Six skills · all live · MIT
An agent starts every session with no memory of the last one.
It doesn’t need a better prompt — it needs a briefing. These skills author the context files a project should have anyway: a changelog of what shipped, a decision log of why, a handoff for where you left off. All Markdown. None of it is code.
$ /plugin marketplace add vidhunnan/agentic-skillsThe context stack
Five tiers, one rule.
Every project doc is tiered by the question it answers — and each tier carries how far it can be trusted. Mix them and you’ve handed an agent contradictory instructions. A human notices the contradiction; an agent just complies, confidently, in both directions.
- 01docs/concepts/What are we even trying to build?Hypothesis · future tense, disposable
- 02docs/prds/What are we still deciding?Proposal · a concept worth building
- 03docs/decisions/Why did we choose that?Truth · past tense, append-only
- 04handoff/Where did we leave off?Snapshot · the latest one wins
- 05changelog/What actually shipped?Truth · generated from git
The rule: never cite a concept or a PRD as proof a feature exists — check the changelog, or check the code. changelog/ is what shipped; everything under docs/ is what we thought.
Install one, or all six
The skills.
Each is a separate plugin. Every fact each one writes is pulled from your repo — git, CLAUDE.md, the files themselves — never invented.
Set up the repo
Start here. It builds the folders the rest of them fill.
- 01repo-setupCode · Chat
Scaffolds your context stack — concepts, PRDs, decisions, handoffs, changelog. Surveys what you already have, adopts your existing folder names, and never moves or overwrites a thing.
/plugin install repo-setup
Keep the record
The three questions a teammate with amnesia will ask.
- 02changelog-trackerCode
What actually shipped? — documents every substantive commit into a per-commit file plus a rolling index. Every fact comes from git.
/plugin install changelog-tracker - 03decisions-loggerCode · Chat
Why did we choose that? — mines the project for decisions that were really made and writes each as an ADR with its evidence. Where the reasoning was never written down, it asks rather than inventing one.
/plugin install decisions-logger - 04handoff-generatorCode · Chat
Where did we leave off? — an interactive, bidirectional Chat↔Code project handoff: progress, timeline, features, decisions, changelog delta, open questions, next actions, and a near-verbatim session log. Interviews you first; on Claude Code it verifies against the repo.
/plugin install handoff-generator
Working conventions
Two small habits, made to stick across sessions.
- 05branch-namingCode
Suggests and creates a branch name that follows your project's convention — read from CLAUDE.md, or inferred from your existing branches. Always confirms before creating.
/plugin install branch-naming - 06model-strategyCode
Builds docs/MODEL-STRATEGY.md — which Claude model for which kind of work, tailored by interview, with a mandatory review rule.
/plugin install model-strategy
Read before you install
This repo runs on its own skills.
Everything below was written by the skills above, while building this repo. Read the output before you trust it with yours.
- docs/decisions/ADRs explaining why this repo is shaped the way it is, each with the evidence it was drawn from.— written by decisions-logger
- changelog/Every substantive commit documented, with the diff and the reason.— written by changelog-tracker
- docs/MODEL-STRATEGY.mdThe model policy this repo actually follows.— written by model-strategy
- handoff/The briefs that carried this work between Claude.ai and Claude Code.— written by handoff-generator
Get started
Install.
Add the marketplace once. Then install the skills you want — each is a separate plugin, so take one or take all six.
$ /plugin marketplace add vidhunnan/agentic-skills$ /plugin install repo-setup$ /plugin install changelog-tracker$ /plugin install decisions-logger$ /plugin install handoff-generator$ /plugin install branch-naming$ /plugin install model-strategy