v0.9.0

Agentic Praxis Grimoire v0.9.0 Release Notes

Agentic Praxis Grimoire (APGR) is a provider-neutral toolkit and canonical skill
corpus for bounded agent engineering. It provides coding-agent systems, LLM
harnesses, and workflow engines with deterministic primitives for task-scoped
guidance selection, immutable evidence collection, curated environment resolution,
repository structure inspection, and context-footprint accounting without
imposing orchestration policy.

This documentation covers 0.9.0. The previous public release baseline was
v0.8.1, published and frozen across Git, GitHub Releases, PyPI, npm, and the
Go module proxy. Once this version is published, packages are available from
standard registries. Prior to formal publication, release qualification, packaging,
and publication remain separately authorized boundaries.

Highlights

• CI-First qualification interface: Added the --summary-file <path> CLI option
  emitting strict 6-field machine-readable JSON evidence (version, subproject,
  suite, test_status, gate_status, source_commit) adhering to JACA's
  subproject evidence contract.
• Closed validation role taxonomy: Established a 4-role taxonomy:
  • policy: fast mechanical validation checking repository inventory, toolchain
    versions, skill library mechanical and Go embedded corpus integrity, and record identity.
  • unit: Python unit testing with exact 80/80 statement and branch coverage gates.
  • integration: Python integration and CLI boundary testing with exact 80/80 gates.
  • unit-integration (combined / apg-dev-gate): combined union gate requiring
    85/85 statement and branch coverage.
• JACA XO compatibility & caller-owned adapter pattern: Verified Go library
  consumption behind a caller-owned internal adapter fixture (testing/fixtures/xo_consumer/)
  utilizing caller DTOs (CallerSkillEvidence, CallerFootprintEvidence,
  CallerComponentEvidence, CallerSourceReference, CallerComparisonDelta,
  CallerProjectionEvidence) to ensure zero APGR domain types leak into caller domain.
  Verified pure in-memory smoke execution without subprocesses, AST containment,
  static dependency boundaries (no os/exec, internal, or cmd packages), context
  cancellation propagation, and sentinel error fidelity.
• Dual-lane verification: Demonstrated clean dual-lane conformance across the
  published v0.8.1 baseline (Lane A via public Go proxy) and the exact development
  candidate (Lane B via disposable local replace).
• Context-footprint accounting: Source-bound observations with explicit
  fidelity and omission disclosure (footprint Go package and apgr footprint
  CLI commands). Reaffirmed key principles:
  • Observations are source-bound measurements, not capacity forecasts.
  • Unavailable metrics remain explicit with stated reasons and nil values; unavailable != zero.
  • Component separation: selected descriptions, selected bodies, support material,
    repository references, and provider prompt overhead are distinct components;
    overlap is not additive.
• Preserved invariants: Retained all 39 canonical skills (14 stable / 25
  provisional), the 9,527-byte discovery integrity limit, the six public Go
  packages (schema, report, skills, envsnap, hotspot, footprint), and
  full backward compatibility with published v0.8.1 APIs.

Truthful status and boundaries

• Published baseline vs 0.9.0 release:
  • v0.8.1 is the preceding frozen baseline on Git, GitHub Releases, PyPI, npm, and Go proxy.
  • Once this version is published, 0.9.0 packages become available on public registries.
    Prior to publication, candidate downloads are not published to registries.
• APGR-local qualification != JACA registration:
  • APGR-local qualification of CI summary emission and test roles is completed
    on Darwin arm64 under phase APG114 (Exit 00159).
  • Downstream runner registration in JACA CI (tools/ci/evidence.go, workflow.go,
    trustedRoleOrder) is consumer-owned in the JACA repository and remains pending.
• XO fixture != production adapter or security proof:
  • The caller-owned adapter in testing/fixtures/xo_consumer/ qualifies APGR Go
    package compatibility and illustrates DTO containment.
  • It is not JACA's production adapter (which belongs in JACA's xo/src/main/go).
  • Bounded passive DTOs do not enforce workflow security or prove universal isolation.
• Platform support and developer qualification gate:
  • Prebuilt runtime targets: macOS Apple Silicon (darwin/arm64), Linux x86_64
    (linux/amd64), and Linux ARM64 (linux/arm64).
  • Developer qualification gate: macOS Darwin arm64 is fully qualified.
    The Linux x86_64 developer qualification gate is currently pending: the policy
    suite is pending runner qualification, and unit/integration/combined
    suites are blocked by whole-inventory preflight binding Darwin arm64 Nix store digests.
• Dispatch and package contents:
  • apgr test (including --summary-file and the policy role) is a repository
    maintenance command requiring an APGR Git source checkout and the developer
    toolchain (Python 3.11+, pytest, coverage, pytest-cov, pytest-xdist, Git 2.40+, Go 1.25+).
  • Prebuilt native Go binaries (cmd/apgr) and bare wheels/npm packages do not
    carry the test runner or test suites. Prebuilt binaries provide runtime commands
    (skills, footprint, env, analyze hotspots, report, response).

Source-checkout first use

To explore the v0.9.0 source, use an APGR Git checkout:

# From the checkout root:
go run ./cmd/apgr --version
go run ./cmd/apgr skills list

To run the v0.9 CI qualification summary interface in a checkout with developer toolchain:

# Run fast mechanical policy validation emitting machine-readable summary
bin/apgr test policy --summary-file "${TMPDIR:-/tmp}/apgr-policy-summary.json"
cat "${TMPDIR:-/tmp}/apgr-policy-summary.json"

Once this version is published, standard registry installation commands (pip install agentic-praxis-grimoire==0.9.0, npm install -g @knowledge-forge-ai/apgr@0.9.0, go get github.com/Knowledge-Forge-AI/agentic-praxis-grimoire@v0.9.0) become active. Prior to formal publication, installation is via source checkout.

Use the existing pinned developer Python environment: pytest 9.1.1, coverage
7.15.2, pytest-cov 7.1.0, and pytest-xdist 3.8.0. Unit and integration preflight
also require the selected TypeScript 7.0.2 compiler and Node runtimes identified
in the CI handoff. The Python frontend
may be installed, but still needs those prerequisites and the checkout; a bare
native/npm executable has no test command even when invoked from a checkout.
combined is the summary suite value and apg-dev-gate is a consumer role alias;
the CLI argument is unit-integration. source_commit records entry HEAD,
so qualification of uncommitted source also needs a separate tested-tree binding.

Upgrade guidance

• Production consumers: Once this version is published, upgrade to 0.9.0 across supported registries. Prior to publication, production consumers remain on the published, frozen v0.8.1 release.
• Go library consumers: All public APIs in skills, footprint, schema,
  report, envsnap, and hotspot remain fully backward-compatible with v0.8.1. Once this version is published, depend on v0.9.0.
• Candidate evaluators: Use disposable local Go replace directives or
  checkout-local scripts to test candidate features without affecting production locks.

Documentation index

• Root README
• Documentation Index
• CLI Reference
• Go Library Reference
• Distribution and Packaging
• JACA CI Integration Handoff
• JACA XO Compatibility Handoff
• APGR–JACA Integration Architecture
• v0.9 Roadmap
• Release Notes (v0.8.1)

Contributing & licensing

Agentic Praxis Grimoire is free software licensed under the
GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later).
Commercial licensing options and enterprise support are available from
Knowledge Forge AI.
Contributions require adherence to the project contribution terms in
CLA.md and CONTRIBUTING.md.

v0.8.1

Agentic Praxis Grimoire v0.8.1

Agentic Praxis Grimoire (APGR) is a provider-neutral toolkit and canonical skill
corpus for bounded agent engineering. It gives coding-agent systems, LLM
harnesses, and workflow engines deterministic primitives for task-scoped guidance
selection, immutable evidence collection, curated environment resolution,
repository structure inspection, and context-footprint accounting without
imposing orchestration policy.

Highlights

• Context-footprint accounting: Full support for measuring, comparing, and
  projecting context budgets across descriptions, bodies, and support material
  via the new public footprint Go package and apgr footprint CLI commands.
• Strict schema versions: Introduces versioned schemas including
  apg.context-footprint/v1, apg.context-comparison/v1,
  apg.context-projection/v1, apg.context-component-registry/v1, and
  apg.capacity-control-mapping/v1.
• 39 Canonical Agent Skills: 14 stable and 25 provisional skill leaves
  curated for test discipline, code hygiene, language profiles, and agentic workflows.
• Reusable Go packages: Public packages for schema, report, skills,
  envsnap, hotspot, and footprint.
• Deterministic multi-registry packaging: Bit-for-bit reproducible wheels,
  source distribution, npm platform packages, launcher, distribution manifest,
  and checksums.
• Enhanced package presentation: Packaged human-facing READMEs and rich
  metadata on PyPI, npm, and GitHub.

Installation

Python (PyPI)

pip install agentic-praxis-grimoire==0.8.1
apgr --version

npm (Node.js)

npm install -g @knowledge-forge-ai/apgr@0.8.1
apgr --version

Or run via npx:

npx @knowledge-forge-ai/apgr@0.8.1 --version

Go library & CLI

CLI executable:

go install github.com/Knowledge-Forge-AI/agentic-praxis-grimoire/cmd/apgr@v0.8.1

Go module dependency:

go get github.com/Knowledge-Forge-AI/agentic-praxis-grimoire@v0.8.1

Release recovery & v0.8.0 predecessor status

During the initial publication attempt of v0.8.0 on 2026-08-31:

• Git commit fc0fd99b41d24951d7db3535402c46ef9c671143 was pushed to main;
• Annotated tag v0.8.0 (aa83b64f019b92e211154cd246b939f3fc62ac91) was created; and
• The Go module proxy (proxy.golang.org) indexed v0.8.0.

Downstream publication to GitHub Releases, PyPI, and npm was stopped before completion
due to interactive credential requirements and stale documentation descriptions.
In adherence to registry immutability and zero-overwrite rules:

• The v0.8.0 Git ref and Go module entries are preserved unchanged as immutable
  predecessor state; no tag is rewritten or force-pushed.
• v0.8.1 is the first complete, synchronized release across all surfaces (Git,
  GitHub Releases, PyPI, npm, and Go module proxy).

Compatibility and breaking-change status

• Additive & backward-compatible: Fully backward-compatible with v0.7.0.
• All public v0.7 APIs (schema, report, skills, envsnap, hotspot) remain
  stable and compatible.
• The footprint package and CLI subcommands are additive.
• No public breaking changes.

Supported targets

APGR provides prebuilt native binaries for:

• macOS Apple Silicon: darwin/arm64 (macosx_11_0_arm64)
• Linux x86_64: linux/amd64 (manylinux_2_17_x86_64, linux-x64)
• Linux ARM64: linux/arm64 (manylinux_2_17_aarch64, linux-arm64)

Verification and checksums

Every official release includes ten canonical assets verified by SHA256SUMS and
apg-distribution-manifest.json:

1. agentic_praxis_grimoire-0.8.1-py3-none-macosx_11_0_arm64.whl
2. agentic_praxis_grimoire-0.8.1-py3-none-manylinux_2_17_aarch64.whl
3. agentic_praxis_grimoire-0.8.1-py3-none-manylinux_2_17_x86_64.whl
4. agentic_praxis_grimoire-0.8.1.tar.gz
5. knowledge-forge-ai-apgr-darwin-arm64-0.8.1.tgz
6. knowledge-forge-ai-apgr-linux-x64-0.8.1.tgz
7. knowledge-forge-ai-apgr-linux-arm64-0.8.1.tgz
8. knowledge-forge-ai-apgr-0.8.1.tgz
9. apg-distribution-manifest.json
10. SHA256SUMS

Verify downloads with:

sha256sum -c SHA256SUMS

Known limitations

• Operating systems: Linux and macOS only. Windows is not supported by the
  prebuilt native binaries or embeddable Go core.
• Context estimates: Provider-specific token estimates are marked explicitly
  as unavailable unless supplied by the caller; APGR measures exact bytes and
  UTF-8 characters locally without guessing tokenizer rules.
• Offline operation: Binary packages and CLI run entirely offline with zero
  telemetry or external network calls.

Documentation

• Root README
• Documentation Index
• CLI Reference
• Go Library Reference
• Distribution & Packaging Guide

Contributing & licensing

Agentic Praxis Grimoire is free software licensed under the
GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later).
Commercial licensing options and enterprise support are available from
Knowledge Forge AI.
Contributions are welcomed under the terms described in
CONTRIBUTING.md
and CLA.md.

v0.7.0

Agentic Praxis Grimoire v0.7.0

APG v0.7.0 adds the embeddable, library-first Go toolkit while preserving the
39-skill corpus and the existing command-line workflows.

Highlights:

• public Go packages for schemas, reports, skill bundles, environment
  snapshots, and structural hotspot analysis;
• a Go-backed apgr CLI with compatible Python adapters;
• platform-specific Python wheels for macOS arm64, Linux amd64, and Linux
  arm64, plus a source distribution;
• an npm launcher and three platform packages with no runtime download or
  JavaScript semantic fallback;
• deterministic task-scoped skill context, portable environment snapshots,
  and provider-neutral hotspot reports; and
• reproducible cross-ecosystem build information bound to version 0.7.0
  and the canonical corpus fingerprint.

The skill library remains 39 canonical skills: 14 stable and 25 provisional.
This release does not deploy or activate APG on a host.

v0.6.0

Agentic Praxis Grimoire v0.6.0

v0.6.0 adds six provisional component profiles to the public skill library:
GoMock, Vitest, JSX, React, MDX, and Astro. None is promoted to stable by this
release.

Project-selected projection remains authoritative. v0.6.0 does not ship
automatic or advisory profile selection. Cross-profile work selects the
narrowest applicable sibling owners for the question at hand; it does not
create a mandatory profile chain or an aggregate stack owner.

The public library contains 39 skills and 39 discoverable projections. Their
descriptions occupy 9,504 UTF-8 bytes, leaving 23 bytes beneath the 9,527-byte
v0.6 context ceiling. Checkout and isolated installed APGR context reports
agree on all 39 skills, 9,504 bytes, 9,492 characters, and zero malformed
metadata rows.

Release qualification exercised real web and Go compositions. The known
Terminal Nova locked-install stop is caused by its checked-in lockfile omitting
declared target dependencies; it is not an APG defect, and qualification did
not use an unlocked bypass.

After PyPI publication succeeds, install with:

python -m pip install agentic-praxis-grimoire==0.6.0

v0.5.0

Agentic Praxis Grimoire v0.5.0

v0.5.0 introduces the apgr Python CLI and distribution, with user
configuration under ~/.apgr, project configuration under .apgr, and
canonical outbox, report, and numbered-response contracts.

The release retains the CSS, Markdown, JavaScript, TypeScript, and Node.js
profile work. The public library contains 33 skills; maturity remains
skill-specific, and provisional profiles are not promoted by this release.
The accepted CSS and JavaScript qualification debt remains documented.

Astro, JSX, MDX, React, Vitest, and GoMock remain v0.6 scope. This release does
not include a Nix deployment adapter.

After PyPI publication succeeds, install with:

python -m pip install agentic-praxis-grimoire==0.5.0

APG v0.4.0

APG v0.4.0 — Manager Topology, Test Profiles, and Reproducible Evidence

Agentic Praxis Grimoire v0.4.0 expands APG from 19 to 28 public skills and strengthens the infrastructure used to select, test, report, validate, and publish them.

This release introduces a dedicated ChatGPT-manager topology, nine new provisional capabilities, a Python-first operational-reporting core, a pytest-based test architecture with explicit coverage gates, and version-aware public-release validation. It preserves APG's existing authority boundaries: skills provide bounded procedures and judgment, but do not grant permission, invent project policy, or silently expand a task.

Release snapshot

• 28 canonical public skills
• 28 catalog rows
• 28 checked-in Codex discovery projections
• 14 stable skills
• 14 provisional skills
• 26 general-router entries
• 1 ChatGPT-local router entry
• 27 mechanically checked route edges
• 9 newly published skills
• Deterministic single-parent release history rooted in the preserved v0.1.0–v0.3.0 public lineage

Release inclusion remains independent of maturity. All nine v0.4 additions are published as provisional, rather than being promoted simply because the release includes them.

ChatGPT-manager topology

v0.4.0 introduces chatgpt-manager-workflow, a dedicated subrouter for ambiguous ChatGPT-manager work.

The new topology separates general APG process selection from ChatGPT-specific management procedures:

• general APG ambiguity continues through the public APG workflow router;
• ChatGPT-manager ambiguity enters one checked local subrouter;
• valid direct skill selection remains authoritative;
• the router selects the smallest sufficient manager capability—or none;
• routing does not authorize dispatch, execution, acceptance, publication, or destructive action.

composing-approved-roadmap-assignments now has a canonical owner under the nested skills/chatgpt/ namespace. Its procedure remains distinct from planning, ordinary worker-assignment composition, dispatch, review, and acceptance.

Codex discovery remains intentionally flat: all 28 public skills are still projected under .agents/skills/, even where their canonical ownership is nested.

Nine new public skills

Management and migration

chatgpt-manager-workflow

Routes ambiguous ChatGPT-manager work to the smallest sufficient manager-specific capability while preserving explicit selection, human authority, and ordinary prompting as valid outcomes.

It is not a universal session bootstrap, mandatory chain, general APG router, or source of action authority.

converting-bash-scripts-to-python

Provides a bounded procedure for migrating an existing Bash executable or script family to Python.

The skill begins with characterization of the observable contract and requires compatibility to be preserved—or any deliberate migration to be made explicit. Generic implementation, testing, review, publication, and lifecycle decisions remain with their existing owners.

Test-framework and test-tool profiles

pytest-test-profile

Adds pytest-specific judgment for:

• discovery and collection;
• fixtures and parametrization;
• mocks;
• xdist behavior;
• coverage;
• warnings and failure interpretation.

The profile does not choose pytest, plugins, Python versions, worker counts, coverage thresholds, project commands, or external actions on behalf of the repository.

minitest-test-profile

Adds Minitest-specific guidance for test and spec organization, assertions, lifecycle hooks, mocks, stubs, fixtures, isolation, parallelism, filtering, runners, plugins, reporters, and external boundaries such as subprocesses, filesystems, and databases.

go-test-profile

Introduces an independent owner for native Go testing judgment. It remains separate from the Go language profile and from third-party assertion or comparison libraries.

go-cmp-test-profile

Introduces a distinct profile for go-cmp comparison semantics and option use.

Native Go testing and go-cmp remain separate owners. v0.4.0 does not introduce a unified Go testing-stack skill.

nix-test-profile

Adds Nix-test-specific guidance while explicitly preserving the distinction between source-reviewed claims and behavior verified in a running Nix environment.

This profile complements the stable nix-language-profile; it does not replace it or imply broad runtime and platform compatibility.

Build and development-environment profiles

dockerfile-profile

Adds Dockerfile-specific judgment for build definitions while leaving image selection, registry policy, runtime orchestration, deployment, CI/CD, secrets management, host configuration, and destructive operations to the target repository and operator.

vagrantfile-profile

Adds Vagrantfile-specific judgment covering configuration loading, machines, boxes, providers, networks, synced folders, provisioners, triggers, Vagrant state, and host-dependent behavior.

It remains bounded by repository policy and does not grant infrastructure-mutation authority.

Python-first operational reporting

v0.4.0 introduces a shared, dependency-free Python reporting core with:

• a common report model and envelope;
• Git CLI adaptation;
• rendering;
• destination and locking ownership;
• operational framing;
• thin command-line entry points.

The existing git-show-report and append-operational-report command families were characterized and migrated through this shared architecture. The release also adds git-diff-report.

The migration preserves defined behavior where required and records deliberate differences rather than treating a language rewrite as self-validating.

Formal phase-commit validation

The new apg-check-phase-commit-message command mechanically validates formal APG phase commit messages.

A compliant formal-phase message must contain:

• the exact canonical phase subject;
• Scope;
• Result;
• Verification;
• Not run;

in the required order.

The checker can validate a message file or an existing Git commit. It verifies the adopted structural contract only; it does not establish that the report is truthful or that the work deserves acceptance.

Pytest architecture and coverage gates

APG's own Python test architecture now provides three explicit modes:

unit
integration
unit-integration

The managed runner:

• defaults to eight xdist workers;
• disables worker restarts;
• accounts for worker and Python child processes;
• preserves deterministic artifact ownership and cleanup;
• measures component coverage independently;
• computes combined statement and branch coverage as a union rather than averaging reports;
• treats missing children, worker crashes, incomplete artifacts, and malformed results as failures.

The accepted gates are:

• unit: at least 80% statement and 80% branch coverage;
• integration: at least 80% statement and 80% branch coverage;
• combined: at least 85% statement and 85% branch coverage.

Bats remains in use where an exact Python supersession has not been established.

Focused corrections to the stable process core

The six stable process skills retain their maturity and trigger boundaries, but v0.4.0 incorporates reviewed corrections around:

• coverage-remediation hierarchy and stop conditions;
• proportionate expansion from focused tests;
• truthful distinctions between mocks and real boundaries;
• formal-phase and ordinary-work defaults;
• Git and operational evidence;
• compression of approved-roadmap manager assignments.

These corrections were exercised through 51 frozen cases. They introduce no new mandatory workflow chain and do not broaden any skill's authority.

Version-aware release validation

The public-release and lifecycle tooling now distinguishes historical and current release surfaces by semantic version.

That means:

• v0.2.0 and v0.3.0 retain their immutable historical inventories;
• v0.4.0 is validated against its own 28-skill policy;
• unsupported, malformed, truncated, retagged, or mismatched identities fail closed;
• a current source cannot silently redefine what an older release was required to contain;
• deterministic candidate construction and checking remain separate from publication authority.

The final v0.4.0 publication is one intentionally squashed Release v0.4.0 commit whose sole parent is exact public v0.3.0, together with an annotated v0.4.0 tag.

Deliberately not included

This release does not publish every candidate investigated during v0.4 development.

In particular:

• matryer-is-test-profile is absent after material defects remained unresolved;
• go-testing-stack is absent, and the proposed unified-stack architecture remains rejected;
• native Go testing and go-cmp remain independent provisional owners;
• no skill is promoted merely to make the release matrix look cleaner.

Deferral and rejection are valid outcomes when available evidence does not support a bounded public capability.

Maturity and compatibility limits

The stable catalog remains at 14 skills. The other 14 skills are provisional, including every capability added in v0.4.0.

Provisional means that a skill is sufficiently bounded and useful to publish, but its evidence still requires explicit limitations and continued observation.

Compatibility is established only where it was:

• exercised on the current macOS environment;
• explicitly source-reviewed;
• fixture-reviewed;
• marked not run; or
• marked not applicable.

Publication does not establish:

• universal applicability;
• automatic invocation;
• production warranty;
• comparative superiority;
• cross-platform compatibility;
• tar...

APG v0.3.0

APG v0.3.0 — Routed Workflows and Language-Aware Guidance

APG v0.3.0 expands Agentic Praxis Grimoire from six foundational process skills into a routed, language-aware engineering operating model with nineteen public skills.

The release adds thirteen capabilities while preserving the authority, evidence, privacy, provenance, and rollback boundaries established by earlier versions.

Release snapshot

• 19 release-included skills
• 14 stable skills
• 5 provisional skills
• 18 non-router capabilities
• 13 newly published skills
• 35 of 35 cross-repository dogfood cases matched
• Fresh-session discovery and explicit-use smoke completed for all nineteen skills

Release inclusion remains independent of maturity: provisional skills are deliberately published as provisional rather than being promoted merely to simplify the release.

New workflow capabilities

APG workflow router

agentic-praxis-grimoire-workflow helps when:

• multiple APG skills appear plausible;
• an earlier routing decision needs audit; or
• APG capability metadata may be incomplete or stale.

The router selects the smallest sufficient process skill or none. It does not bootstrap every session, mandate skill chains, override an explicit applicable selection, invent task authority, or own another skill's procedure.

Its checked capability map covers eighteen non-router skills and fails validation when the catalog grows without an explicit routing disposition.

Repository-guidance synthesis

synthesizing-repository-guidance analyzes dense, duplicated, mixed-scope, private, or source-derived guidance before rewrite.

It separates ownership, provenance, reuse rights, privacy, migration, rollback, and destination decisions. It does not silently rewrite source material, mirror private guidance, remove existing skills, or authorize a migration.

Approved-roadmap manager assignments

composing-approved-roadmap-assignments translates an already human-approved roadmap phase into a reviewable top-level manager assignment without expanding its authority.

Planning, routing, worker-assignment composition, dispatch, execution, review, and acceptance remain separate responsibilities.

Language, test, and database profiles

v0.3.0 introduces profile-based guidance for technology-specific judgment beyond repository policy.

Stable profiles

• python-language-profile
• bash-language-profile
• bats-test-profile
• zsh-language-profile
• zunit-test-profile
• nix-language-profile

The ZUnit profile is intentionally limited to the exact verified pairing of ZUnit v0.8.2 with Zsh 5.9.2. Zsh 5.3.1 is unsupported in the tested environment, and no broader compatibility range is claimed.

Provisional profiles

• go-language-profile
• ruby-language-profile
• postgresql-database-profile
• sqlite-database-profile

These profiles passed their retained evaluation requirements but remain provisional where the evidence does not yet justify stable maturity.

No generic SQL profile is introduced. PostgreSQL and SQLite retain distinct ownership because their transaction, locking, migration, recovery, operational, and file-ownership models materially differ.

Semantic phase and record identity

The new apg-check-record-identity command validates APG's semantic phase and durable-record conventions.

v0.3.0 formalizes:

• semantic phase identifiers assigned before implementation;
• independent ADR and exit-record sequences;
• durable semantic references instead of development commit identities;
• precommit record finalization; and
• separation of public records from private operational Git evidence.

Source-specific lifecycle support

The release and user-management tools now support version-specific skill sets while retaining state schema version 1:

• v0.2.0 sources declare six skills;
• v0.3.0 sources declare nineteen skills;
• update computes exact retained, added, and removed names;
• rollback restores the previous source's exact declared set; and
• historical v0.2.0 remains a valid immutable public source.

This avoids treating the current nineteen-name policy as though it had always applied to older releases.

Upgrading from v0.2.0

For an installation managed by apg-user-skills:

bin/apg-user-skills update --source <v0.3.0-public-checkout>
bin/apg-user-skills check

The update adds the thirteen v0.3.0 skills while preserving source identity, ownership state, retained links, rollback information, and unrelated user content.

Project-local installations can select all nineteen skills or an explicit subset:

bin/apg-project-skills install --repo <target-repository>
bin/apg-project-skills check --repo <target-repository>

A full Codex restart remains the fallback when discovery changes do not appear automatically.

Documentation

• v0.3.0 skill catalog
• Language-profile contract
• Phase and record identity
• Public release process
• Full changelog: v0.2.0…v0.3.0

APG v0.2.0

APG v0.2.0 — Stable Core and Reproducible Delivery

APG v0.2.0 turns the initial six-skill foundation into a stable, mechanically validated release and lifecycle system.

The release preserves the focused process model introduced in v0.1.0 while substantially strengthening maturity evidence, public distribution, installation safety, regression coverage, and long-term maintainability.

Highlights

All six process skills are now stable

Following repeated real-project use, representative non-triggers, edge and stop behavior, complete regression testing, post-Superpowers evidence, and fresh non-author review, all six original skills are promoted to stable:

• composing-bounded-worker-assignments
• designing-significant-changes
• planning-repository-work
• implementing-with-test-discipline
• debugging-systematically
• reviewing-and-verifying-repository-work

Stable means suitable for routine bounded use within each documented trigger and project-owned authority boundary. It does not mean universal applicability, automatic invocation, production warranty, or comparative superiority.

Focused implementation-discipline correction

implementing-with-test-discipline now accounts for locally owned code or test artifacts whose sole purpose ends because of an authorized change—when removal policy and authority are clear.

This is a narrow consistency rule, not permission for unrelated cleanup or independently owned lifecycle decisions.

Reproducible public releases

The new apg-public-release command builds and validates deterministic public candidates with:

• exact non-private source projection;
• byte- and mode-preserving path equality;
• one-parent append-only release history;
• annotated version tags;
• reproducible manifests, trees, commits, and tag objects;
• critical-owner and wrapper checks;
• licensing and confidentiality validation; and
• configured unit and integration tests.

The release process now detects the omitted-wrapper defect present in public v0.1.0.

Managed user-scoped lifecycle

The new apg-user-skills command supports:

• list
• install
• adopt
• check
• update
• rollback
• uninstall

It uses direct skill links, strict source identity validation, state-last updates, persistent locking, exact ownership records, and conservative rollback. Unrelated user skills and unowned containers remain outside its authority.

Mechanical skill-library validation

apg-check-skill-library provides a dependency-free, read-only checker for the adopted mechanical skill contract, including canonical leaves, catalog consistency, link containment, and checked-in discovery projection.

A passing check establishes mechanical consistency—not semantic usefulness, maturity, privacy, provenance sufficiency, or runtime discovery.

Distribution and testing improvements

v0.2.0 also:

• restores the missing bin/apg-project-skills wrapper;
• establishes an exact public-surface policy;
• separates unit and integration test layouts;
• strengthens public lineage and read-only validation;
• formalizes skill authoring, correction, maturity, deprecation, and removal;
• closes the legacy candidate-theme queue with explicit dispositions; and
• preserves v0.1.0 as the immutable root of the public release chain.

Superpowers is retired from the maintainer workflow. A subsequent fresh RepoMap session discovered all six public skills, applied the review skill successfully, passed managed checks, and preserved the target repository.

Upgrading from v0.1.0

Upgrading is strongly recommended because v0.2.0 restores the missing project-local wrapper and introduces strict release, installation, update, and rollback validation.

For a managed user-scoped installation from an exact public checkout:

bin/apg-user-skills install --source <v0.2.0-public-checkout>
bin/apg-user-skills check

For project-local installation:

bin/apg-project-skills install --repo <target-repository>
bin/apg-project-skills check --repo <target-repository>

A full Codex restart may be necessary when newly linked skills do not appear immediately.

Documentation

• v0.2.0 skill catalog
• Public release process
• User-scoped integration
• Full changelog: v0.1.0…v0.2.0

APG v0.1.0

APG v0.1.0 — Public Foundation

APG v0.1.0 is the first public release of Agentic Praxis Grimoire, a modular engineering operating model for coding agents.

This release establishes an evidence-oriented foundation for planning, design, implementation, debugging, review, and bounded delegation—without turning any single methodology into a mandatory universal workflow.

Highlights

Six foundational process skills

APG v0.1.0 introduces six canonical skills:

• composing-bounded-worker-assignments
• designing-significant-changes
• planning-repository-work
• implementing-with-test-discipline
• debugging-systematically
• reviewing-and-verifying-repository-work

Each skill owns one focused, reusable procedure with explicit trigger, non-trigger, authority, evidence, and stop boundaries.

Evidence-backed bootstrap

All six skills completed the v0.1 bootstrap standard:

• structural validation;
• representative positive scenarios;
• material non-trigger scenarios;
• edge or stop scenarios;
• independent review; and
• initial real-world dogfooding in APG and RepoMap.

This evidence supports bounded experimentation. It does not establish automatic selection, comparative superiority, or production readiness.

Codex discovery and project-local adoption

The repository includes a checked-in .agents/skills projection for Codex repository discovery. APG also defines a conservative project-local adoption model based on direct symbolic links, Git-local ownership state, narrow exclusions, verification, and rollback.

The tooling is designed to preserve tracked repository content and unrelated user-owned skills rather than silently claiming or replacing them.

Governance and provenance

The initial public release establishes:

• explicit human, ChatGPT, Codex-manager, and worker authority boundaries;
• the manager-worker delegation and reporting protocol;
• source provenance and reuse-rights requirements;
• architecture decision and truthful phase-exit records;
• a candidate-to-adoption lifecycle; and
• separation between private development evidence and the public release surface.

Licensing

APG is available under AGPL-3.0-or-later. Commercial licensing is available for proprietary embedding, private deployments, OEM distribution, support, warranty, indemnity, and other negotiated terms.

Maturity and limitations

All six v0.1.0 skills are provisional.

Release inclusion does not imply stable maturity, universal applicability, production warranty, automatic invocation, or clean comparative superiority. Superpowers may coexist with this release but remains reference evidence rather than APG authority.

Known issue

The v0.1.0 public projection accidentally omitted the documented executable wrapper:

bin/apg-project-skills

The underlying implementation and documentation were present, but the documented project-local command could not be invoked through its intended public entry point. This omission is detected and corrected by APG v0.2.0.

Users seeking managed project-local or user-scoped installation should use v0.2.0 or later.

Documentation

• Skill catalog
• v0.1 bootstrap model
• Project-local projection model
• Browse the v0.1.0 release