Changelog

All notable changes to the quality-strategy plugin, newest first. The format follows Keep a Changelog, and versions follow Semantic Versioning; while the version is below 1.0, a minor release can remove or replace skills. To pick up a new version in Claude Code, run claude plugin update quality-strategy@tollens (or /plugin update quality-strategy@tollens inside Claude Code).

0.5.2 - 2026-11-10

Added

• A terms page, TERMS.md, and a link to it in the plugin manifest.

Changed

• The plugin's icon is a PNG in place of an SVG.

0.5.1 - 2026-11-10

Added

• A privacy page, PRIVACY.md. The plugin runs inside your own agent and collects nothing.
• A page for known issues, KNOWN-ISSUES.md.
• Listing details in the plugin manifest for OpenAI's plugin directory: a name, a description, starter prompts, an icon, and links for support and privacy.

Changed

• The README opens by saying what the plugin does and what you end up with, puts installing it next, and is shorter.
• This changelog is shortened to a list of changes for each version. The known issues that were in it moved to their own page, grouped by what to check, what it costs you and where to steer.
• The roadmap and the open questions are rewritten for someone new to the project.

0.5.0 - 2026-11-10

The strategy is now a set of steps, each a skill of its own, in place of one long interview, and the plugin also runs in Codex. Most of the earlier skills are replaced; strategies they saved remain readable.

Added

• Three skills for the first steps of a strategy: establish-project-context, stakeholder-deepdive and prioritise-release-quality.
• connect-quality-strategy writes the strategy up straight after the priorities, with a section for each priority. A second agent reviews it before you get it.
• priority-deepdive works one priority further: what good looks like, how you will know, where it stands now and what to do about it.
• quality-interview, an open interview about what is on your mind. It saves a transcript, a record for each topic and a summary, and you can pause and resume it.
• pre-read reads a project's files and reports the major issues, what the tests cover and how friendly the code is to agents.
• quality-setup agrees once where the quality documents are kept, who they are for and whether the agent may commit them.
• draft-strategy-steps drafts steps without you and marks them as the agent's drafts.
• quality-report makes visual reports, a page to explore, and versions for a particular reader.
• A decision log, decision-log.md, records each change to the saved documents.
• A template for the written strategy, references/strategy-template.md.
• Codex support: it installs as a Codex plugin from the same repository.
• A package for OpenAI's plugin directory.

Changed

• Node.js is now needed: quality-interview and the strategy steps use small scripts to keep their records.
• quality-strategy no longer runs the whole strategy itself. It says where your strategy stands and which step comes next.
• Stakeholders can be grouped, nested and overlapping.
• There is a fourth level for how a stakeholder would take a result: Ugh, meaning active dislike short of rejection, beside Delight, Good Enough and Dealbreaker.
• A stakeholder's view is recorded as your understanding of them, and what you do not know about them becomes an open question.
• You put the release's priorities in order yourself. The High, Medium, Low and None ratings are taken from your order, for you to correct.
• What anyone says about how good something must be is kept in their words.
• Records mark what you agreed, what you said, what the agent inferred and what it guessed.
• Open questions stay in the document they are about, each with a next action.
• When you do not know something, the agent asks how you could find out before it suggests a way.
• The agent asks one question at a time.
• quality-strategy-review reviews any quality document, not only a whole strategy.
• Reviews and reports work from partial records.
• The effective-comms plugin, 0.4.0 or newer, is needed alongside this one.

Removed

• The single seven-part strategy flow and its required document layout. Strategies in that layout remain readable.
• Ratings and confidence levels that the skills assigned automatically, the scoring of reports, and the mandatory final review cycles.
• The separate list of accepted risks. An accepted risk is kept in the document it is about.
• The worked example, examples/fernly/.
• The skills in this table. Their source remains under retired/ for reference.

Removed skill	Use instead
quality-artefacts, strategy-variants, operational-distillation	quality-report
test-strategy, evaluation-strategy, oracle-adequacy, tooling-strategy, tooling-adequacy, process-strategy	priority-deepdive, one priority at a time. Detailed test planning and build plans are outside the plugin
test-strategy-review, contradiction-check	quality-strategy-review

0.4.0 - 2026-07-10

Added

• A choice of document structure when a session covers more than one release.
• /quality-strategy can recommend a change to how the team works as one of the actions in its plan.
• The context interview asks for a recent example of how human-led and agent-led work goes.

Changed

• A non-goal names a level of quality the release is deliberately not reaching, not a feature that was deferred.
• Each quality dimension is scoped to a release, a stakeholder and a part of the product.
• How much a dimension matters is asked separately from how good it is now.
• Findings are grouped by theme when there are enough to group.
• When a stakeholder's bar could mean "once" or "every time", the skill asks which.
• The first question, where the document should live, says why it matters and recommends a private first pass.
• When you correct something, the skill re-reads the passage before changing it.
• Ideas you volunteer are tracked through the session.
• /strategy-variants no longer makes a variant from an unfinished strategy, and asks who a client-safe version is for.

Fixed

• Internal and agent stakeholders are no longer skipped because they already appear in another role.

0.3.7 - 2026-06-26

Added

• The strategy skills run /effective-comms on their documents before finishing.
• The review skills flag communication failures such as buried recommendations.
• The effective-comms plugin is a declared dependency.

0.3.6 - 2026-06-15

Added

• Strategy documents name the version of the skill that produced them.

0.3.5 - 2026-06-12

Changed

• /quality-artefacts describes its slideshow format in its own words, without another product's branding. No change in behaviour.

0.3.4 - 2026-06-11

Changed

• /quality-artefacts, asked without a brief, opens a gallery of starting points.
• /quality-artefacts asks what tone you want when you have not said.
• /quality-artefacts puts the strategy's most striking finding in the title.

0.3.3 - 2026-06-11

Added

• /quality-artefacts is offered when the strategy is complete.

Changed

• /quality-strategy proposes each cut in scope and asks you to confirm it, instead of assuming it.
• /quality-strategy includes security, data integrity and runaway spending in the strategy by default. You can take one out only by accepting the risk explicitly.
• When agents do the work, /quality-strategy adds quality dimensions that matter to agents.
• Before calling a behaviour a defect, /quality-strategy asks what purpose it serves.
• Human and agent perspectives are considered for each dimension, and only meaningful differences are shown.
• The current state comes from results, tests and your own evidence before code reading.
• Each strategy skill says how to resume after a break.
• /test-strategy-review checks a strategy for notes about how it was produced that do not belong in it.

0.3.2 - 2026-06-11

Added

• Revising a document archives the current version first.
• A revision revisits earlier concerns with evidence and looks for new gaps.
• A revision review asks for evidence behind each "fixed".
• The strategy skills say how far along you are, and offer a place to stop at the end of each part.

Changed

• A strategy with many High ratings is no longer challenged for their number. Each High must be justified.

0.3.1 - 2026-06-11

Added

• /test-strategy: exploratory testing and testing in production as named methods.
• /quality-strategy: a targeted design review where evidence is thin.
• Observability, fixability and recoverability in the list of quality dimensions.
• This changelog.

Changed

• /quality-strategy treats a goal that is met as met, and no longer asks for more than the release needs.

0.3.0 - 2026-06-11

Added

• /quality-artefacts, which designs a shareable SVG or HTML view of your strategy.
• examples/fernly/, a worked sample on a fictional plant-care app.

0.2.9 - 2026-06-11

Added

• Interactions between features, in the list of quality dimensions.
• A standing search for defects by fresh agents, in /test-strategy.

0.2.8 - 2026-06-10

Added

• First public alpha release: /quality-strategy, /test-strategy, /tooling-strategy, /strategy-variants, and their review and adequacy skills.