retail-init
Skill2 starsChanged 4 months agoApache-2.0View on GitHub
Bootstrap the Compass-Driven kit and lead a NEW user to a first visible result on THEIR own table in the Seshat BI repo. Use when someone asks to "set up the kit", "initialize seshat", "onboard me / get me started", or runs `retail init`. The AGENT performs this: it writes the backstage substrate (compass router + fenced AGENTS.md/CLAUDE.md regions + manifests) via the substrate-only `retail init` CLI, DELEGATES the worked-example offer + human-seam list to `first-hour-compass`, and ROUTES into `retail-onboard-table` for the Stage-1 read-only profile (grain candidates + column types over a live DB, or `[PENDING LIVE PROFILE]` without one). It is NOT a terminal wizard, stores no run-state, fetches no remote, self-grants no approval, and emits no confidence score.
- Reads credentials
- Installs packages
In this repository60 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- approval-consoleSkill
- approval-evidence-packSkill
- business-knowledge-interviewSkill
- capabilitiesSkill
- consumer-data-dictionarySkill
- cross-table-lineageSkill
- dagster-orchestration-adapterSkill
- dashboard-designSkill
- dashboard-intelligenceSkill
- dbt-transformation-adapterSkill
- evidence-pack-generatorSkill
- first-hour-compassSkill
- friendly-pr-reviewerSkill
- grain-confidence-reviewerSkill
- kpi-contract-builderSkill
- pbi-mcp-write-adapterSkill
- pbip-workflowSkill
- pbip-xraySkill
- pbir-authoring-adapterSkill
- portfolio-watchSkill
- powerbi-dashboard-designSkill
- pr-readiness-reviewerSkill
- readiness-viewerSkill
- release-notes-generatorSkill
- report-intent-interviewSkill
- retail-build-warehouseSkill
- retail-control-roomSkill
- retail-discover-portfolioSkill
- retail-governSkill
- retail-onboard-tableSkill
- retail-orchestrateSkill
- retail-scaffoldSkill
- retail-semantic-checkSkill
- retail-validateSkill
- run-next-readinessSkill
- seshat-studioSkill
- showcase-buildSkill
- source-mappingSkill
- speckit-analyzeSkill
- speckit-checklistSkill
- speckit-clarifySkill
- speckit-constitutionSkill
- speckit-git-commitSkill
- speckit-git-featureSkill
- speckit-git-initializeSkill
- speckit-git-remoteSkill
- speckit-git-validateSkill
- speckit-implementSkill
- speckit-planSkill
- speckit-specifySkill
- speckit-tasksSkill
- speckit-taskstoissuesSkill
- bi-analyst-knowledgeSkill
- bi-bigdata-knowledgeSkill
- bi-dax-knowledgeSkill
- bi-python-knowledgeSkill
- bi-sql-knowledgeSkill
- retail-kpi-knowledgeSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 47
Bootstrap the substrate: Run retail init (or retail init --repo <path>).
- Distinctive: line 71
Does not profile, open a DB, prompt, or show a menu (that is the CLI's hard boundary — substrate only).
- Very specific: line 73
Does not advance a readiness stage, write an approvals[] entry, or emit a numeric health / confidence / percent-ready score (FR-010.
- Very specific: line 75
Does not fetch from a remote or auto-execute pulled content — kit self-update (sync, Phase-3) and channel-driven fetch (Phase-4) are later, …
- Distinctive: line 78
Does not rewrite any hand-authored / constitution-owned content.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
retail-init
The install-time front door for the Compass-Driven kit. Its one job is to turn a
fresh pip install into a first visible result on the user's own table — leading
with analyst value, keeping the substrate backstage. You (the agent) perform this
flow; the retail init CLI only writes substrate and prints the next step. This is
Phase-1 Step 1-2 of docs/roadmap/distribution-ideas.md.
Scope + non-negotiables (read first)
- Agent-first, not a wizard (Principle I). The
retail initCLI is substrate-writing ONLY — it writes.seshat/+ the fenced regions and prints the next step. It never prompts, shows a menu, or emits a profile. The delegate → route → profile flow is YOU performing this skill over the existing prose verbs. - Delegate, never fork (anti-fork). The worked-example offer, the human-seam
list, and the single-table orientation card belong to
first-hour-compass— its single source. The Stage-1 profile belongs toretail-onboard-table. Do NOT restate any of them here; route into them. - No run-state (FR-005).
compass.yamldeclares the orientation protocol and points at per-tablereadiness-status.yaml; it stores nocurrent_stage. There is no repo-level stage. - Fence-only writes (FR-006/FR-007). The substrate write touches only the
<!-- SESHAT-KIT START -->…<!-- SESHAT-KIT END -->region of AGENTS.md / CLAUDE.md; everything outside is hand-authored / constitution-owned and stays byte-identical. - Stop at judgment (Principle V). Grain, PII publish-safety, business
rollup/segment, product identity are the human's — surfaced (via
first-hour-compass) and STOPPED on, never self-granted. - ASCII only, UTF-8 no BOM.
The flow (what you perform)
- Bootstrap the substrate. Run
retail init(orretail init --repo <path>). It writes.seshat/compass.yaml+ manifests and projects theSESHAT-KITfenced regions of AGENTS.md / CLAUDE.md, then prints the next step. Do NOT narrate the substrate as steps — it is backstage. - Set expectations honestly (FR-009). State up front: the agent handles the
sequence and the plumbing; you still own the judgment seams. Surface the seam
wording from
first-hour-compass(its single source) — do not re-type a divergent list. - Delegate the worked-example offer. Route into
first-hour-compass, which presentsretail-store-sales(full seven-stage spine,docs/worked-examples/ retail-store-sales.md) and takes the user's pick as a narrative pattern to steer by (not a file template). - Route into the profile. Hand the user's named table to
retail-onboard-table(the Source → Mapping front door, which owns the Stage-1 read-only profile). The agent-visible first result is the grain candidates + column types it returns. - Degrade honestly when there's no DB. The Stage-1 profile is DB-backed
(
profile.pyover aQueryRunner). If nodbextra / DSN is configured, report the boundary and the enable steps (pipx inject seshat-bi psycopg2-binaryorpip install "seshat-bi[db]"; setDATABASE_URLorANALYTICS_DB_*in the gitignored.env), mark profile numbers[PENDING LIVE PROFILE], author the source-map / orientation structure, and STAY USEFUL — never traceback, never fake a pass. There is no CSV/Excel profiler (YAGNI).
What init does NOT do
- Does not profile, open a DB, prompt, or show a menu (that is the CLI's hard boundary — substrate only).
- Does not advance a readiness stage, write an
approvals[]entry, or emit a numeric health / confidence / percent-ready score (FR-010; hard rule #9). - Does not fetch from a remote or auto-execute pulled content — kit self-update
(
sync, Phase-3) and channel-driven fetch (Phase-4) are later, gated slices (FR-011). - Does not rewrite any hand-authored / constitution-owned content; on a malformed fence it STOPS and reports rather than force a rewrite.
Idempotent re-run
Running retail init again on a bootstrapped repo re-projects only the fenced
regions, reports "already bootstrapped", and never duplicates a fence or clobbers the
hand-authored region.
See also
- The delegated verbs:
.claude/skills/first-hour-compass/SKILL.md(offer + seams + card),.claude/skills/retail-onboard-table/SKILL.md(Stage-1 profile). - The router it writes:
.seshat/compass.yaml; its canonical source:.seshat/kit-source.yaml. - The design + resolved decisions:
docs/roadmap/distribution-ideas.md;specs/070-retail-init-bootstrap/. - The conductor for the later medallion stages:
.claude/skills/retail-orchestrate/SKILL.md.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

