@skyf0xx/hedgehog 3.0.10 → 3.0.12

package/README.md

@@ -48,6 +48,18 @@ The build order is encoded into the project. The AI does not have to remember wh
 
 ![Small steps, big leverage: small context loops, continuous verification, traceable evolution, sustainable velocity](https://raw.githubusercontent.com/skyf0xx/hedgehog/master/docs/images/small-steps.png)
 
+## Your build order is a graph
+
+**Every task** Hedgehog generates **is a node** with explicit dependencies in sqlite.
+
+Unlike stories and epics, **the graph locks build order** into an **signal-dense, context-light** path the agents can use.
+
+```bash
+npx @skyf0xx/hedgehog graph
+```
+
+![The Hedgehog build graph](https://raw.githubusercontent.com/skyf0xx/hedgehog/master/docs/images/graph.png)
+
 ## What Hedgehog builds
 
 ### Full-stack applications
@@ -156,15 +168,6 @@ from them. It never touches the instructions file, the build graph, the
 core workspace, or `skills/BMAD`, since those carry project-specific or
 write-once content.
 
-To see the build graph:
-
-``` bash
-npx @skyf0xx/hedgehog graph
-```
-
-Starts a small local server and opens a live, read-only diagram of every
-task, status and its dependencies.
-
 ## Why Hedgehog
 
 Most AI coding tools improve prompting.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@skyf0xx/hedgehog",
-  "version": "3.0.10",
+  "version": "3.0.12",
   "description": "Install the Hedgehog build discipline (agents + skills) into a repo, for Claude Code, Cursor, or Gemini CLI.",
   "type": "module",
   "repository": {

package/src/skills/hedgehog-core-design/blueprints/data-pipeline.md

@@ -15,9 +15,16 @@ schedule  — the orchestration wiring: what runs when, retries, backfill entry
 
 ## Adaptation points
 
-- Merge `schedule` into `load` when the pipeline is invoked externally
-  (cron calling a script, an orchestrator defined outside this repo) —
-  there's no orchestration surface here to build.
+- On the Dagster stack, `schedule` is Dagster's own schedule/sensor
+  definitions plus the retry policy on each op — real orchestration
+  surface that belongs in this repo and earns its own layer.
+- On the stdlib/argparse stack, or when a pipeline is invoked externally
+  regardless of stack (cron calling a script, an orchestrator defined
+  outside this repo), there is no in-repo orchestrator to retry or
+  backfill through: merge `schedule` into `load` and scope it to
+  whatever the script itself can do (an idempotent upsert key, a
+  `--since`/`--backfill` flag `load` reads), and record retries/
+  scheduling as owned by the caller rather than as a gap in this layer.
 - Split `extract` per source (`extract/{module}`) when the pipeline pulls
   from several genuinely different systems with independent failure modes
   and auth; keep one `extract` layer for a single source.

package/src/skills/hedgehog-core-design/blueprints/infra-deploy-tool.md

@@ -15,9 +15,16 @@ cli      — argv, output rendering, and the confirmation gate in front of apply
 
 ## Adaptation points
 
-- Merge `plan` into `apply` only for a tool that is genuinely fire-and-forget
-  (a one-shot bootstrap script) — and note that this gives up the dry-run
-  surface, which is the main reason to reach for this shape.
+- On the TypeScript substitute (a thin wrapper generating config/manifests
+  with no systems-level need), `provider` isn't a live target system —
+  there's no observed state to read, so `plan` and `apply` collapse into
+  one `render` layer producing the config/manifest text, and `provider`
+  drops entirely. The dry-run safety property below doesn't apply: the
+  tool's output *is* the printable, diffable artifact, so there's nothing
+  further to preview.
+- Merge `plan` into `apply` only for a Go-stack tool that is genuinely
+  fire-and-forget (a one-shot bootstrap script) — and note that this gives
+  up the dry-run surface, which is the main reason to reach for this shape.
 - Split `provider` per target (`provider/{module}`) when the tool spans
   several systems with independent auth and failure modes; keep one layer
   for a single target.
@@ -31,4 +38,6 @@ what a plan named. A tool that mutates during planning can't offer a
 trustworthy dry run, and dry run is the safety property this shape exists
 to provide — an infra tool without it is a script that edits production
 with no preview. `plan` must be printable and diffable on its own, with
-no credentials required beyond reading observed state.
+no credentials required beyond reading observed state. This boundary is
+moot on the TS/`render`-collapsed substitute above, where there's no live
+target to mutate in the first place.