal corporate authors while failing closed on conflicting identities.
• Adds optional local langdetect support and preserves the dependency-free trigram language detector.
• Keeps the D Research ledger, research gateway, sealed roleplay boundary, external HMAC refusal, and assurance rules unchanged.
• Forces optional Hugging Face/Transformers model resolution into cache-only mode unless the caller explicitly grants network access.

No new mandatory dependency is introduced. Aleph does not bundle or auto-install sentence-transformers, langdetect, Node.js, Playwright, Chromium, browser binaries, credentials, or node_modules.

Compatibility

• Aleph package and validator versions advance to 2.1.1.
• Workspace schema_version remains 2.0.0.
• Numerical formula_version remains 2.0.0.
• D Research workspaces require no schema migration.
• Previously finalized Aleph workspaces must be revalidated because validation, quality, finalization, and component-binding receipts are version-bound.
• The seven Aleph domain packs remain experimental; this release does not authorize probability claims.

Upgrade from v2.1.0

Install or extract v2.1.1 into a clean skill directory, keep the v2.1.0 package and workspace as rollback copies, and resolve all commands from the installed Aleph root.

Bash preflight

export ALEPH_SKILL_ROOT="$(cd "/absolute/path/to/aleph-skill" && pwd -P)"
export ALEPH_WORKSPACE="/absolute/path/to/workspace"

python "$ALEPH_SKILL_ROOT/scripts/preflight.py" --json
python "$ALEPH_SKILL_ROOT/scripts/research_gateway.py" research:preflight --json

PowerShell preflight

$env:ALEPH_SKILL_ROOT = (Resolve-Path 'C:\absolute\path\to\aleph-skill').Path
$env:ALEPH_WORKSPACE = (Resolve-Path 'C:\absolute\path\to\workspace').Path

python "$env:ALEPH_SKILL_ROOT\scripts\preflight.py" --json
python "$env:ALEPH_SKILL_ROOT\scripts\research_gateway.py" research:preflight --json

Require package preflight status: pass and D Research status: available, source: bundled, package_version: 3.2.1, path: aleph-component://d-research, and lock_ok: true.

For a finalized workspace that contains D Research 3.2.0 component-binding receipts, use the non-mutating check first, then create a sibling workspace with refreshed 3.2.1 binding. The source workspace is not modified, and the destination <workspace>-bundled-bind must not already exist.

Bash component rebind

python "$ALEPH_SKILL_ROOT/scripts/migrate_workspace.py" --source "$ALEPH_WORKSPACE" --bind-bundled-d-research --check --json
python "$ALEPH_SKILL_ROOT/scripts/migrate_workspace.py" --source "$ALEPH_WORKSPACE" --bind-bundled-d-research --json
export ALEPH_TARGET_WORKSPACE="${ALEPH_WORKSPACE}-bundled-bind"

PowerShell component rebind

python "$env:ALEPH_SKILL_ROOT\scripts\migrate_workspace.py" --source "$env:ALEPH_WORKSPACE" --bind-bundled-d-research --check --json
python "$env:ALEPH_SKILL_ROOT\scripts\migrate_workspace.py" --source "$env:ALEPH_WORKSPACE" --bind-bundled-d-research --json
$env:ALEPH_TARGET_WORKSPACE = $env:ALEPH_WORKSPACE + '-bundled-bind'

The rebinder verifies retained research ledgers with both Aleph and the locked upstream canonicalizer, refreshes portable component identity, and invalidates stale finalization and assurance receipts. If the ledger has an HMAC sidecar, provide D_RESEARCH_LEDGER_KEY through the environment only. Do not put secrets in the workspace or command line.

Re-run every affected import, compile/run/replay, sensitivity, hindcast, and report step against the sibling, then finalize and require clean validation. Workspaces that never imported D Research do not need component rebinding, but finalized copies still need new 2.1.1 receipts; set ALEPH_TARGET_WORKSPACE to a separate non-overwriting working copy before running the lifecycle commands below.

python "$ALEPH_SKILL_ROOT/scripts/validate_simulation_artifacts.py" --workspace "$ALEPH_TARGET_WORKSPACE" --mode draft --write-report --json
python "$ALEPH_SKILL_ROOT/scripts/render_simulation_report.py" --workspace "$ALEPH_TARGET_WORKSPACE"
python "$ALEPH_SKILL_ROOT/scripts/finalize_simulation.py" --workspace "$ALEPH_TARGET_WORKSPACE" --json
python "$ALEPH_SKILL_ROOT/scripts/validate_simulation_artifacts.py" --workspace "$ALEPH_TARGET_WORKSPACE" --mode final --require-report --write-report --json
python "$ALEPH_SKILL_ROOT/scripts/evaluate_simulation_quality.py" --workspace "$ALEPH_TARGET_WORKSPACE" --threshold 90 --enforce --json

Rollback to v2.1.0

Restore the verified v2.1.0 skill package together with the untouched v2.1.0 workspace. Aleph v2.1.0 contains D Research 3.2.0 and cannot validate receipts bound to the 3.2.1 component lock. There is no automatic reverse migration; do not hand-edit component paths, hashes, or signed receipts.

Provenance and release scope

The D Research snapshot is derived from the upstream verified annotated tag v3.2.1:

• Tag object: 4f797f0cb0f75539edfc9bc9332ca4dd041e881c
• Commit: dc07d4902361ddf15ff0dd093faa0784b2fd47ab
• Git tree: 4943fec2d4fa6bda1f4beecfd9177a630af0ab8b
• Canonical git archive SHA-256: e91837e6d2d38cde4f055f54d382e58c7b2094220b87e77dff695f92a63cdd71
• Bundled component-tree SHA-256: 0276b3f4b6230b65e75c664edd642b9b2b7ce30e534a88a8328899b3d6fc7edd
• Boundary: 730 upstream tracked files, 529 exact repository-only CI/release/dogfood-evidence exclusions, and 201 locked runtime files
• Upstream package boundary: 199 paths

Aleph release CI independently fetches the pinned tag and verifies its tag object, peeled commit, Git tree, reproducible archive, exact exclusion recipe, package manifest, and every bundled byte. It then runs the cross-platform Python and Node matrices, no-browser acceptance, real Chromium gateway smoke, strict static checks, lifecycle/replay/adversarial gates, and two byte-identical release builds before publication.

Per maintainer direction, no new model dogfood or manual multi-model forward-test campaign is claimed for this dependency patch. This omission does not skip deterministic package, security, provenance, browser, lifecycle, or reproducibility gates.

Verify the release

Download aleph-skill-v2.1.1.zip, distribution-manifest.json, and SHA256SUMS.txt into one directory:

sha256sum -c SHA256SUMS.txt
gh attestation verify aleph-skill-v2.1.1.zip \
  --repo d-init-d/aleph-skill \
  --signer-workflow d-init-d/aleph-skill/.github/workflows/release.yml \
  --source-ref refs/tags/v2.1.1
for asset in distribution-manifest.json SHA256SUMS.txt; do
  gh attestation verify "$asset" \
    --repo d-init-d/aleph-skill \
    --signer-workflow d-init-d/aleph-skill/.github/workflows/release.yml \
    --source-ref refs/tags/v2.1.1
done

The build-provenance records are stored by GitHub's attestation service rather than attached as separate release files.

See the full changelog, the Aleph comparison, and the upstream D Research comparison for the complete change set.

Aleph Skill v2.1.0

Aleph Skill v2.1.0

Aleph 2.1.0 ships D Research as a locked internal component, so users install one Aleph package while retaining the full bundled D Research skill surface, subject to the host capabilities and optional dependencies described below. Workspace schema_version and formula_version remain 2.0.0; there is no schema-number migration, but external D Research integrations and previously finalized workspaces must follow the compatibility and migration guidance below.

Highlights

• Bundles D Research 3.2.0 under components/d-research/ and addresses it portably as aleph-component://d-research.
• Verifies the component with an annotated upstream tag, commit, Git tree, reproducible source archive, exact snapshot recipe, per-file SHA-256 digests, and a component-tree digest.
• Routes every Aleph-to-D Research subprocess through one allowlisted gateway with shell=False, filtered environment variables, workspace path controls, timeouts, and output limits.
• Preserves full D Research ledger functionality, including 14/19/22/23-column contracts, HMAC verification, claim-only evidence import, process/blocker audit records, and portable component_binding receipts.
• Keeps research and roleplay separated by a sealed packet/receipt boundary. Host-enforced network and tool denial remains mandatory for verified roleplay assurance.
• Adds byte-exact workspace rebinding, upstream/Aleph canonical dual-run checks, sibling staging, and automatic invalidation of stale finalization artifacts.

Compatibility and behavior changes

• Users need to install only the Aleph package. Hosts should register the root SKILL.md; Aleph invokes nested D Research internally. Do not invoke components/d-research/SKILL.md as a second Aleph entrypoint.
• The verified bundle wins by default. D_RESEARCH_SKILL cannot silently override it.
• Selecting an explicit external installation requires both --external-d-research <path> and --allow-external. Automation that expects an explicit path to be selected must supply both flags and inspect the reported source.
• A healthy verified bundle remains the default. An incompatible explicit candidate fails closed instead of being hidden by the bundle. When the bundle is unavailable, external opt-in can also admit configured legacy discovery roots.
• External D Research remains compatibility-limited and cannot receive bundled-verified assurance or privileged secret forwarding.
• schema_version and formula_version remain 2.0.0; the seven domain packs remain experimental and do not authorize probability claims.
• Aleph and bundled D Research remain licensed under CC-BY-NC-4.0. See LICENSE, THIRD_PARTY_NOTICES.md, and components/d-research/LICENSE.

Bundled D Research provenance

• Package: d-research-skill-tools 3.2.0
• Upstream: d-init-d/d-research-skill
• Tag: v3.2.0
• Annotated tag object: 091997d33825a29ed6f30b089c33de9acf897ae3
• Commit: 8156b460b358636e69d66c10c16800045d258446
• Git tree: 2a77a48bd98126fec5812398afd98033e03d4709
• Canonical git archive SHA-256: 7942008985c288dd276bee459de61400662eb5a327febf84ebef9e953f2df1d7
• Component snapshot SHA-256: aac64ef9ae80f5343818a03cdacdc003b4aee0bf27b0c4798d66ad37ac219bcd
• Snapshot recipe: all 209 upstream tracked files minus the exact 11 repository-only paths listed in component-lock.json, with included text normalized to LF
• Bundled snapshot: 198 locked files; D Research package manifest: 196 archive paths

From a clone of the upstream D Research repository, reproduce the canonical archive digest on a POSIX shell with host line-ending conversion disabled and the TAR permission mask pinned:

git -c core.autocrlf=false -c core.eol=lf -c tar.umask=0002 \
  archive --format=tar 8156b460b358636e69d66c10c16800045d258446 | sha256sum

Release CI fetches only the pinned annotated tag into a bare shallow repository, then verifies the tag object, peeled commit, Git tree, canonical archive, snapshot recipe, and every bundled byte before publication.

Upgrade from v2.0.1

Back up each workspace and install or extract v2.1.0 into a clean skill directory. Do not overlay unrelated files or credentials. Resolve commands from the installed Aleph root rather than the current working directory.

Bash

export ALEPH_SKILL_ROOT="$(cd "/absolute/path/to/aleph-skill" && pwd -P)"
export ALEPH_WORKSPACE="/absolute/path/to/workspace"

python "$ALEPH_SKILL_ROOT/scripts/preflight.py" --json
python "$ALEPH_SKILL_ROOT/scripts/research_gateway.py" research:preflight --json
python "$ALEPH_SKILL_ROOT/scripts/research_gateway.py" research:manifest --json
python "$ALEPH_SKILL_ROOT/scripts/research_gateway.py" research:route --json

PowerShell

$env:ALEPH_SKILL_ROOT = (Resolve-Path 'C:\absolute\path\to\aleph-skill').Path
$env:ALEPH_WORKSPACE = (Resolve-Path 'C:\absolute\path\to\workspace').Path

python "$env:ALEPH_SKILL_ROOT\scripts\preflight.py" --json
python "$env:ALEPH_SKILL_ROOT\scripts\research_gateway.py" research:preflight --json
python "$env:ALEPH_SKILL_ROOT\scripts\research_gateway.py" research:manifest --json
python "$env:ALEPH_SKILL_ROOT\scripts\research_gateway.py" research:route --json

The package preflight must report top-level status: pass together with d_research.status: available, d_research.source: bundled, d_research.path: aleph-component://d-research, and d_research.lock_ok: true. Each gateway probe must exit successfully, report status: ok, and retain the bundled component_binding. A successful process exit alone is not sufficient evidence of a verified bundle.

Every finalized v2.0.1 workspace needs revalidation because validator_version changes from 2.0.1 to 2.1.0; old version-bound receipts are stale even when the workspace does not use an absolute D Research path. Preserve the original as the rollback copy and create a separate, non-overwriting working copy. For a workspace that does not need rebinding, select that copy and begin with draft validation:

Bash revalidation without rebinding

export ALEPH_TARGET_WORKSPACE="/absolute/path/to/workspace-v2.1-revalidate"
test -d "$ALEPH_TARGET_WORKSPACE"
python "$ALEPH_SKILL_ROOT/scripts/validate_simulation_artifacts.py" --workspace "$ALEPH_TARGET_WORKSPACE" --mode draft --write-report --json

PowerShell revalidation without rebinding

$env:ALEPH_TARGET_WORKSPACE = (Resolve-Path 'C:\absolute\path\to\workspace-v2.1-revalidate').Path
python "$env:ALEPH_SKILL_ROOT\scripts\validate_simulation_artifacts.py" --workspace "$env:ALEPH_TARGET_WORKSPACE" --mode draft --write-report --json

For a workspace that stores an absolute external D Research path, first run the non-mutating equivalence check. The rebinder writes a sibling <workspace>-bundled-bind directory and never mutates the source workspace.

Bash migration

python "$ALEPH_SKILL_ROOT/scripts/migrate_workspace.py" --source "$ALEPH_WORKSPACE" --bind-bundled-d-research --check --json
python "$ALEPH_SKILL_ROOT/scripts/migrate_workspace.py" --source "$ALEPH_WORKSPACE" --bind-bundled-d-research --json

export ALEPH_TARGET_WORKSPACE="${ALEPH_WORKSPACE}-bundled-bind"
python "$ALEPH_SKILL_ROOT/scripts/validate_simulation_artifacts.py" --workspace "$ALEPH_TARGET_WORKSPACE" --mode draft --write-report --json

PowerShell migration

python "$env:ALEPH_SKILL_ROOT\scripts\migrate_workspace.py" --source "$env:ALEPH_WORKSPACE" --bind-bundled-d-research --check --json
python "$env:ALEPH_SKILL_ROOT\scripts\migrate_workspace.py" --source "$env:ALEPH_WORKSPACE" --bind-bundled-d-research --json

$env:ALEPH_TARGET_WORKSPACE = $env:ALEPH_WORKSPACE + '-bundled-bind'
python "$env:ALEPH_SKILL_ROOT\scripts\validate_simulation_artifacts.py" --workspace "$env:ALEPH_TARGET_WORKSPACE" --mode draft --write-report --json

Inspect migration-bind-report.json and the draft validation report before continuing. Rebinding rewrites component identity and deliberately removes the stale artifact index, validation receipt, quality receipt, and finalization.

For either upgrade path, re-run every affected original lifecycle command against ALEPH_TARGET_WORKSPACE, including compile/run/replay, sensitivity, or hindcast where applicable. Aleph cannot infer the scenario-specific arguments for those commands. Render a fresh report before rebuilding final receipts.

After those workspace-specific stages succeed, complete the lifecycle:

Bash finalization

python "$ALEPH_SKILL_ROOT/scripts/render_simulation_report.py" --workspace "$ALEPH_TARGET_WORKSPACE"
python "$ALEPH_SKILL_ROOT/scripts/finalize_simulation.py" --workspace "$ALEPH_TARGET_WORKSPACE" --json
python "$ALEPH_SKILL_ROOT/scripts/validate_simulation_artifacts.py" --workspace "$ALEPH_TARGET_WORKSPACE" --mode final --require-report --write-report --json
python "$ALEPH_SKILL_ROOT/scripts/evaluate_simulation_quality.py" --workspace "$ALEPH_TARGET_WORKSPACE" --threshold 90 --enforce --json

PowerShell finalization

python "$env:ALEPH_SKILL_ROOT\scripts\render_simulation_report.py" --workspace "$env:ALEPH_TARGET_WORKSPACE"
python "$env:ALEPH_SKILL_ROOT\scripts\finalize_simulation.py" --workspace "$env:ALEPH_TARGET_WORKSPACE" --json
python "$env:ALEPH_SKILL_ROOT\scripts\validate_simulation_artifacts.py" --workspace "$env:ALEPH_TARGET_WORKSPACE" --mode final --require-report --write-report --json
python "$env:ALEPH_SKILL_ROOT\scripts\evaluate_simulation_quality.py" --workspace "$env:ALEPH_TARGET_WORKSPACE" --threshold 90 --enforce --json

When a research ledger has an HMAC sidecar, set D_RESEARCH_LEDGER_KEY in the environment before any step that verifies the ledger. Never place the key in the workspace or command line.

If --check reports COMPONENT_DRIFT, do not force the rewrite. Keep the workspace unchanged and either align the external installation byte-for-byte ...

Aleph Skill v2.0.1

Aleph Skill 2.0.1

Aleph 2.0.1 is a portability, protocol-correctness, security, and release-integrity update. Schema and numerical formula identifiers remain at 2.0.0, so there is no schema-number migration. The validator intentionally tightens several 2.0 contracts; existing workspaces must be revalidated and may require artifact regeneration before they pass.

Highlights

• Promotes current native Agent Skills discovery for supported IDEs and CLIs, with exact project/user paths and non-global fallbacks.
• Makes every documented command resolve from an explicit installed skill root instead of assuming the process working directory.
• Reconciles the likelihood contract: uncalibrated branches use relative_weight; calibrated probabilities require the calibration and hindcast gates.
• Adds a provenance-preserving, assurance-limited research fallback when D Research is unavailable.
• Adds resumable research checkpoints and an honest unsaturated handoff when a host execution window ends.
• Makes lifecycle acceptance part of the release gate and enforces genuine strict mypy checks.
• Keeps self-tests non-mutating by placing lint, type-check, and coverage caches outside the attested package tree.
• Reports every refused or failed installer transaction with ok: false, including Windows symlink privilege failures.
• Builds the official ZIP from the verified distribution manifest only, verifies byte-for-byte reproducibility, and publishes GitHub build-provenance attestations.
• Verifies the remote annotated tag object independently of actions/checkout, binds its commit to both the checked-out source and main, then rechecks the exact tag object and target immediately before publication.

Compatibility notes

• schema_version remains 2.0.0.
• formula_version remains 2.0.0.
• This is not input-compatible with every artifact accepted by the v2.0.0 validator. Revalidate existing workspaces before relying on them.
• In uncalibrated workspaces, node/branch/actor likelihood must use the exact relative_weight contract and must not retain probability claims.
• Material-actor packets, roleplay outputs, execution rows, and receipts must be regenerated when they do not satisfy the new packet-only semantic binding and anti-replay checks; do not hand-edit or copy hashes.
• Re-render reports that use the old “probability updates” heading, then recompile/replay numerical artifacts whose source hashes changed and finalize again to regenerate receipts.
• migrate_workspace.py remains the 1.x-to-2.0 schema migrator; it is not a substitute for this v2.0.1 revalidation/repair pass.
• The seven bundled domain packs remain explicitly experimental; they do not authorize probability claims.
• External CLI JSON profiles describe a portable adapter contract. They are not a claim of turnkey orchestration for every third-party host.

Check an existing workspace in draft mode before changing it:

python "$ALEPH_SKILL_ROOT/scripts/validate_simulation_artifacts.py" \
  --workspace <run-dir> --mode draft --write-report

After repairing reported contract errors, rerun the workspace's original compile/run/replay commands, render the report, finalize, and require a clean final validation. Preserve an untouched backup throughout the process.

Verification

The tagged release workflow must pass the full Linux, Windows, and macOS verification matrix before publishing. It runs the cross-contract release gate on Python 3.10 through 3.13, plus strict static checks, lifecycle acceptance, adversarial validation, deterministic replay, adapter drift checks, domain-pack validation, and manifest-exact packaging.

Manual cross-host forward check

The reproducible release ZIP was also exercised read-only on 2026-07-14 with Grok Build 0.2.101 (grok-4.5), OpenCode 1.17.18 through Ollama Cloud (glm-5.2, kimi-k2.6, and minimax-m3), and Cline 3.0.40 through all ten Cline Pass models exposed by the installed binary (deepseek-v4-flash, deepseek-v4-pro, glm-5.2, kimi-k2.6, kimi-k2.7-code, mimo-v2.5, mimo-v2.5-pro, minimax-m3, qwen3.7-max, and qwen3.7-plus). Every model recovered the exact likelihood, sealed-roleplay, fallback-assurance, absolute-root, package-version, and schema-version contracts and observed passing preflight and fixture validation results.

Every listed model passed the semantic audit, including a second adversarial prompt that required it to reject uncalibrated probability, roleplay networking, verified fallback assurance, and current-working-directory helper resolution. Exact single-object JSON adherence varied by host, model, and prompt: Grok prepended progress text; OpenCode GLM changed the requested schema on the adversarial form while citing all four correct violations; and members of the MiMo/MiniMax Cline Pass set intermittently wrapped or reshaped otherwise correct output. These are recorded as model instruction-following warnings, not silently counted as strict-format success. The checks demonstrate semantic interpretability of the packaged instructions; they do not grant simulation assurance or calibrate any domain pack.

Download the ZIP, distribution-manifest.json, and SHA256SUMS.txt into the same directory, then verify both downloadable files:

sha256sum -c SHA256SUMS.txt

The build-provenance attestation is stored by GitHub's attestation service rather than attached as a separate release file. With GitHub CLI installed, verify the ZIP against this repository:

gh attestation verify aleph-skill-v2.0.1.zip \
  --repo d-init-d/aleph-skill \
  --signer-workflow d-init-d/aleph-skill/.github/workflows/release.yml \
  --source-ref refs/tags/v2.0.1

Repository-level immutable releases are enabled before the tag is pushed, so GitHub locks the published tag and assets. The workflow also refuses to replace or edit an existing v2.0.1 release.

See CHANGELOG.md for the detailed change list and retained engineering limits.

Aleph Skill v2.0.0

Aleph Skill 2.0.0

Aleph 2.0 is a production-oriented rewrite of the evidence-grounded causal timeline simulation skill. It remains portable across AI CLIs and IDEs and does not require a unified CLI or MCP server.

Highlights

• Strict schema 2.0 contracts, typed public issue codes, recursive unknown-field rejection, bounded loaders, formula replay, and fail-closed assurance gates.
• Deterministic and Monte Carlo simulation with counter-based RNG, intervention handling, convergence checks, sensitivity analysis, hindcasting, and reproducible run hashes.
• Numerical run/replay binding to the declared model, trace, saved results, engine trajectory, and branch ledger.
• Sealed D Research 3.x integration with package discovery, claim-only ledger import, evidence regeneration, HMAC verification, and provenance receipts.
• Explicit separation of research and roleplay tracks, with public-role actor dossiers, knowledge packets, privacy controls, and Tier A receipt verification.
• Secure transactional installer for Codex, .agents, supported IDEs, and external AI CLIs, backed by a distribution allowlist and rollback receipts.
• Migration from schema 1.2 to 2.0 with sibling output, check mode, source-tree digest, and canonical dual-run verification.
• Seven validated data-only domain packs and eleven generated adapter targets.

Breaking changes

• New artifacts are written as schema 2.0.0; migrate 1.2 workspaces with sim:migrate.
• Assurance tiers are now experimental, limited, verified, and calibrated; a score cannot override a hard gate.
• Uncalibrated scenarios use relative_weight. Probability output requires declared calibration and hindcast evidence.
• Propagation traces are recomputed and replayed; forged or stale effects are rejected.
• Run, trace, branch, research, and roleplay integrity gates are stricter and may reject previously accepted workspaces.

See CHANGELOG.md and Upgrade plan.md in the source archive for the complete contract and migration details.

Verification

• 178 automated tests pass.
• GitHub Actions passes on Windows, macOS, and Ubuntu with Python 3.10, 3.11, 3.12, and 3.13, plus the quality job.
• Release gates include Ruff, strict mypy, 80% branch-coverage threshold, package validation, adapter drift checks, all seven domain packs, deterministic replay, adversarial rejection, and complete lifecycle acceptance.
• Release commit: f5af157f5c2bd3ec0b8fcc408e161ff30994729f.
• CI run: https://github.com/d-init-d/aleph-skill/actions/runs/29318391629
• Distribution tree: 209 files; tree SHA-256 30d946076d1a2a8f86a0eb94d368b0c86358d73f2416cb8302853b74d70e1ea8.
• Archive SHA-256: d2c2a2dd4e9b9afa9e8edba8f68d77ba975012c06ae2c8c959d31e4852323d0a.
• Manifest SHA-256: 0f8fe27f68729bfe7be42b9152f14049b2c6004559799a4b9507bebb53fdfab6.

Residual engineering limits

• Installer commit-time path rechecks materially narrow TOCTOU exposure, but a fully atomic OS-level guarantee would require platform-native directory-handle and no-follow APIs.
• Model/run pair promotion is exception-safe and rolls back ordinary failures; power-loss atomicity would additionally require a durable recovery journal.

These limits are documented explicitly so downstream users can choose controls appropriate to their threat model.

Aleph Skill v1.2.0

Adaptive evidence-saturated causal simulation across counterfactual pasts, alternate presents, and branching futures. Removes execution profiles and fixed caps; adds automatic retrospective/prospective/hybrid temporal modes, present-to-future projection, complexity-scaled research floors, strict human research/roleplay separation, professional reports, and a 90-point excellent release gate. Verified with 20 unit tests, Agent Skills validation, and OpenCode regressions across DeepSeek V4 Flash, GLM 5.2, Kimi K2.7 Code, MiniMax M3, and Qwen 3.7 Max.

Aleph Skill v1.1.0

Aleph Skill v1.1.0

This release hardens causal simulation quality and makes human-node separation auditable.

Highlights

• Schema 1.1 with cross-artifact ID integrity.
• Mandatory distinct Human Research and Human Roleplay subagents when the runtime exposes subagent tooling.
• Auditable human-track ledger with timestamps, agent references, knowledge cutoffs, and normalized hypotheses.
• Quick, standard, and deep execution profiles with immutable source and repair-loop budgets.
• Direct-source, source-tier, contradiction, confidence, and search-snippet quality gates.
• Strict final report validation and a 100-point quality evaluator.
• Ten validator regression tests covering common failure and gaming modes.

Forward-test evidence

Baseline workflows were exercised with DeepSeek V4 Flash, GLM 5.2, Kimi K2.7 Code, and MiniMax M3 through Ollama Cloud, plus Qwen 3.7 Max through Cline Pass. The final upgraded DeepSeek regression produced separate research/roleplay subagents, passed independent validation with zero errors and warnings, and scored 98.75/100. Later provider calls for several models stalled or queued after initialization; those runs are treated as provider limitations rather than successful regressions.

Aleph Skill v1.0.0

Changelog

v1.0.0 - 2026-07-01

Initial production release.

Added

• Portable skill core with SKILL.md frontmatter limited to name and description.
• Runtime guidance for Codex, Claude Code, OpenCode, and generic .agents skill directories.
• D Research integration guide for evidence ledgers, source discovery, contradiction checks, and public-role actor research.
• Mandatory Human Research / Human Roleplay split for material human decision nodes.
• Seven-phase timeline simulation workflow: define, research, construct, link, propagate, branch, validate.
• Node, edge, human-node, propagation, branch, safety, and reporting reference guides.
• JSON/CSV/JSONL templates for simulation manifests, nodes, causal edges, actor dossiers, evidence maps, branch ledgers, propagation traces, and validation reports.
• Stdlib-first helper scripts for package validation, workspace initialization, artifact validation, butterfly scoring, report rendering, preflight checks, and adapter installation.
• Local release-gate self-test workflow documented in README.

Verification

• npm run self-test
• python scripts/validate_skill_package.py .
• python scripts/validate_simulation_artifacts.py --examples
• end-to-end workspace lifecycle smoke test.